Posts

Posted on

AI Prompting Guide: Get Better Results with Copilot and ChatGPT

A person using a laptop

Generative AI is now woven into the way we work both at work and home.

Whether you’re drafting content, analysing data, or simply trying to get unstuck, tools like Microsoft Copilot can be transformative – but only if you give them the right instructions.

Great prompts lead to great outcomes.

Weak prompts lead to generic, vague, or unusable results. And in a world where AI is becoming a core productivity skill (almost like Internet search was in the late 90s), knowing how to prompt effectively is quickly becoming as important as knowing how to search the web.

Don’t worry it’s not rocket science but it does take some thinking before you type or speak your prompt.

Below is a practical, real‑world guide to writing better prompts using my place of zen and peace (the Isle of Wight) as a theme throughout to give it some meaning. This is written with Copilot in mind, but this will work across any AI tool just the same.

What is an AI prompt?

A prompt is the instruction or request you give to an AI model. It can be a question, a task, a description, or a request. A prompt can also be recursive, meaning, you can ask follow on prompts as you go.

For example:

“Give me three ideas for an AI community tech workshop in Ryde on the Isle of Wight.”

The clearer and more specific your prompt, the better you will find the AI’s response.

Why Writing Good Prompts Matter

Whilst you may have an idea in your mind of what you want from your prompt/ask – your AI is not a mind reader (not yet anyway!).

AI tools like ChatGPT and Copilot work with the information (instructions) you give it. If your prompt is vague, the output will be vague. If your prompt is rich with context, constraints, and intent, the output becomes sharper, more relevant, and more actionable.

Taking a little time to create good and thorough prompts:

  • reduce rework
  • save time
  • improve accuracy
  • produce content that feels closer to what you would write
  • Increase your confidence with working with AI.

How to Write Better AI Prompts

You’ll see loads of these scattered around social media, and most of these hold truth.

The aim here to simply these and give some simple advice that will improve every interaction you have with Copilot whether at home or at work.

1. Be Clear and Specific

Vague prompts produce vague answers.

Imagine you’re organising a local event. You type:

“Give me some AI event ideas “

With prompt like this, you will get generic suggestions that could apply to anywhere across any topic related to AI.

Instead, a little bit of detail can make a lot of difference:

“I’m organising a fundraising event for a coastal conservation charity on the Isle of Wight. I want to make this focused at small businesses and around Microsoft Copilot. Suggest some learning-based drop-in sessions I could run to help improve skills and knowledge around this whilst also raising money. I aim to run these in cheap to hire venues in towns like Ventnor, Yarmouth, or Sandown on the Isle of Wight”.

This prompt is much more thorough gives a Goal, Context around what I need and examples of what I am looking for, whist still giving the AI work to do.

👌Why not try the Prompts above and see the difference you get.

2. Telling is what you don’t want is Important

AI responds brilliantly to boundaries. Many of us tell Copilot what we want it to do, but we also need to pay as much attention to what we do not want it to do.

For example:

“Write a short social post for X and Facebook promoting a new ferry discount for commuters travelling between Cowes and Southampton.
Do keep it professional and concise.
Do not use emojis.
Do not use American spelling.”

These constraints help Copilot stay focussed, stay on‑brand and on‑tone.

👌Why not practice the above with something in your interest area or work?

3. Examples Make a lot of Difference

Examples are one of the most powerful prompt ingredients.

For example, let’s say you need an intro written for your next blog or newsletter and you need it written in a certain style. You can give Copilot a sample paragraph you like or point at an existing web/blog page. Watch to see how it will mirror the tone, structure, and rhythm.

For instance:

“Here’s an example of the tone I want — friendly, conversational, and lightly humorous. [or insert URL to existing post] Now write a similar intro for a guide to the best walking routes around Freshwater Bay.”

Examples act as creative guardrails.

👌Have a go - play around and have some fun. How about a different language, dialect or even as a poem!

4. Role, Tone and Audience matter

If your content is going to be read by other people, it is important to tell Copilot who those people are and how it should be written. It is also useful to give Copilot a role which can make a big difference in how it “thinks” about it’s output.

For example:

“You are an internal communications writer. Write an internal announcement for staff across our Isle of Wight offices about a new Head of Operations joining the team.
Tone: warm, optimistic, and inclusive.
Audience: a mix of office‑based and frontline colleagues, aged between 20–65.”

The AI will adapt its language accordingly.

👌Focusing on the output of the prompt when preparing content for others to consume makes it more readable and relevant. You can even try "Write multiple versions, one for my CEO, one for the Marketing Manager and one for the rest of the employees"

5. Conversations can yield better results

AI can perform multi-step tasks, but it does still need clarity. If you are not sure what the next step is (since you are exploring ideas), conversations (multi-iterative chats) work best . This is similar to how you might have a conversation with a co-worker or friend.

Below is a really simple example:

Instead of:

“List ways to promote our new cafe in Shanklin and tell me list which channels will reach tourists based on each option”

Since you might get over-whelmed with information here, why not turn it into a chat.

  1. First Prompt: “List ways to promote our new AI themed cafe in Shanklin.”
  2. Second Prompt: “Ok, I like idea number 3, can you suggest the best channels to reach tourists visiting the Isle of Wight in summer and tell my this will work.”
  3. Third Prompt: “Great, I need a professional post-card style image to market this. Can you suggest 3 ideas I can use”
  4. Forth Prompt: Generate me an image based on the first example.

This produces cleaner, more accurate results. Copilot remembers the theme of the conversation between each “turn”. 

👌Try the above - see where the conversation takes you. If you like what it has produced, you could ask Copilot to create you a prompt based on the conversation you could use next time!

Treat it like a creative partner, not a robot vacuum (which I still want to get by the way!)

👌If you don't like the revised version, you can also ask Copilot to go back the earlier version.

6. Experiment, test, refine

Prompting is a skill you build through practice.

Try variations. Adjust constraints. Add detail. Remove detail.
The more you experiment, the more you’ll understand how AI interprets your instructions.

Remember: AI is powerful, but your judgment is still essential. Always review, refine, and verify — especially when using AI for public‑facing content or factual information.

👌Check out Microsoft's Prompt Coach. This can give you feedback back on your prompt or ask you recursive questions to help you build a shape a re-usable prompt.

7. Use Memory and Instructions to make prompting easier

One of the biggest shifts in how we work with AI is the move from one‑off prompts to persistent personalisation. Copilot (both Consumer version and Business Version) gives you two features that dramatically improve the quality of your outputs: Copilot Memory and Copilot Instructions.

Used well, they reduce the amount of prompting you need to do, keep your outputs consistent, and help Copilot understand you rather than just the task in front of it.

7.1 Copilot Memory: teach Copilot what matters to you

Memory lets Copilot remember stable, long‑term information about you or your work so you don’t have to repeat it every time.

Think of it as giving Copilot the background context you’d normally give a new colleague in their first week.

For example, you might store:

  • the tone you prefer for public‑facing content
  • your role and responsibilities
  • your organisation’s audience
  • your writing preferences (like your preference for hyphens)
  • recurring details you use in content

This means you can stop repeating things like:

“Write this in a professional tone without emojis.”
“Keep the language accessible for a mixed audience.”
“Use UK spelling.”

Instead, Copilot already knows what you like.

Example:

If you regularly create content for local audiences, you could store:

“I work with organisations across the Isle of Wight, so when I ask for examples, keep them relevant to local towns like Newport, Ryde, Cowes, and Ventnor.”

“Remember that when I am planning an event, I will always need a marketing hook to sell the event to my audience who are small business owners”.

Now, when you ask Copilot for event ideas, marketing examples, or case studies, it automatically localises them without you having to specify it every time and since you asked it to always give you marketing hooks – it will do so too!

👌Have a go adding things to Copilot Memory. These are personal to you and you can view and delete them from the Copilot Settings or simply tell Copilot not to remember that "thing" any more.

7.2 Copilot Instructions: set the rules of engagement

If Memory is about who you are, Instructions are about how you want Copilot to behave.

Instructions let you define:

  • your preferred writing style
  • how formal or informal you want responses
  • how detailed you want explanations
  • whether you want step‑by‑step reasoning
  • How Copilot should structure outputs

This is especially powerful if you produce a lot of content or work across teams. These do not replace your prompts, but they reduce the number of times you tell it the same thing so help make your prompts more focussed on the ask or task.

Example

You might set an instruction like:

“When I ask for travel or tourism examples, prioritise locations on the Isle of Wight and keep suggestions practical for visitors.”

Or:

“When generating blog content, keep the tone friendly, clear, and audience‑first and always in UK english.”

Now Copilot behaves consistently across tasks, without you having to restate your preferences.

👌Instructions can be simple of really in-depth. My advice is to keep them generic otherwise they can negatively impact your day to day prompting. If you always ask for bullets, that is what you will get unless your prompt explicitly states otherwise!

7.3 Why Memory + Instructions matter for better prompting

Together, these features:

  • reduce repetitive prompting
  • improve consistency across your content
  • help Copilot understand your context
  • make outputs feel more like your voice
  • save time on every interaction

And crucially, they allow you to use shorter, simpler prompts because Copilot already knows the background and how you like it to work.

Instead of:

“Write a LinkedIn post about our new community workshop in Ryde. Keep it friendly, avoid emojis, use UK spelling, and keep it under 120 words.”

You can simply say:

“Write a LinkedIn post about our new community workshop in Ryde.”

Copilot fills in the rest because you’ve already taught it how you work.

Final Thoughts

Prompting or “Prompt Engineering” can be over-whelming but it is really just about planning what you want to achieve. Clear, specific instructions unlock better results, save time, and help you work smarter.

As tools like Copilot become a standard part of our daily workflows, the ability to prompt effectively is becoming a core digital skill. Investing a bit of time in learning and practising how to prompt well now will pay dividends across your work, creativity, and productivity.

