# Deliver better Training

> Canonical: https://deliverbetter.co.uk/what-we-do/training/
> Published: 25 Sep 2026. Last updated: 30 Sep 2026.
> Author: David Bain, Deliver better.

## In short

Deliver better Training is hands-on AI training for teams. We show what AI can and can't do for your work, set up tools like Claude properly, get a first agent running, explain how Claude, ChatGPT and Gemini differ, and agree the rules for using them safely with company information. Vendor-neutral, and pitched at the people in the room.

Deliver better Training 

 
 
111 Get your team working with AI, safely.
 

 
 Most people have tried an AI tool. Far fewer use one well, and fewer still know what they're allowed to put into it. We fix both, hands-on, with the tools your team will actually use.

 

 

 

 
 
 
 

 

 

 

 
 
 
 The problem

 

 
 
222 Everyone's using it. Hardly anyone was shown how.
 

 
 The pattern is familiar. A few people use AI every day and quietly get more done. Most tried it once, got a confident wrong answer and gave up. Someone pasted a customer list into a free chatbot, and nobody’s sure whether that was allowed.

The numbers say this is normal. In a Microsoft survey of 2,003 UK employees, 71% said they had used unapproved consumer AI tools at work, and 51% still do every week. Yet only 11% of UK businesses with 10 or more staff say more than half their workforce has had any AI training.

The tools have moved faster than the training. The AI companies’ own courses are free and often good, but each one teaches its own product. None of them tells you which tool suits which job, or what not to paste into any of them.

 

 

 

 
 
 
 What we do

 

 
 
222 From "we should try AI" to "this is how we work now".
 

 
 
333 Show what's actually possible
Live, on your kind of work, not a slide of logos. We take real tasks from your team, such as a report, a policy or a messy spreadsheet, and show what AI does well, what it does badly, and where a person still has to check.

333 Get you started with Claude
Set up properly from day one: projects that hold your context, skills that teach Claude how your team does a job and how your brand sounds, and the settings that keep your data where it should be.

333 Get your first agent running
An agent is AI that carries out a series of steps rather than just answering a question. We build a small, useful one with you, and agree what it's allowed to touch and when a person signs off.

333 Cut through the names
Claude, ChatGPT, Grok, Gemini, Copilot, OpenClaw. We explain what each is good at, what happens to your data, and which to use for what. We don't sell any of them.

333 Make it safe to use
Short, clear rules people will follow: what can go into which tool, how to check an answer before relying on it, and what to do about agents and add-ons downloaded from the internet.

333 Keep it going
A session is a start. You keep the prompts, skills and rules you built, and we come back to see what's being used and what isn't.

 

 

 

 
 
 
 What you can actually buy

 

 
 
222 Six sessions, each with a clear point.
 

 
 Any of these can be the only one you ever buy. Each is shaped to your team after one conversation.

 

 
 
333 The art of the possible
For leadership teams. What AI can do for your organisation now, what it can't, and the questions to ask before anyone buys anything.

333 Getting started with Claude
For a team that is going to use it. Accounts set up, projects built around your real work, first skills written, and everyone leaving having done a real task with it.

333 Your first agent
For the people who want to build. One small agent on a real task, with sensible permissions and a person signing off. You leave with it working.

333 Which AI, for what?
A plain-English comparison of Claude, ChatGPT, Gemini and Copilot, tried on your tasks and judged against your rules. Useful before you choose a tool for everyone.

333 Safe use for everyone
For all staff. What can and can't go into an AI tool, how to spot a confident wrong answer, and why some add-ons are a security risk. Matched to your policy.

333 Rules people will read
We help you write a short AI-use policy, then match the training to it, so the policy and the sessions say the same thing.

 

 

 

 
 
 
 

 

 

 

 
 
 
 Signs this is for you

 

 
 
222 You might recognise one of these.
 

 
 *01*People are already using AI tools and nobody's sure what's allowed.

*02*You bought licences and most of them sit unused.

*03*A few people are flying with AI and everyone else is watching.

*04*Your team keeps asking whether to use Claude, ChatGPT or Copilot.

*05*Someone wants to try an agent and IT is nervous, with good reason.

*06*The board has asked what your AI training plan is.

 

 

 

 
 
 

Good AI training ends with people *doing* the work, not watching a demo.

 

 

 
 
 
 Safe use

 

 
 
222 The risk isn't the AI. It's what people give it.
 

 
 In KPMG and the University of Melbourne’s study of UK workers, 39% said they had uploaded company information, such as financial, sales or customer data, into a public AI tool. 58% had relied on AI output at work without checking whether it was accurate.

Agents raise the stakes, because they act rather than answer. OpenClaw, the open-source agent that runs on your own computer, is a useful example. CrowdStrike warned that staff were installing it on work machines with wide access to files and the command line, and that it can be steered by instructions hidden in the content it reads. The National Cyber Security Centre says that kind of attack, prompt injection, may never be fully fixed. The practical answer is to limit what an agent can reach.

None of that is a reason to avoid these tools. It’s a reason to set them up properly, and to teach people the handful of rules that matter.

 

 

 

 
 
 
 How it works

 

 
 
222 From first try to everyday use.
 

 
 01

333 Talk
One conversation about your team, your tools and what you want them doing.

02

333 Shape
Sessions built around your real tasks and your rules, not a generic course.

03

333 Do
Hands-on. Everyone does real work with the tools before they leave.

04

333 Keep
The prompts, skills, agents and rules you built stay with you.

05

333 Check back
See what's being used, fix what isn't, and plan what's next.

 

 

 

 
 
 
 
222 Questions people ask
 

 
 Aren't the AI companies' own courses free?Many are, and some are good. Anthropic, OpenAI, Google and Microsoft all publish free training, and we'll point you to the useful bits. Each one teaches its own product. We add what they leave out: which tool suits which job, your data rules, and hands-on work on your own tasks.

Do you only teach Claude?No. Claude is where we'd start many teams, and we'll say why, but we cover ChatGPT, Gemini and Copilot too. We don't resell any of them, so the advice isn't tied to a licence.

Do we need technical people?No. Getting started and safe use are for anyone who writes, analyses or decides. Building a first agent goes better with someone who's comfortable with a spreadsheet formula. They don't need to be a programmer.

What about the EU AI Act?If you operate in or sell into the EU, the AI Act expects organisations that use AI to take steps to build their people's AI literacy. Training matched to the tools you actually use and the rules you actually have is the most direct way to show you've done that. We're not lawyers, so check the detail with yours.

How is this different from Deliver better AI?Deliver better AI decides which uses of AI are worth doing and how to govern them. Training gets people using the tools well and safely, whether or not there's a bigger AI programme behind it. Plenty of organisations need both.

How much does it cost?It depends on the size of the group and what you want them able to do afterwards. David gives a range after one conversation.

 

 

 

 
 
 
 

 

 
 
222 Want your team using AI properly?
 

 
 Tell me who’d be in the room and what you’d like them able to do afterwards. I’ll suggest where to start.

 

 
 
 [
 
 Start a conversation
 
 ](/contact)
 

 
 [
 
 Take the AI readiness check
 
 ](/free-tools/ai-readiness-check)
 

 

 
 Sources: Microsoft UK and Censuswide, October 2025 (2,003 UK employees); ONS, Artificial intelligence in UK businesses, July 2026; KPMG and University of Melbourne, Trust in AI, UK findings, April 2025; CrowdStrike, February 2026; National Cyber Security Centre, December 2025.
