Skip to page content
Client Hub Tel 01892 830111

Blog

Byte-sized AI: Watching Paint Dry

By Nick Ellis, AI and Innovation Lead | Published 18 Mar 2026

I’m back for another Byte-sized AI issue, and as promised here are my three things: 😁

  • Something that you can do to improve how you work
  • Something techy from the news
  • Something that made me laugh 

Byte-sized Ai: IT consultancy in Kent at its best

Every time I think I’ve got a handle on what AI can do for SMEs, the rug gets pulled from under me. And honestly? I love it.

The latest thing to shift my thinking is the rise of Small Language Models – SLMs, since we apparently need another acronym. For the past couple of years, the assumption has been that bigger is better: the largest, most powerful models (GPT, Claude, Gemini) are the ones worth paying attention to. That assumption is now being seriously questioned, and the implications for smaller businesses are – I think – genuinely exciting.

Here’s the bit that caught my attention: these smaller, more focused models can be trained on your data and run on your own hardware. No subscription ticking away in the background, no data leaving your building, no dependency on a US tech giant staying friendly with your budget. You own it.

What does that look like in practice? Imagine a manufacturer with a complex product catalogue. Instead of customers waiting for Derek to come back from lunch to answer a technical query, they’ve got a model trained on their own specs, lead times and compatibility data – available instantly, accurate to their actual products, and running entirely on their own kit. Their customers get faster answers, their sales team gets breathing room, and nothing sensitive goes anywhere it shouldn’t.

So, how could you use something like that in your business? 

This week I’m going to break my own rules, because this didn’t make me laugh – but I think you should know about it anyway. AI Confidential with Hannah Fry is a BBC iPlayer series that tackles the stuff that doesn’t get enough airtime in the AI conversation – ethics, accountability, morality, behaviour. Who’s responsible when AI gets it wrong? How do we build systems we can actually trust? It’s thoughtful, accessible, and well worth an hour of your time.

I should declare an interest: I would happily watch eight episodes of Watching Paint Dry with Hannah Fry. So make of my recommendation what you will.

Following neatly on from the AI News section – if the idea of a manufacturer giving customers instant, accurate answers from their own product data sounds appealing, here’s some good news: if you’re on a Microsoft 365 Copilot licence, you’ve already got the tools to build something very close to that.

Microsoft’s Copilot Agent Builder lets you create your own focused AI agent – trained on documents, data and information you choose – without writing a single line of code. You point it at your source material (product catalogues, spec sheets, FAQs, whatever’s relevant), give it some instructions about how you want it to behave, and publish it for your team or your customers to use.

For our fictional manufacturer, that might look something like this:

  • Upload your product catalogue, lead time data and compatibility guides
  • Set the agent up to answer customer queries accurately, in your tone of voice
  • Publish it to your website, your Teams environment, or both

It won’t know anything outside what you’ve given it – which sounds like a limitation but is actually the point. It won’t guess. It won’t hallucinate specs it doesn’t have. It’ll just answer what it knows, accurately.

You’ll need a Microsoft 365 Copilot licence to access Agent Builder. If you’re not sure whether you’re on the right plan – or you’d like a hand setting something like this up – give us a shout.

Byte-sized Ai
Loading

Check out past Byte-sized AI issues:

Nick Ellis

Nick Ellis is the AI and Innovation Lead at Select Technology. He is the driving force behind our Future Workplace service,  which enables SMEs across all industries to use AI and automation to enhance business operations

Throughout his career he has been instrumental in helping organisations across London and the South East, to unlock the power of AI and automation to build smarter workflows that remove manual administration and connect systems. He has helped organisations replace inefficient tools with simplified processes, freeing up teams to focus on what they do best and enabling business growth without increasing costs.

With over 30 years’ experience in business technology, spanning FTSE 500 organisations and SMEs, Nick specialises in Microsoft 365, cloud software, AI solutions, and business intelligence.


Let's get started

Book a call today to discuss
how we can work together.

Book a call Let's talk
Contact Us