If you’re helping teams adopt Copilot, training them in prompt‑writing — and showing them how to use Memory and Instructions — is one of the highest‑value steps you can take. And if you’re experimenting on your own, keep refining — the more you play, the better you get.

Posted on

Microsoft Expands Copilot Connectors: What the New Integrations Mean for Microsoft 365 and Enterprise AI

Banner Image for New Copilot Connectors

Microsoft has expanded the Copilot Connector catalog with 35 new connectors so Microsoft 365 Copilot can reach many more external systems and data sources directly, with new developer tooling and enterprise controls to keep governance front and centre. This wave includes a General Availability release of new connectors and tools to validate connector readiness.

New Copilot Connectors in App Groups

What are Copilot Connectors?

Copilot connectors allow organisations to bring external enterprise data directly into Microsoft 365 Copilot. Once connected, this enables Copilot to ground its responses in more of data beyond what is the Microsoft 365 environment thereby creating a more unified and data relevant rich experience for users and ultimately making Copilot a more effective AI tool for all line of business information.   These refreshed Connectors aim to provide:

  • Broader connector coverage: Microsoft announced a GA wave that adds dozens of new, first‑party connectors across categories such as developer tools, project management, content management, and IT service platforms.  
  • Indexing into Microsoft Graph: Copilot connectors bring external content into the Microsoft Graph so Copilot can ground responses in your organisation’s content rather than only relying on previously indexed snapshots. 
  • Developer and validation tooling: Microsoft published templates and a lightweight Copilot Connector Checker to help validate third‑party prerequisites and speed up integration.

Quick comparison of integration approaches

AttributeCopilot ConnectorTraditional Indexing / Sync
Access ModelOn‑demand connector access to external systemsPeriodic indexing or one‑way sync
ActionabilityQuery plus action patterns; can trigger workflows where permitted.Mostly read‑only search results .
Developer surfaceAPIs, templates, validation toolsGraph connectors, custom indexing
Governance Tenant admin controls, consent flows, auditing hooksVaries by connector, often separate tooling
Data RefreshNear real‑time or on‑requestDependent on indexing cadence

Connector Core Concepts

  • Contextual reach: Copilot can pull live context from multiple systems at the moment of the user’s request, reducing the “information in too many places” problem.  
  • Actionable intelligence: Connectors are designed so Copilot can do more than summarise where policies and permissions allow, it can suggest or initiate actions.  
  • Friction reduction for builders: Templates and the Connector Checker lower the barrier for teams to build Copilot‑native experiences.  
  • Enterprise Data Protection & Governance: The Copilot Connector expansion includes admin controls and consent mechanisms so organisations can enforce least‑privilege and auditing from the get-go!

Practical examples for Copilot Connectors

  • Sales: Draft follow‑ups in Outlook using live third-party CRM data via the connector.
  • Support: In Teams, Copilot can query third party ticketing systems and knowledge bases in real time to propose next steps.
  • Engineering: Copilot can surface GitHub/GitLab context (issues, PRs, knowledge) while drafting release notes or triage summaries.
  • Finance: Copilot can connect to third party finance systems and CRM for up-to-date quote to cash tracking, sales data and opp cash forecasting.

Microsoft have categorised these as follows:

Risks and trade‑offs to be explicit about

There are sometimes trade-offs to consider when expanding Copilot’s reach with connectors or APIs. These can include:

  • Expanded surface area increases the places data can flow – as such organisations should review and enforce scoped consent and least‑privilege. 
  • API usage and cost: Depending on the service connecting to, real‑time access can raise third‑party API calls and costs; plan quotas and caching. Start with a test and monitor.
  • Governance complexity: IT admins need will need to ensure there are clear policies, monitoring, and audit trails to avoid drift.
  • User expectations — when Copilot can act, build confirmations and safe defaults to avoid unintended changes.

Common Concerns?

A common question I expect here is: “Will this expose our internal data to Microsoft or third parties?”

Short answer: No, not by default. Connectors require explicit configuration and tenant consent; data flows only when you enable and scope them. Really, organisations should treat connector setup like any other integration – Zero Trust First: Restrict permissions, Enable logging, and Test in non‑production first.

Summary

Overall, I think these will provide much easier integration and use of third-party systems with Copilot. The announcement which is here covers more detail on:

  • General Availability of new connectors is expanding coverage across developer tools, and many line of business apps including project management, content management, IT services, and vertical systems.
  • These Copilot Connectors allow the indexing of external content into the Microsoft Graph so Copilot can ground answers in your organisation’s data outside of Microsoft apps and services.  
  • These “Action‑capable connectors” will enable Copilot to suggest or trigger workflows where permitted in line of business applications.
  • Developer tooling and validation including templates and the Copilot Connector Checker to speed integration.  
  • Enterprise governance features for consent, policy enforcement, and auditing.

Posted on

Is AI Forcing the Network conversation organisations have been avoiding?

Artificial intelligence is no longer just changing applications – it’s exposing the limits of enterprise network infrastructure.

In their Q2 earnings report, Cisco’s leadership team described the current campus refresh cycle as “the top of the first inning”, signalling a long‑term shift in how organisations should think about “networking in an AI‑driven world.”

The networks many enterprises rely on today were never designed for AI workloads.

Why AI Is Reshaping Enterprise Network Infrastructure

Traditional campus networks were built for predictable user traffic, cloud and client-server communications and incremental growth. AI changes those assumptions entirely.

AI is forcing enterprises to rethink campus networking. Cisco’s AI‑driven refresh success highlights why legacy networks are no longer fit for purpose. AI‑driven environments introduce:

  • Latency‑sensitive workloads that expose weak network design especially when we have agents working with other agents!
  • Increased east‑west traffic across campus and edge locations
  • Security requirements that must be embedded, not layered on. Security must be part of the network aka Secure Networking
  • Operational complexity that manual network management cannot scale with driving vendor consolidation.

The above are the reasons why Cisco are seeing renewed demand across switching, routing, wireless, and edge networking – and why this refresh cycle is being positioned as multi‑year rather than opportunistic.

Is this just a Hyperscaler Problem?

While hyperscalers dominate AI headlines, Cisco say that the more significant shift is happening inside enterprise environments, especially in environments that have not refreshed in the last 5 or so years.

As organisations move AI from experimentation into production, they quickly discover that:

  • Legacy campus networks struggle under AI‑driven traffic patterns
  • Performance issues become business‑critical risks
  • Security gaps are amplified by AI workloads
  • Network visibility becomes essential, not optional or bolt on.

The enterprise network is no longer background infrastructure. It is becoming part of the AI platform and fabric itself.

What This Means for Cisco Partners

This moment should an ideal opportunity for partners to position and service the refresh demand, but also creates pressure as demand increase, coupled supply chain challenges create a perfect storm.

Cisco’s recent changes to partner contract terms, driven by rising component and memory costs, highlight the commercial realities partners are navigating alongside technical transformation.

For partners, AI‑driven network modernisation is not about faster hardware refreshes. It is about:

  • Designing AI‑ready campus and edge networks
  • Embedding security into the network fabric
  • Modernising network operations and observability
  • Aligning infrastructure decisions to long‑term outcomes

Partners who lead with outcomes – resilience, performance, security, and readiness – will remain relevant. Those who focus solely on speeds and feeds will struggle to differentiate.

The Real Shift Is Strategic, Not Technical

The most important signal from Cisco was not about revenue growth or new product announcements (though there were plenty of those). It was more the acknowledgement that old networking assumptions no longer hold true and that this AI Transformation wave needs a thinking reset.

AI is forcing organisations to confront technical debt they have tolerated for years. And the enterprise network is often where that debt becomes impossible to ignore.

This campus refresh cycle will be as big as the Data Centre and Hypercaler demand. It is not simply about chasing AI hype (even though much of a Cisco’s new kit is baked with AI naturally) but it is about empowered organisations to invest in the secure network foundations that can support their current and future AI workloads without collapsing under them.

Cisco say that this is a conversation organisations  – and partners need to have early, and with intent. Cisco have assessments and resources to help shift the conversation from speeds and feeds to business readiness.

References: https://www.crn.com/news/networking/2026/cisco-ceo-ai-infrastructure-campus-refresh-opportunities-will-propel-networking-giant-toward-strongest-fiscal-year-yet

Posted on

What are Copilot connectors?

A Copilot connector is the plumbing that brings your organisation’s external content into Microsoft 365 so Copilot can ground answers in real, company-specific data.

Practically speaking, a connector extracts or fetches items from a third‑party system (files, tickets, PRs, CRM records, etc.), indexes those items into the Microsoft Graph, and makes them available to Copilot and Microsoft Search under your tenant’s security and governance rules.

How it works in plain terms

A connector usually has three moving parts: the source adapter (code that knows how to read the external system), an agent or orchestration layer that schedules crawls and handles incremental updates, and the ingestion into Microsoft Graph where items are stored and security‑trimmed. Microsoft provides a Connectors SDK and a lightweight connector agent so you can build custom connectors or run Microsoft’s prebuilt ones; the agent handles full and incremental crawls, delete detection, ACL stamping, and ingestion into Graph.

Real connector examples

Copilot connectors already cover a wide range of enterprise systems. Here are practical examples you’ll see in the wild and the kinds of prompts they unlock:

  • Salesforce – pull opportunity summaries and recent activity to prep for a customer call.  
  • Jira / Azure DevOps – surface sprint status, open bugs, or backlog items when you ask for release readiness.  
  • GitHub / GitLab / Bitbucket – summarise open pull requests, list failing CI runs, or extract changelog notes.  
  • Box / Dropbox / Google Drive / SharePoint – compare versions of a spec or summarise the latest product doc.  
  • ServiceNow / Zendesk / Freshservice – analyse incident trends and recommend process changes.

Each connector turns source content into indexed, searchable items so a prompt like “Summarise the top five customer issues from Zendesk this month and suggest fixes” returns grounded, citeable results instead of generic advice.

Connector versus MCP server versus calling an API

Copilot connector (indexed or federated). 
A connector’s job is to make content available to Microsoft Graph and Copilot.

Many connectors perform a crawl and index model (content is ingested into the Microsoft  Graph), while newer federated or user‑level connectors let Copilot fetch live data on demand using the user’s credentials. The indexed model is great for broad search and fast responses; federated connectors are useful when you need real‑time, user‑scoped access.

MCP server (Model Context Protocol). 

MCP is a protocol and server model that standardizes how AI assistants call external tools and fetch context in real time. An MCP server exposes a set of tools/endpoints that agents (like Copilot or Teams Channel Agent) can discover and invoke; it’s about runtime tooling and live interactions, not indexing. Think of an MCP server as a live automation or tool host — for example, a calendar MCP server that creates events or a custom MCP server that exposes internal business actions to an agent. MCP servers are registered so Microsoft 365 agents can discover and securely call them.

Direct API Calls

Calling a service’s API is the most basic integration pattern: your app authenticates and requests data or performs actions directly. That’s perfect for bespoke workflows or when you need full control over data flow, but it doesn’t automatically make that data searchable inside Microsoft Graph or available to Copilot across the tenant. If you want Copilot to use that data as part of its knowledge, you either build a connector that indexes it into Graph or expose it via an MCP/federated connector for live access.

How they differ…

  • If you index Jira with a connector, Copilot can quickly answer questions such as “What’s blocking Product release X?” using indexed issues.  
  • If you run an MCP server that exposes a ticket‑triage tool, Copilot (or an agent) can interface and directly open a triage workflow and assign owners in real time without needing to open the app.
  • If you call the Jira API from a custom app, you can build any workflow you want – but Copilot won’t automatically see that data unless you also surface it via a connector or MCP endpoint.

Security, governance, admin controls

Connectors are designed to respect tenant boundaries and Microsoft 365 governance. Indexed content is ingested under your tenant, encrypted, and security‑trimmed so users only see what they’re allowed to see. Admins manage connectors from the Microsoft 365 admin center, control which connectors are available, and can stage rollouts or disable federated connectors if they don’t meet policy. The connector agent and SDK also let you enforce crawl schedules, incremental updates, and ACL mapping so indexing behaves predictably at scale.

Connector, MCP server, or API?

When you need Copilot (or an agent) to work with an external service, you usually have three realistic integration patterns to choose from: indexing via a Copilot/Graph connector, exposing live tools through an MCP server, or calling the service’s API directly from your app. Each option solves different problems.

Below I will try to explain the tradeoffs in plain language, give concrete examples, and finish with a short decision checklist you can use immediately. NB: I’m still learning this myself so bear with me!

  • Use a Copilot connector when your goal is to make a lot of content searchable and citeable across the tenant so Copilot and Microsoft Search can answer questions quickly (think documents, tickets, PRs).
  • Use an MCP server when you need runtime tooling – live, discoverable actions an agent can call (for example: triage a ticket, kick off a workflow, or run a business action) rather than just read data. 
  • Call the API directly if  you are or need to build a bespoke app or workflow that needs full control, custom logic, or real‑time two‑way operations and you do not need Copilot to automatically see that data unless you also surface it via a connector or MCP. 

These are not mutually exclusive, and sometimes you might need a combination of both. For example you may index the data for search and also expose a small set of live actions via MCP or direct API calls.

Deciding which you need

I found this check list helpful which some of my developer friends use…

  1. What do you need Copilot to do? If it’s answer/search/summarise → connector. If it’s act/execute/triage → MCP or API. 
  2. Do you need tenant‑wide, security‑trimmed search? If yes, prioritise a connector. 
  3. Do you need live user‑scoped actions or interactive workflows? If yes, design an MCP server or expose specific API endpoints as MCP tools. 
  4. How fast do you need it? Real‑time → MCP/API. Fast search across many items → connector. 
  5. What’s your governance posture? If you need Microsoft 365 controls and auditing, connectors and Power Platform connectors give built‑in alignment; MCP requires explicit registration and security design.

The other tip from my limited experience so far (I did say bear with me) is

  1. Use prebuilt connectors first – there are loads. Check the Microsoft Copilot connectors gallery for existing connectors before building custom ones.  
  2. If you need to build (or get someone to build) a custom connextor, use the Connectors SDK and agent to build and test your connector; the agent handles crawling and ingestion. 
  3. Pilot / test with real prompts and measure relevance, latency, and trust (are results correctly security‑trimmed and useful?).

Posted on

Cisco updates contract terms in Response to Market Volatility

Cisco’s recent update to partner contract terms, prompted by rising memory prices, has caught attention across the partner and customer community. Of course any change that vendors make that has potential to impact pricing protection or order certainty will get the channel talking.

But taken in context – and with Cisco’s own commentary in mind – this looks less like a shift in philosophy and more like a pragmatic response to a market that remains anything but stable.

Market-Wide Challenge, Not a Cisco-Specific One

The backdrop here matters.

As Cisco’s global partner sales SVP Tim Coogan outlined in a note to partners, “the industry is dealing with sustained supply constraints driven largely by AI-led demand for memory and storage” . Production capacity simply hasn’t kept pace, leading to longer lead times and rising component costs across the server and infrastructure ecosystem.

This isn’t unique to Cisco. Other major vendors have already taken similar – and in some cases more aggressive steps (we have seen this and expect to see more in the devices and desktops space). Against this challenge, Cisco’s approach is arguably measured and something they need to do given the volume of hardware product lines this is likely to affect.

What did Cisco announce around pricing changes?

As Cisco announced to partners last week and the channel has loudly commented on, the key headlines were:

  • Cisco are changing partner terms and now reserve the right to cancel compute orders up to 45 days before shipment
  • Cisco also reserve the right to adjust pricing if component or manufacturing costs materially change.
  • Cisco are revising the quote validity but they have said they work work with partners to operationalise the changes and will clearly communicate the changes.

For partners like us, none of this is ideal as customers expect prices to be help subject to deal reg and partner pricing terms. As such this means partners will likely need to change their terms too.

Certainty and confidence is of course preferable, but there are some important nuances worth calling out in these annoucements.

Why are there changes happening?

One of the more overlooked aspects of Cisco’s position here is the why these changes are being made.

On Cisco’s Q2 earnings call, CEO Chuck Robbins was explicit that Cisco is leaning heavily into its supply chain scale and financial strength to secure memory supply – including advanced purchase commitments that have increased significantly year-on-year. They are also assuring partners that they will keep them up to date around pricing and supply chain challenges.

For partners, that matters a lot since it’s partners customers talk to about what is happening so important for partners to be kept up to date.

Any technology vendor that can secure supply, even at higher cost, is better positioned to fulfil customer demand than a vendor (compete) that simply freezes or cancels orders at the last minute.

According a CRN report, Cisco’s 45‑day cancellation window is actually more generous than policies being introduced by other vendors in the market. Cisco’s approach does means that partners should get:

  • Earlier visibility of potential changes impacting deals and projects.
  • More time to manage customer expectations and look at alternatives or options.
  • Greater confidence that committed orders are backed by real supply planning.

Of course from a customer perspective, the alternative to this approach is often worse.

The worst thing for an organisation is when they see poorly (or not communicated) silent price increases, delayed shipments, or last-minute order cancellations. These can erode trust far more quickly than upfront conversations about market volatility.

Cisco’s CFO Mark Patterson said that that the “we are adjusting terms to control what we can control” while staying close to market dynamics .

This transparency will hopefully gives partners the opportunity to have informed, proactive discussions with their  customers with statements backed up by the Cisco rather than fobbing them off. Something many organisations learned the hard way during the pandemic-era supply crunch with stuff cancelled, prices increased and supply chain lies.

Trust and Execution

Communication will of course be how Cisco are measured and trusted here (as will Cisco Partners).

Partners need clarity and will demand it from Cisco if it’s not given proactively, particularly around revised quote price protection periods — and they’ll need it quickly.

Parnters need to be able to meet there commitments too, so know whether a quote is valid for 45 days, 30 days or three days materially changes how customer conversations are framed by partners.

Final Thoughts

It’s not perfect, but it’s happening and expect others to follow.  In a market defined by uncertainty, pragmatism may be the most partner-friendly option available.

Will organisation change or shift vendors? Maybe. Will those others compete vendors be as open and honest with partners and customers? Who knows. Will organisations look to do more in cloud to mitigate these risks? Perhaps.

Posted on

Cisco Live Amsterdam 2026: AI‑Ready Networking to Reshape Enterprise Infrastructure

Cisco Live

Whilst I wasn’t able to make it this year,  Cisco Live EMEA in Amsterdam was full of annoucements and updates to their products sets. The message to customers and partners goes back to their roots “the network is now the backbone of the AI era” .

To that effect, Cisco unveiled a new generation of their silicon, systems, optics, software and collaboration devices designed to scale AI clusters, cut energy and operational costs, and bring real‑time AI and secure access into production.

For a mid-year event this was a huge set of announcements, so let’s take a look at these.

Cisco Live EMEA 2026 – Announcements at glance

Cisco Silicon One G300 and the new systems

With Cisco’s latest home-grown Silicon upgrades they can now deliver a whopping 102.4 Tbps programmable switching ASIC built to power gigawatt‑scale AI clusters and reduce job completion time while increasing GPU utilisation.

Cisco positions the G300 as the networking foundation for “agentic” AI workloads, with features such as a fully shared packet buffer, path‑based load balancing and proactive telemetry.

This is important of course for the global Hyper Scalers but also for enterprise organisations running or building AI workloads on their own data center hardware, since AI training and inference are increasingly limited by data movement. The new G300 and the G300‑powered Nexus 9000 and Cisco 8000 systems have been designed to treat the network as part of the compute fabric, improving throughput and reducing stalls that waste GPU cycles. Cisco said that organisations should expect a 28% improvement in job completion time in their benchmarks.

Optics and cooling for energy efficiency

Still in the DC switching space, Cisco announced new high‑density optics (including 1.6T OSFP and 800G LPO) plus 100% liquid‑cooled system options for the Nexus N9000/8000 family. Cisco claim these advances can improve energy efficiency by close to 70% for AI scale‑out deployments. which need increased cooling technologies.

This is an importnant impact with the awareness and concerns abou the engery consumption that AI data centers command. On top of that, energy and cooling are major operational costs for any organisation running AI data centers. Higher‑density optics and liquid cooling reduce power per bit and allow denser GPU/network co‑location, lowering TCO for large AI clusters.

Nexus One: new Unified management for AI fabrics

Next up – Cisco announced that Nexus One will now deliver a unified management plane intended to simplify fabric deployment, scale predictably, and operate securely across both on‑premises and cloud environments. Cisco positon that as a key step to remove operational friction when scaling AI data centers.

For any organisation that runs their own on-premises data centres, predictable scale and simpler operations are often the gating factors for enterprises that want to run real production AI workloads rather than experiments. The move towards a true single management plane reduces integration risk and speeds time to value.

Updates to Cisco SASE and AI Defense

Initially announced last year, Cisco announced what it calls its biggest update yet to their AI Defense and SASE stacks which forms part of the “Cisco Secure Cloud Suite”. This latest update brings AI‑driven detection, automated workflows and tighter integration between network telemetry and security enforcement across both “bought AI” and an inhouse developed AI tools and services.

Amongst the extra capabilities announced was the new ability to catalog all Model Context Protocol (MCP) servers an business uses in-house and elsewhere. This helps SecOp teams spot systems and AI agents that might not be caught by existing security policies and AI guardrails.

As organisations move past the PoC and experimental phase of AI agents and workloads proliferate and attack surfaces change. Combined with the amount of Shadow AI (such as consumer AI services in the workplace) and AI built into everyday apps, embedding AI into Cisco’s core detection and response and tying that to SASE policy enforcement helps organisations maintain security posture at the speed of AI.

New Webex Devices – Room Kit Pro G2 and Desk Pro G2

Announced initially at ISE 2026, Cisco have announced new AI‑ready collaboration endpoints in form of the Room Kit Pro G2, Desk Pro G2. Cisco are bringing real‑time speech translation in Webex that boldly claims to also preserves tone and emotion (something many AI translation services fail to do). Cisco also emphasised real‑time, human‑centric collaboration features powered by their mix of on‑device and cloud AI.
As Cisco continues to re-invest the Webex brand, hybrid work continues to demand better meeting experiences; real‑time translation and improved endpoint intelligence reduce friction and broaden collaboration across languages and locations.

Final Thoughts

Cisco’s announcements shift the conversation from “can we run AI” to “how do we run AI efficiently, securely and at scale.”

The core messaging from the keynotes [“the network is now the backbone of the AI era“] was clearly reminding their customers and parters of Cisco’s role in powering and securing networks for the AI era. AI was clearly the agenda and what Cisco aim to help customers achieve – shifting from speeds and feeds (although there was lots of speed and feed talk) to creating measurable business outcome opportunities. This is about faster model runs, lower operational cost, more sustainable IT, combined with stronger security for the AI era.

it certainly does feel like Cisco have got their innovation hats on and are once again doing what they did at the start of the cloud era – driving innovation and leading by example with a joined up, integrated approach. It’s a fun time to be a Cisco Partner. In fact, it’s a great time to work in the tech world – even though it feels like it’s almost impossible to keep up-to-date!

Posted on

2026 Role‑Based training for Microsoft 365 Copilot users

I’m a huge believer in role‑based learning because it gives people practical, relevant ways to bring AI into the work they already do. Wherever you are on your Copilot journey – curious beginner or confident user – these sessions can help you, your team mates or your friends understand how to get the best from Copilot in your line or work.

There’s FREE virtual training available for executive leadership, finance, HR, IT, marketing, operations, legal, and sales. Each session focuses on real workflows, real scenarios, and great examples

Training sessions commence from Tuesday 10th Feb, so take a look and register below. https://msft.it/6047Q3DET

Posted on

How OpenAI and Anthropic Just Triggered the Next Big Shift in Software

The last week has felt like a turning point in the AI landscape. Not because of a single product launch, but because two of the biggest players in the industry effectively declared that the next frontier isn’t chat, search, or reasoning. It’s agentic coding. It’s the potential to change the landscape of SaaS applications really quickly (especially for SMBs) which has caused the world to wobble a bit.

This agentic code (or AI Codex) is the word that has been buzzing tech news and social.

That said, the term has been around for a while, but last week it became shorthand for a new class of models designed not just to write code, but to understand systems, navigate repositories, fix issues, and build software with a level of autonomy that edges closer to genuine engineering assistance.

The second narrative and the “concern” that has been rising alongside it is the idea that AI isn’t just transforming software development – it’s going to (and is already beginning to) challenge the software industry itself.

Analysts and commentators have been exploring whether AI agents could eventually replace entire categories of SaaS tools, reshape how companies buy software, or even enable organisations to build their own alternatives on demand.

These two threads – the rise of agentic coding and the pressure on traditional software models – collided last week, and that’s why the conversation has become so loud.

What is an AI Codex?

AI Codex refers to a specialised family of models optimised for programming tasks. Where a general model like ChatGPT is trained to converse, explain, and reason across domains, Codex models are tuned to:

  • read and interpret codebases
  • generate new code with high precision 
  • refactor and optimise existing logic 
  • follow patterns, libraries, and frameworks accurately 
  • maintain context across multi‑file projects 
  • act as an agent that can plan, execute, and iterate on tasks

You can think of ChatGPT as a generalist and Codex as the engineer. Both are powerful, but they’re built for very different jobs.

Why It Exploded Into the News

Timing… There were three things that happened almost simultaneouslyast week and it was the timing if these as well as what happened that brought it front and centre to the tech news.

1. OpenAI and Anthropic launched competing coding models within minutes

It started with OpenAI releasing their latest Codex model at almost the exact moment Anthropic unveiled its new Claude coding variant.

Whether coordinated or coincidental, it created a sense of a head‑to‑head sprint. The press framed it as an AI coding arms race, and the narrative stuck.

2. OpenAI and Anthropic both claimed major breakthroughs in agentic coding.

This was the next thing.. Both Anthropic and OpenAI both positioned these are more than just incremental updates. Instead both said these models are a leap forward and their models can now:

  • plan multi‑step coding tasks
  • call tools 
  • run and evaluate code
  • fix their own mistakes
  • work across entire repositories 

This is essentially a shift from “autocomplete on steroids” to more of a a “junior engineer that can take a ticket and run with it”.

3. The question of could AI replace software?

A wider industry conversation about AI replacing software intensified last week too with commentators and analysts exploring the potential of whether these AI Codex models AI could eventually:

  • reduce the number of SaaS subscriptions companies need
  • automate workflows traditionally handled by enterprise software
  • enable organisations to build their own tools instead of buying them
  • reshape the economics of “seats” and licences

4. A viral comment poured fuel on the fire

Thank social media for this one as OpenAI’s Sam Altman’s remark that using the new Codex made him feel “a little useless” hit a nerve and sent it viral.

Of course, developers amplified it, commentators debated it, and suddenly the story wasn’t just about new models. It was about what these models mean for the future of software development. But it dominated tech feeds and news stories last week. So much I wanted to know more…

ChatGPT vs GPT‑Codex

So again.. For clairy…The simplest way to explain the difference between these two models is:

  • ChatGPT is built for natural language. 
  • GPT‑Codex is built for code.

ChatGPT is the base LLM most users are familiar with. Used by ChatGPT, Microsoft Copilot base LLM etc. It’s designed for most things… brilliant for language, strategy, communication, architecture, and ideation.

Codex is built to read your repo, understand your patterns, and produce code that fits your environment. It’s not for the general user.

Why This Matters for Engineering Teams and MSPs

The shift to agentic coding of course sparks concerns over the future role of the developers but both are trying to assure the world that it’s not about replacing developers, it’s about changing the shape of work.

This means:

  • faster delivery of repeatable patterns 
  • more consistent infrastructure‑as‑code 
  • automated remediation and optimisation 
  • better documentation and handover 
  • reduced toil across migration and modernisation projects

It also means of course that governance becomes even more important. If these models can now act and not just suggest, then organisations need a clear governance process to define the rules, guardrails, and patterns that define how AI participates in engineering workflows.

At the same time, the broader industry conversation about AI replacing software should be seen as a signal, not a threat. The opportunity for technology partners is to help organisations navigate this shift….(just as they are helping to adopt Gen AI tools like Copilot). This is still about  exploring where AI augments existing tools, where it replaces them, and where it enables entirely new workflows.


This is only the beginning.

The Real Story Behind the Headlines
The noise last week wasn’t just about OpenAI and  Anthropic competing (publically), it was about the potential for a new category of AI becoming mainstream.

Here, coding models are being treated not as assistants, but as participants in the software lifecycle – and that has implications far beyond engineering.

This is the beginning of a new phase in how software gets built, bought, and delivered
That’s why the term “AI Codex” suddenly feels everywhere. It captures the moment we moved from conversational AI to operational AI. From chat to action. From suggestion to execution.

What do you think? Fad or as SaaS apps going to be impacted?

Posted on

Create Agents in One-Click from your OneDrive 

Microsoft has now made it possible to create grounded knowledge “agents” directly from OneDrive. If you’ve not seen this yet, it allows you to select up to 20 OneDrive for Business files and create an agent you can use to ask natural‑language questions across all of those documents at once. You can even share it with others to use.


It’s works just like creating an agent from scratch that is grounded in specific OneDrive files or folders, but you can do it with a quick click create. This is a simple way to create an agent for reasoning over a bunch of project, customer or research files.

It lives directly in OneDrive folder as well as an agent pinned in Copilot Chat… It makes agent building (if you can really call it that very simple and quick).

Once created, it works like any other agent – meaning you can ask questions based on yoru files such as:

  • What decisions have we made so far? 
  • What risks keep coming up across these documents? 
  • Give me a summary of themes across all project files.

Microsoft describes the result as “complete, grounded responses”, and in the demos and tests I have done so far, that seems about right.

How to create an Agent from OneDrive

To Create an agent from OneDrive (Microsoft 365), head over to your OneDrive and choose a folder. Open it and simply click the “Copilot” button and then “Create an agent”.

One-Click to Create an Agent from your OneDrive files or library

Optionally, you can name the agent, change its logo, give it custom instructions, starter prompts and a role.

The agent creates a .agent file in the OneDrive folder.

When you open it, it installs the agent into the M365 Copilot pane which you then need to select to use it.

The agent responds and works like any other agent.

Do we need a OneDrive agent?

I think it demands. You could easily just create a agent and point it a these same OneDrive files. You could also use a Copilot Notebook and drag word files into the Notebook as references.

What the OneDrive agent does, however, from an ease of use and productivity standpoint, is makes it really simple (almost no thinking just click) to create a agent that works across the docs you need and want.

Being able to interrogate the entire project pack in one go is a meaningful step toward the “AI-powered workplace” Microsoft has been forecasting for the last few years.

For organisations already deep into Microsoft 365 and Copilot, this is a simple addition. It won’t change the world but does make Copilot a little more accessible in a really quick and simple way.

Posted on

Microsoft Q2 FY26 Earnings: Cloud & AI Power Record Growth

Microsoft has opened 2026 with a landmark quarter that firmly positions them as the global leader in enterprise cloud and AI adoption.

In the earning report, Microsoft has achieved milestones with cloud revenue surpassing $50 billion for the first time, record growth in Copilot usage, and major acceleration in commercial bookings, this earnings report signals a new era for Microsoft’s growth and might put the AI Bubble Burst chats to bed?

Top standouts were:

  • Revenue of $81.27 billion for FY26 Q2 ended December 31, 2025
  • Cloud revenue surpased $50B for first time, up 26% to $51.5B
  • Microsoft 365 Copilot seats up 160% with 15 million paid seats
  • Commercial bookings up 230%
  • Productivity & Business Processes up 16%
  • Intelligent Cloud grew 29%

Cloud Revenue Surges 26% to $51.5B

Microsoft Cloud revenue reached $51.5 billion, up 26% year‑on‑year – its strongest performance to date.

This is a pivotal number for analysts and CIOs alike, reinforcing Microsoft’s dominance as enterprises scale AI workloads, hybrid cloud infrastructures, and GPU‑accelerated services.

Supporting this momentum, their Intelligent Cloud revenue grew 29%, reaching $32.9 billion.

This remains Microsoft’s fastest‑growing segment, driven largely by Azure and AI services.

Microsoft AI: Copilot Growth Explodes

If 2024–25 marked AI experimentation, Q2 FY26 confirms enterprise AI at scale. Microsoft Q2 results were.

  • Microsoft 365 Copilot seats up 160% year‑on‑year, now at 15 million paid seats
  • GitHub Copilot subscribers up 75%

This data supports Microsoft narrative that more organisations are moving from pilot phases to widespread (well wider) deployment. These Copilot adoption rates are among the strongest indicators Microsoft has ever reported for a Copilot. They have previously avoided too much focus on the actual Copilot seats sold.

15 million paid Copilot seats (users) is a lot of seats… Let that sink in a moment!

Another interesting state was “the number of customers spending one million dollars-plus per quarter on Foundry grew nearly 80%, driven by strong growth in every industry….. over 250 customers are on track to process over one trillion tokens on Foundry this year. “

Long term AI Growth is strong

Q2 FY26 demonstrates that Microsoft has completed its transition from a cloud‑first business to an AI‑first hyperscaler, with every major metric — revenue, bookings, cash flow, and product adoption — propelled by AI demand.

Microsoft said the majority growth of their Azure Cloud was fuelled by AI Cloud Infrastructure and area Microsoft continue to invest. This is also reflected in the GitHub Copilot and Microsoft 365 Copilot growth.

Microsoft FY26 Q2 message shows that reinforces AI isn’t the future of Microsoft’s business  it’s the present, and the the future and it’s scaling fast.

You can read the earning report transcript here.

Posted on

Microsoft Teams Licensing Updates: More Premium Features for Everyone

Microsoft has announced some big changes to the Teams licensing model, aimed at making more  advanced features available to everyone along with updates to their Places products set which is now firmly more rooted as a key part of Teams.

These changes reflect Microsoft’s commitment to  democratising advanced collaboration tools and to keep adding value to their core products without the need for premium licenses to unlock core value.

These updates, effective 1 April 2026, will impact organisations of all sizes. Here is what is changing.

1.Microsoft Places Now Broadly Available

Yes, I thought they had forgotten about it as well! But not..

Microsoft Places which was originally only part of Teams Premium will now be available to anyone with access to Microsoft Teams and Office users making it a core subcomponent of Microsoft 365 for anyone with the following license.

  • Microsoft 365 E3 / E5
  • Microsoft 365 Business Basic / Standard / Premium
  • Office 365 E1 / E3 / E5
  • Exchange Online

What is Microsoft Places?

Microsoft Places is a set of apps and services designed to provide rich meeting and space management to enhance and enable the hybrid meeting experience. It includes

  • Places Finder: Provides rich meeting location details, including images and floorplans.
  • Places Explorer: Adds map-based reservations, people, and space visibility.

The key value of places (and why it needed to just be part of Teams) include the ability to auto rebook rooms if more people are attending in person that the room can fill, notify you if you are joining a meeting remotely when you and others are in the same building and tools to help you navigate and find places, people and spaces more effectively in the workplace.

2. Teams Shared Devices becomes Teams Shared Spaces

This change of name from the Teams Shared Devices license to Teams Shared Space also provides IT with the following management capabilities under the common name. It includes.

  • Bookable desks
  • Teams panels
  • Shared spaces


There also new capabilities which include:

  • Space management
  • Desk booking
  • Space analytics
  • And more integration with third-party APIs including Building Management systems.

The ability for Places to integrate with (more) 3rd-party sources of spatial data, floorplans, check-in signals should make adoption better by streamlining the onboarding and management of the spaces in the Places Directory.

3. Town Halls and Webinars for Enterprise

Previously exclusive to Teams Premium Users only, Town Halls and Webinars style meetings  will now be available to all Teams Enterprise customers. With this everyone gets default meeting capacity or up to 3,000 attendees and up to 10,000  for stream only events.

Microsoft have also introduced “attendee packs”
2hich can be purchased to increase limits up to 100,000 attendees

This is a major enhancement for organisations hosting large-scale events without needing Teams Premium Licenses.

Read more here.

https://techcommunity.microsoft.com/blog/microsoftteamsblog/licensing-updates-extend-access-to-advanced-capabilities-in-microsoft-teams-and-/4488312

Posted on

M365 Copilot Image Generation Levels Up, and Video Summaries are coming to Copilot Notebooks

Microsoft has kicked off 2026 with two significant enhancements to Microsoft 365 Copilot – both providing a clear signal to how improvements in multi-media creation in AI will support creativity, communication, and knowledge sharing across the workplace.

The first is a major upgrade to Copilot’s image generation capabilities with the rollout of OpenAI’s GPT‑Image‑1.5 model. The second is a new capability that turns Copilot Notebooks into automatically generated video summaries (in addition to the voice / podcast over views). These are both in preview for organisations enrolled into the Copilot Frontier Preview.

Whilst subtle – these make Copilot more useful, more expressive, and more multi-modal in the flow of work. Read on for more detail.

GPT‑Image‑1.5 Creation comes to Microsoft 365 Copilot

Microsoft is good at getting new features into Copilot Quickly. Just before Xmas 2025, Copilot was updated to support the latest GPT5.2 models and now they are replacing OpenAI’s GPT‑4o with GPT‑Image‑1.5 across the Copilot’s image generation experiences. This will gradually roll out through January 2026. This includes Copilot Chat and the wider “Create” module in Copilot.

For organisations already using Copilot to create internal comms assets, presentation visuals, campaign concepts, or quick mock‑ups, this upgrade is most welcomed. The quality gap between “AI‑generated” and “designer‑produced” continues to narrow, and the speed improvements make Copilot even more viable for rapid ideation. Image creation in AI tools has come on massively in just a few months.

One of the key upgrades is the ability to updates aspects of an image (via a prompt) or take over and edit with Microsoft’s Image Designer Tools directly from Copilot. You can see the difference in the example below (you’ll need to zoom in sorry)!

What is GPT-Image-1.5?

This latest model from OpenAI is their answer to Google’s highly regarded Nano Banana image models – This is (according to experts), “on par” in terms of fidelity, instruction following, and realism, plus it’s included at not additional cost to Microsoft 365 Copilot users.

According to Microsoft, once rolled out, users can expect:

  • Sharper prompt adherence – especially for composition, style, and on‑image text
  • More precise region‑specific editing with fewer unintended changes
  • Higher‑quality visuals with more realistic lighting, textures, and detail
  • Faster generation – up to 4× quicker for many prompts
  • Better consistency when iterating on faces, colours, and lighting.

Video Overview in Copilot Notebooks

Microsoft is introducing Video Overviews – the ability for Copilot to automatically generate a short, narrated video summary of a Notebook’s content. This adds to the current audio overview feature and is rolling out to organisations enrolled in the Frontier Preview, Other organisations will get this in due course – keep an eye on the official Roadmap for this one.

This enhanced the existing overview feature, allowing Copilot Notebook users to:

  • Analyse the full Notebook
  • Extract key insights
  • Generate visuals
  • Produce a narrated video summary in first person, interview or podcast style

Think of it as a dynamic, visual executive summary – ideal for sharing updates, explaining concepts, or turning long‑form thinking into something more digestible.

Copilot Notebooks are a powerful space for iterative thinking, brainstorming, and structured problem‑solving in solo or shared mode. But they’ve also been static, plus you still had to read them. The ability to have audio or video overviews make these much more digestible, quicker to consume and are great for helping consume content in their preferred way.

Whilst Google’s NotebookLM has had this feature for a month or so, this is the first time Copilot is turning your content into multimodal output without requiring any video editing skills (or even prompting) at all. It’s a glimpse of a future where:

  • Documentation becomes auto‑summarised
  • Content is consumable in ways that meet the user’s need.
  • Knowledge becomes more accessible

I’m personally really interested to see how well Copilot handles narrative flow, visual selection, and pacing. If Microsoft gets this right, it could become one of the most impactful features in the Copilot.

Summary

In short, these subtle but impact updates point to the same trajectory of where Copilot is heading – becoming a fully multi-modal assistant, not just a text‑based one.

  • GPT‑Image‑1.5 – higher‑fidelity visual creation
  • Video Overviews – automated multimedia storytelling
Posted on

Satya Nadella’s Call for an “AI Reset” – What Business Leaders Must Prioritise in 2026

Cut out image of Sayta Nadella

As I finished my first week back to work of 2026, I was thinking about Microsoft’s Satya Nadella’s end‑of‑year reflection blog post “Looking Ahead to 2026“ which he wrote at the end of December 2025.

This for me landed at an important moment for the industry. After a year dominated by “AI slop,” concerns over whether the AI boom was ending and stock prices in key AI stocks becoming a bit rocky. In the Microsoft space, there has been some “backlash” about how much AI Microsoft are “pushing” in their products and the industry in general failing to really get big AI initiatives off the ground. We then of course have the upcoming chip shortages (which remind us back to COVID times) with AI cited as the cause of these shortages (memory especially) with AI Data Centres demand exceeding supply.

We have seen huge advances in AI models from Open AI, Google, Anthronic and Microsoft this year but the news has really been coined with the term AI Slop – with AI appearinf to be baked into everything whether people want it or not – who ever thought Notepad in Windows would get the Copilot treatments!!

Despite the negative press, AI is still very much the buzz word (heck, its driving all those component and chip shortages). At work, our clients are still talking about it, driving it forward and investing more into adoption, enablement and consultative engagements. Businesses remain “excited” about the opportunity AI can bring to businesses, consumers and the world!

In his blog post, Satya Nadella argues that we’re finally moving from discovery to diffusion. The message is clear: 2026 must be the year we separate spectacle from substance.

This rest of my blog looks at Nadella’s framing, the wider industry commentary, and what it means for organisations that need AI to deliver measurable value – not just noise with AI Slop or no slop!

From “bicycles for the mind” to “scaffolding for human potential”

In his blog, Satya Nadella revisits Steve Jobs’ famous metaphor – computers as “bicycles for the mind” – and argues that it no longer captures the scale or nature of modern AI. Instead, he positions AI as scaffolding: a structure that supports, amplifies and extends human capability rather than replacing it.

This is a subtle but important shift. It reframes AI not as a tool we operate, but as a system that surrounds us — shaping how we work, decide, and collaborate.

Coverage from the Economic Times reinforces this by highlighting the need to move beyond the novelty phase and focus on real‑world impact. In short, the industry must stop debating “slop vs sophistication” and start designing systems that genuinely improve human cognition and outcomes.

From what I see with my engagement with business leaders, it means moving away from app and model‑centric thinking (what is better ChatGPT or Copilot) and towards systems engineering within our line of business. We need to look at how our business works today, where AI fits and where automation counts. Businesses need to work with Microsoft, with their technology partners and most importantly from within, to plan and deliver measurable outcomes.

“Aiming to please” is not alignment

One of the most important – and least discussed issues in business led AI is the tension between:

  • AI that aims to please (reinforcement learning from human feedback, training, safety tuning, helpfulness scoring) where AI is just that assistant, that Copilot!
  • AI that is aligned (truthful, reliable, predictable, value‑consistent and therefore really useful) which is much more important as we look for it to help us make decisions, take action and do stuff!

These are not the same thing.

Research throughout 2025 showed that models tuned to be more “helpful” often become more sycophantic, more agreeable, and more likely to produce confident but incorrect answers. An interesting article I found from “PC Gamer” cites Microsoft‑linked research suggesting that over‑reliance on AI tools can actually reduce user capability over time due to the mind set of “AI can do anything” whilst at the same point “we don’t trust AI”. So, which is it?

This re-enforces Satya’s view that as we enter 2026, the priority must shift from “does it sound good?” to “does it stand up to scrutiny?”

The AI race is real – and so is the concentration of power

AI is now a currency and we are seeing this everywhere. Leaders cite it, big tech companies are embracing it and the models are doubling in capability in less than a 12 month cycle (some say 8 months). Compute, data, and model access are becoming strategic assets – we are seen this through component shortages! These global hyper-scalers and a handful of model labs will capture disproportionate value, and enterprises will increasingly depend on them for:

  • model access
  • agent frameworks
  • safety and alignment layers
  • orchestration platforms
  • compute and optimisation pipelines

This shouldn’t be seen as negative, but it does require deliberate strategy. Vendor lock‑in, opaque safety layers, and proprietary agent ecosystems will shape enterprise risk profiles for years.

Then there is the vast number of other tech venders embracing the security and safety layers – in the dozens these are and will appear to sell you AI Safety and Security tools, while we wait for the leading AI models and companies to bake this is. Firms like Microsoft will likely do this well and make it part of their security and compliance approach across their wider platforms. others will or may lean to other vendors through partnerships.

The biggest gap perhaps is still user awareness, adoption and controlling shadow-IT. This becomes very more important as organising and the world begin the next pivot shift from AI assistants to autonomous AI.

Again, Satytas’ blog acknowledges this indirectly: the industry does not yet have “societal permission” for widespread autonomous AI. That permission will depend on trust, governance, and measurable outcomes – not just capability.

Two strategic paths for Corporate and Enterprise AI?

We look at two different paths for how organisation use and leverage AI. The oath for superior intelligence and the path that recognises AI’s purpose as cognitive scaffolding.

Path A: Pursue superior intelligence

  • Focus on emergent capabilities and autonomy
  • High potential upside
  • High systemic risk
  • Harder to govern, audit, and predict

Path B: AI as cognitive scaffolding

  • Focus on augmentation, reliability, and workflow integration
  • Measurable value
  • Lower risk
  • Requires strong systems engineering

Satya calls out that most organisations will need to operate somewhere between these paths, but that the balance must be intentional.

What IT leaders should prioritise in 2026

There have been lots of posts around this from many different players, analyst firms and now little ol’ me! For me, nothing much has changed since last year (though the models have got cleverer, faster, cheaper!)

Outcome‑based evaluation: Benchmarks and demos are not enough. Businesses need to look at real use cases, go deep with proof of value production pilots. they need to determine what good looks like. It will likely require solutions/use cases that deliver task‑level accuracy, has no (minimal) hallucination, and longitudinal user‑impact studies which they can be loud and proud of.

Human‑centred design: This partly means no AI for AI’s sake. That said user familiarisation and confidence (like Internet Skills once were) are key for trust and understanding Departments need to surface provenance, uncertainty, and verification steps in putting AI to work. Moving beyond assistive to authorative AI should not be done lightly – trust is everything and humans being in the loop is vital in order to avoid “auto‑execute” defaults for high‑risk actions.

Governance and entitlements: IT and compliance need to have (and enable in the case of Copilot for example) auditing to track which agents can access which data and enforce least‑privilege by default. Audit everything, Trust nothing, Training everyone and Pilot, Pilot, Pilot, test, test, test.

Red‑team testing and observability – This goes beyond security and governance – this is about trusting AI to act like a human would. Use cases need to ensure they test for sycophancy, bias, adversarial prompts, and silent failure modes – for AI to world in place of human roles, there needs to be trust, failback and no “time outs” in the middle of a conversation with an agent.

Skills protection – Training is key to ensure organisations pair automation with training to prevent deskilling of people and upskilling of how inference and AI models work. We need to document everything, know what expected outcomes are and re-evaluate each business process, AI model and task our AI is performing. We need to preserve the entity of what our business is about this is the human factor) and retain our rooted institutional knowledge.

Final reflection

Sayta Nadella’s call to move beyond “AI slop” is timely and what many leaders are thinking. It’s hard to beleive it is less that 2 years since Copilot make it’s general availabiloity to the masses and the industry has spent these two years chasing and demaning more capability, bigger models, flashier demos, and viral outputs.

As we sit here, closer to 2030 than we are to 2020, 2026 demands something different. AI needs to deliver on the promises. This is less about the AI and more about how organisations re-shape, re-invest and rebuild business modes, systems, and processes with AI in mind. Technology change is not quick and simple and AI is not (in most cases) not a quick fix answer to underlying issues. For IT pros and business leaders, the message is clear:

  • Demand measurable value in pilot and projects
  • Prioritise trust and governance and use a model/tool that fits into what you have
  • Treat AI as scaffolding, not a substitute for good processes and people
  • Build for augmentation, not automation (in the immediate term)
  • Evaluate outcomes, not optics and do in phases.

The AI race will continue to get faster, we will hear more success stories, more doom and gloom, feel left behind because we are not moving fast enough and people will worry more about job losses, environmental impact and cost. The economics will only intensify, but I believe that the organisations that win will be those that operationalise AI responsibly, at the right pace, deliberately, and with a clear understanding of where the real risks and opportunities lie.

Thanks for reading or listening. Happy 2026!

References / Sources in this doc

  • Economic Times — “Microsoft CEO Satya Nadella calls for a big AI reset in 2026…”
  • PC Gamer — “Microsoft CEO Satya Nadella says it’s time to stop talking about AI slop…”

Posted on

What is Copilot Checkout? Microsoft and PayPal’s new AI-Powered Commerce Experience.

Microsoft and PayPal have officially joined forces to launch Copilot Checkout, a groundbreaking integration that redefines the online shopping experience. This collaboration merges Microsoft’s conversational AI capabilities with PayPal’s trusted payment infrastructure, creating a seamless journey from product discovery to purchase all within the Copilot interface on Windows, app and the Web.

It’s rolling out initially in the USA and works with partners including PayPal (and later), Shopify, Stripe, and Etsy.

Intelligent Discovery Meets Trusted Payments

Copilot Checkout will allows users to browse,  evaluate, and complete purchases (check out) directly from within Microsoft Copilot experience, eliminating the need to switch between different apps, devices or websites. By embedding PayPal’s agentic commerce services, the experience becomes not only faster but also more secure and flexible.

Key Features of Copilot Checkout

  • Conversational Shopping: Users interact with Copilot to explore products, compare options, and make informed decisions.
  • Integrated Payments: PayPal will power the end to end checkout flow, offering branded and guest checkout, credit card support, and even PayPal wallet integration.
  • Merchant Enablement: Retailers are able to surface and promote inventory, and reach high-intent customers at the moment of decision without loosing to the “buy later crew”

Benefits for PayPal sellers and consumers

For sellers:

  • Let’s them reach customers during high-intent moments
  • Reduce friction and cart abandonment and avoids the buy later mind set.
  • Expand visibility across Microsoft touch points such as Bing, Edge, and of course Copilot.

For Consumers:

  • Enables fast, flexible and in the moment checkout options (Amazon experience)
  • Stay within the Copilot experience from start to finish to ensure customers do not have to shift platforms, devices or apps.
  • Benefit from PayPal’s buyer protections and payment versatility.

Early Adoption and Ecosystem Impact

In this early stage of the partnership the early adopters include major brands including Etsy and Urban Outfitters.

What’s more Shopify merchants are auto-enrolled, while PayPal and Stripe sellers can opt in for this experience (Microsoft will need to convince them and sell the value).

This rollout marks a significant step in the evolution of agentic e-commerce, where AI agents act on behalf of users to simplify and accelerate transactions.

PayPal and Microsoft Partnership.

In the annoucement, Michelle Gill, GM of Small Business and Financial Services at PayPal, said:

“By integrating PayPal’s agentic commerce services with Copilot’s intelligent shopping platform, we are enabling seamless, reliable transactions for both merchants and consumers.”

Kathleen Mitford, CVP of Global Industry at Microsoft shared:

“Retailers can automate what slows them down and amplify what sets them apart.”

What and when?

Copilot Checkout will be come more than just a new hidden feature. Following OpenAI (and expect Google to follow), Microsoft and PayPal are looking to reshape consumer behaviour and drive more footfall to Copilot and PayPal. This is just the start and Microsoft plan to work with other e-commerce providers.

The joint goal is to redefine how brands connect with customers in real time, with intelligence and trust at the core from both PayPal and Microsoft.

For more insights, visit PayPal’s announcement or explore Microsoft’s commerce vision.

Posted on

What is Copilot Real Talk Mode? And how to use it.

Back at the “Fall update in October”, Microsoft announced that a new talk mode called “Real Talk” was coming to Copilot. This has been available in the US for a while in preview but is now available in the UK and other regions.

This (currently) optional chat mode allows Copilot to better adapt to the user’s tone by adding more personality, wit, and even “playful” challenges in its responses. This new mode “real talk” offers a more balanced and engaging conversational style that is designed to be more honest like a true “friend” whilst still staying polite. One of the critiques of many chat bots is that they “tell you what you want to hear”. Real Talk, aims to address this by providing a more balanced, open and honest conversation but still without being rude!

Jacob Andreou, Microsoft’s CVP of product and growth for AI, talks about this mode as being able to understand and “mirror” the user’s attitude, offer its own perspective, and not simply agree with everything said by the human. “Real talk aims to create a more interactive, human-like dialogue that goes beyond providing straightforward information. It will provoke users to think by occasionally questioning or offering alternative views, making conversations with Copilot more stimulating“.

How to use Copilot Real-Talk mode

This new mode is accessible through a dropdown menu and is currently limited to text chats (so not yet available in Copilot’s voice chat interface).

Copilot Real-Talk Moe

It’s available in the Windows Copilot App, Web (at https://copilot.microsoft.com) and mobile.

You can also switch to Real Talk mode at any point, but this does start a new conversation. To do this simply click the drop down and change the voice mode:

Copilot Real-Talk in Action

In this example, I gave Copilot a simple statement where I said, “Many people still think the earth is flat – what do you think about that?“. I run this in Copilot’s default chat mode and then again in “Real Talk” mode.

I’ve also created a video on this where I go into more detail.

Copilot Standard Chat Response

The response tone and length was quite typical – a balanced view, not wanting to say the wrong thing and looking at both angles whilst still stating facts.

"It’s a fascinating example of how belief can diverge from evidence.

From a scientific standpoint, the Earth being spherical isn’t up for debate. It’s supported by centuries of observations and measurements, including satellite imagery, GPS systems, gravity measurements, time zones, airline routes, and even how shadows behave at different latitudes. These aren’t abstract theories; they’re practical realities that modern life depends on every day.

What is interesting is why flat‑Earth beliefs persist. They tend to be less about geography and more about psychology and trust:

- Distrust of institutions – Some people see flat‑Earth ideas as a way to reject governments, scientists, or perceived elites.
- Community and identity – Belonging to a group that “knows the truth” can be powerful.
- Misunderstanding of science – Scientific methods can feel opaque or intimidating, making alternative explanations seem appealing.
- Algorithmic reinforcement – Online platforms can amplify fringe ideas by repeatedly showing similar content.

I don’t see flat‑Earth belief as a failure of intelligence. It’s more a failure of critical thinking skills being nurtured and a breakdown in trust between experts and the public. When people feel excluded from knowledge, they’re more likely to reject it entirely.

The real challenge isn’t correcting the shape of the Earth; it’s rebuilding curiosity, scientific literacy, and trust in evidence-based reasoning. That’s a much harder problem—and a far more important one."

Copilot Real-Talk Mode Response

This was much more concise and to the point. What do you think and what do you prefer?

It's wild that in 2025, with satellite photos and physics we use daily (like GPS literally requiring round-earth calculations), some folks still reject basic science. The psychology behind it fascinates me though—it's less about evidence and more about wanting to feel special, distrust of institutions, or finding community in shared beliefs.
Posted on

Copilot gets free GPT‑5.2 upgrade with “Smart Plus” mode

Microsoft has begun rolling out GPT‑5.2 across Copilot on the web, Windows, and mobile as a free upgrade. It sits alongside GPT‑5.1 rather than replacing it, giving users a clear choice between speed and deeper reasoning.

GPT‑5.2 is described as OpenAI’s strongest model series so far, designed to accelerate real work: building spreadsheets and presentations, writing and reviewing code, analysing long documents, using tools, and working with images. Copilot only moved to GPT‑5.1 for Smart mode in November, and Microsoft is now introducing GPT‑5.2 as a new “Smart Plus” option.

GPT5.2 gives higher‑reasoning model for complex tasks

The updated version of Open AI’s GPT arriving in Copilot is the “Thinking variant” of GPT‑5.2 – the same model Microsoft highlights for complex, multi‑step tasks and strategic reasoning.

GPT5.2 – Benchmark performance data

OpenAI’s own release notes describe GPT‑5.2 Thinking as expert‑level for well‑specified office tasks and significantly more reliable than previous models. Industry benchmarks reinforce this:

  • On knowledge‑work tasks across 44 occupations (GDPval), GPT‑5.2 Thinking beats or ties industry professionals 70.9% of the time (vs 38.8% for GPT‑5). 
  • Coding performance is significantly higher, scoring 55.6% on SWE‑Bench Pro and 80% on SWE‑Bench Verified — both ahead of GPT‑5.1 Thinking. 
  • It posts strong reasoning scores: GPQA Diamond 92.4%, AIME 2025 100%, and CharXiv Reasoning with Python 88.7%.

Microsoft Makes Model Choice a core Copilot feature

Microsoft has also rolled out GPT‑5.2 into the commercial Microsoft 365 Copilot and Copilot Studio products, where it appears in the model selector alongside GPT‑5.1 and GPT‑5.2 Instant. This aligns with Microsoft’s broader strategy of giving users explicit model choice depending on the task.

In Microsoft 365 Copilot, GPT‑5.2 connects to Work IQ (aka Microsoft Graph) – Microsoft’s intelligence layer that reasons across meetings, emails, documents, and organisational data to unlock insights and strategic planning scenarios.

More Value for Copilot users

Microsoft frames this as part of its ongoing commitment to model choice across Copilot experiences, ensuring users can pick the right model for the job rather than relying on a single default. For consumers this is huge value as Open AI still make the 5.1 and 5.2 models exclusive to premium subscribers whereas Microsoft give the latest models for free (both in Windows 11 and on the web and mobile apps).

For the non commercial / enterprise version, Smart Plus becomes the “high‑reasoning” lane inside Copilot. It’s designed for:

  • Multi‑step workflows 
  • Strategic analysis and planning 
  • Complex document understanding 
  • Code generation and review 
  • Tasks that benefit from slower, more deliberate reasoning 

GPT‑5.2 Instant remains the fast, everyday model, while GPT‑5.2 Thinking powers Smart Plus for deeper work.


Posted on

Automatic Alt Text on Copilot+ PCs: A Small Feature with a Big Accessibility Impact

Microsoft has rolled out a useful update to the Office Suite apps like Word, Excel and PowerPoint which creates automatic, “on‑device” Alt Text generation for images. This is a great way of helping content become more inclusive by providing automatic (but editable) descriptions of images, graphics, graphs etc. This helps screen readers and also AI tools better understand documents.

Alt Text Auto Generation is for Copilot + PCs only

This great feature leverages the on-chip NPU of Copilot+ PCs and is currently not available on older (non-AI) PCs. Sometimes the most valuable improvements are the ones that make good practice the default — especially for accessibility and inclusion.

Image of someone with hands in air emotionally speaking

This is a great way of ensuring content becomes meaningfully accessible by default.

You don’t need to do anything to enable this feature. You will need to be running the latest version of the Office apps on your device.

  • Word, Excel and PowerPoint now generate Alt Text instantly as you insert images.
  • The processing for this happens on‑device, so nothing is sent to the cloud.
  • It’s currently only for Copilot+ PCs, taking advantage of the NPU
  • Authors can still edit or override the generated text, keeping humans in control.
Posted on

How to get free security updates for Windows 10

If you are a home/consumer user using Windows 10 – because you are unwilling to, or unable to (due to hardware restrictions) to upgrade to Windows 11, and not able or wanting to buy a new modern PC, this blog shows you how to take advantage of Microsoft’s free Security updates for Windows 10 until 13th October 2026.

First…head over to Windows Update. Here you may be offered the last Windows Update, but will also see the option to “enroll in extended security updates”.

Click [enroll now].

Microsoft will check your eligibility – must be Windows 10, must not be a corporate device (therefore not Active Directory or Entra ID Joined) and must be a licensed version of Windows.

Assuming you see the message above (which you should), click [Enrol].

Thats it – enjoy free Windows 10 updates until 13th October 2026 whilst you get ready to upgrade, buy a new device or move off Windows to another OS as Windows 10 will be 11 years old when your free security updates expire…

Windows 10 came out in 2015…..that’s an old OS now – older than COVID-19 🙂

Posted on

GPT-5.2 now available in Microsoft 365 Copilot

Microsoft has just (11th Dec) started rolling out OpenAI’s GPT‑5.2 across Microsoft 365 Copilot and Copilot Studio, marking another significant leap in AI-powered productivity.

The differences between GPT5 and GPT5.2 provide “a significant upgrade in performance across various metrics”. For example, GPT-5.2 has achieved higher scores on benchmarks like ARC-AGI2 and GPQA Diamond which indicate improved abstract reasoning and scientific knowledge. In coding tasks, GPT-5.2 outperformed GPT 5.0 and 5.1 in accuracy and speed. The model also excelled on CharXiv Reasoning an AIM 2025 scores which measures LLMs capabilities in advanced mathematics and problem-solving.

The update delivers two models under one banner:

  • GPT‑5.2 Thinking, ideal for deep reasoning and complex problem-solving, and
  • GPT‑5.2 Instant, tailored for everyday tasks like writing, translation, and learning – all now integrated into Work IQ for actionable insights across emails, meetings, and documents.

    Within Microsoft 365 Copilot, (as this rolls out), users are able to manually select GPT‑5.2 from the model menu in both Copilot Chat and Copilot Studio, enabling smarter decision-making for tasks such as:
  • Preparing insights ahead of meetings.
  • Conducting comparisons and analysis such as market research of reviewing reports.
  • Extracting strategic takeaways linked to objectives and milestones.
Model Selection for GPT5.2 in Copilot Chat

Microsoft has sad that Copilot Studio agents will transition automatically from GPT‑5.1 to GPT‑5.2 in early-release environments and early in 2026, the default model in Copilot Chat will shift from GPT 5 to GPT5.2

These improvements therefore deliver improvements for tasks requiring complex reasoning and problem-solving skills. How their fair in day to use will be down to you as user to evaluate and determine. For now – we get the choice to dip in and out.

This rapid launch reinforces Microsoft’s commitment to offering model choice allowing users to access the latest innovation tuned for enterprise-grade security, compliance, and performance.

Posted on

When the Cloud Sneezes: a look at the ‘Outage Season’

The past few months have been a bruising reminder that even the biggest cloud providers can stumble. AWS, Microsoft Azure, and Cloudflare have all suffered major outages, disrupting any services that rely on them from websites to shopping sites to CRM and Finance systems and AI tools. For businesses (and their customers / users) there cause huge problems, lower confidence, impact services and reputation. Many are asking what happened, why and what can be done about it?

The Outages – What Happened?

There’s been a fair few this autumn, some causing performace issues, others huge impacts bringing  down lots of services. Most recent examples include:

  • Amazon Web Services (20 October 2025): A DNS automation bug in the US‑East‑1 region corrupted internal records for DynamoDB, cascading into failures across Lambda, API Gateway, and thousands of downstream apps which then replicated across other regions. 
  • Microsoft Azure (29 October 2025): A faulty configuration in Azure Front Door’s CDN nodes caused global downtime, impacting Microsoft 365, Xbox Live, and airline systems. 
  • Cloudflare (two recent ones – 18 November & 5 December 2025): According to their support pages, the first outage was the result of a bot‑management configuration file which got too big and grew beyond expected limits, crashing traffic‑handling software worldwide. The recent one last week, was different, caused by a firewall update triggered a bug, disrupting almost a third of (29%) of HTTP traffic globally. 


Whilst we hear a lot about cyber threats (similar to the CrowdStrike problem a couple years ago) these were not cyber-attacks or capacity limit failures. They were internal configuration and metadata errors. This may have been bad change management, inadequate fail back and validation controls or something else!

What are these Providers doing about it?

As you’d imagine, these major cloud providers need to tell their customers what is going on, provide assurance as to their stability and ability to recover. Following these, each have (and remember every cloud service does suffer outages from time to time for a number of reasons) has a requirement to update customers on the root cause and long term strategies to improve resilience. As an example:

  • AWS have said they are rolling out their “Route 53 Accelerated Recovery” and will be partnering with partnering with Google Cloud for multi‑cloud failover. This will be part of their premium resiliency services.
  • Microsoft have committed to strengthening their change‑management processes, checks and controls and are publishing detailed post‑mortems.
  • Cloudflare, in similar vein to Microsoft, are adding extra guardrails to monitor more controls and configuration parameters to prevent issues such as oversized configuration files. They are also committed to improving their Web app firewall testing and failover services.

How can organisations can be better prepared

It’s always tough when a business relies on Cloud providers to host and power their business as their hands can be tied and them left a little helpless when outages or performance issues occur – often it’s the visibility (or lack of) of what is happening that is most daunting for organisations and for IT. Whilst the benefits of cloud are not being disputed here, businesses can find themselves in tricky situations, having to answer to their customers, the board and shareholders as outages to their business can cause everything from mild inconveniences to value and reputational damage.

A multi cloud strategy can of course help, mitigating some of the impact of an outage of one, a hybrid (on prem/cloud) approach can do the same, but each of these add huge amount of complexity to their infrastructure management and cost so it’s about weighing up the impact and cost vs impact cost.

So, what else can organisations do then?

Visibility and Early Warnings

With any cloud service, it is often the visibility (or lack of) of what is happening that is most daunting for organisations and for IT. Just like tsunami warnings or those sniffles you get that let you know you have a cold or flu coming, there are services that can provide a holistic view across all your cloud services and help IT understand the continuous, predicted and previous performance and reliability of such services.

Cisco ThousandEyes as an example, is an incredible powerful service that provides end‑to‑end visibility across Internet Service Providers (ISPs) SaaS, and all cloud providers.

It helps detect outages early and quickly, can pinpoint root causes (DNS, CDN, SaaS, routing), and prove whether performance issues or service outages are external or internal. There are also plug-ins for desktops devices and browsers that can do this for remote and cloud users.

ThousandEyes also offers digital experience monitoring for apps like Microsoft 365, Service Now, Salesforce, and Zoom, and they have AI‑driven insights to flag risks before they cascade.

It can’t fix it as such, but if do have failovers, contingencies or DR in place, it can help organisations prepare, understand where the issue is and how to help communication with leadership, customers and other stakeholders and sooth the stress (slightly) of diagnosing the issues.

Failover Services

Of course, visibility and understanding are great but depending on how big the risk and cost of these outages, many organisations (many giant content providers do this) are looking at ways to add more contingency, failover, etc.

Where the issues are with DNS or content delivery providers, many organisations look at leveraging Multi‑CDN/DNS providers with auto Failover. This can help keep traffic flowing by rerouting when one provider fails. 

These tools often provide their own native cloud monitoring. Whilst not as in-depth as Thousand Eyes, they will typically provide dashboards that provide telemetry and status of these services and can instigate failover automatically or on-demand.

Cloud providers also provide their own status and performance monitoring tools often with in-depth telemetry but limited of course to their own platforms.

Chaos Engineering

This is really the approach, organisations will ypically employ as part of their DR/BC planning. Chaos Engineering is essentially the process organisations use to simulate outages to enable them to validate (and improve) their recovery plans before the real thing happens. This of course needs cultural buy-in, a continuous improvement approach and awareness across different aspects of their environment to ensure the impact of such outages can have. Again – visibility tools can help.

Insurance Safety Net

Even with resilience tools in place, redundancy and well architected fail-over solutions, outages will always happen, and this can mean lost revenue. Many “tech” insurers offer cyber insurance with many also offering “business interruption cover”, compensating for lost earnings due to outages caused by IT fails. I’ve not gone into details here, but some of these include:

  • Hiscox UK: Offers optional cyber business interruption cover, including income loss if systems fail. 
  • Clarke Williams Insurance Brokers: Provides Dependent Business Interruption (DBI) coverage for third‑party service failures, such as internet or cloud outages. 
  • Grove & Dean Insurance Brokers: Offers cyber insurance with explicit business interruption protection, covering lost income due to downtime from cyber incidents. 

This means UK businesses can not only plan for resilience but also insure against the financial impact of outages. Of course, a continuous increase in outages may increase these insurance premiums….

Conclusion

The cloud outages at prove that cloud isn’t “always on”, despite the various up-time (sometimes financially backed) SLAs providers promise. Providers are tightening controls, but businesses must assume failure and design for resilience. Whilst observability tools like Cisco ThousandEyes can’t stop an outage, they do help you see them faster, understand them better, and recover more effectively.

Combined with multi‑cloud strategies, failover planning, chaos testing, and cyber insurance with business interruption cover, organisations can turn inevitable outages into manageable events — protecting both operations and earnings.

Proudly powered by WordPress