The AI Adoption Framework

Why AI Stalls in Smaller Businesses

The pattern is remarkably consistent, and it has very little to do with the technology.

Licences bought, no plan for what happens next

Someone signs off the spend. A group of people get access. There is no rollout plan, no agreed first use case, and no way of telling six months later whether any of it was worth the money.

No clear place to start

Leadership knows AI matters. Nobody can say which process to tackle first, so the decision keeps getting deferred and nothing actually moves.

Access without skills

Staff are handed a tool and an onboarding email. Without real training they try it twice, get an underwhelming result, and quietly go back to working the way they always did.

Your staff have already started without you

People under pressure to get work done will find their own way, and AI takes about a minute to sign up for. The decision is being made for you, one person at a time.

THE PART NOBODY PLANNED FOR

If you do not decide, your employees will decide for you

The gap left by not having a plan does not stay empty. Staff who are measured on getting work done will use whatever helps them do it, and consumer AI is free, instant and available in any browser. No approval required, and no reason for them to think they are doing anything wrong.

What that means in practice is company information going into services your business has no agreement with, no visibility of and no ability to retrieve. Client names, contract terms, pricing, staff data, board papers. Whatever was in the document someone needed summarising at half past four on a Friday.

Different terms apply

Personal accounts are governed by consumer terms, not the business ones. On several platforms that difference includes whether your content can be used to improve the underlying models.

No record, no recourse

You cannot audit what you cannot see. If a client or a regulator asks what was shared and with whom, there is no answer available to you, and no way to get the information back.

It walks out with them

Everything built in a personal account belongs to that account. When the person leaves, the setups they created and the material they uploaded leave with them.

The uncomfortable part is that the people doing this are usually your most motivated staff, working around a gap you left. Banning it tends to push the same behaviour further out of sight rather than stopping it. What works is giving people a managed alternative that is better than what they found on their own, and being clear about why it matters. That is the first thing a plan does.

Most advice on AI adoption is written for large organisations with dedicated IT and AI teams. Smaller businesses need something else: a plan that fits a business their size, a low commitment way to get started, proper training for their people, and a credible route to the more advanced work once they are ready for it.

The Framework

Three stages, moving from having a plan, to real usage, to AI doing and building work for the organisation.

STAGE 1

Foundation

"Do we have a plan?"

A clear, realistic plan for adopting AI, grounded in how your business actually works rather than generic use cases.

STAGE 2

Adoption

"Is our team actually using it?"

Real, embedded usage across teams, with people building their own agents and internal champions supporting them.

STAGE 3

Automation

"Can AI do the work, and build tools for us?"

AI connected into the systems you already run on, and where it is justified, purpose-built applications of your own.

Foundation and Adoption are close to universal starting points. Almost every business needs a plan, and needs its people using the tools well. Automation follows naturally once adoption has taken hold, and it covers a spectrum: some organisations stop at connecting AI into their existing systems, others go on to commission a fully custom application once they know exactly what they want built.

What Each Stage Involves

The framework is deliberately platform independent. The sequence does not change based on which tool you use.

Foundation: do we have a plan?

Organisations rarely fail at AI because the tools are bad. They fail because they never had a plan. Foundation exists to fix that before anything gets rolled out.

What we do: assess where you actually stand across tools, workflows, data practices and staff sentiment, including an honest look at what people are already using on their own accounts. Identify two to four high value, low risk use cases specific to your business. Review the risk and governance position, covering data privacy, acceptable use and any compliance obligations. Then set out a 90 day rollout plan with a defined first win.

What you end up with: a documented plan, leadership behind it, a clear first step, and a sanctioned route for staff who were otherwise going to carry on using their own accounts.

Delivered as: readiness workshop and governance training, plus a readiness assessment and rollout plan

Adoption: is our team actually using it?

This is where most of the value is won or lost. Buying a tool does not change how work gets done. People do.

What we do: role based training, because a finance team and a sales team need different sessions rather than the same generic one. Advanced sessions that take people past prompting into building their own agents and setting simple automations to run on a schedule, none of which needs code. A champions model, training a small group of internal power users who then support their own teams. And measurement, using real usage data and staff confidence to see what is working and adjust.

What you end up with: sustained usage you can actually measure, people building their own assistants for their own repeat work, and internal champions who reduce how much you need us over time.

Delivered as: fundamentals, advanced and champions courses, plus adoption reviews

Automation: can AI do the work, and build tools for us?

Once staff are comfortable using AI as an assistant, the next step is applying it to real business processes and systems. This moves you from individual productivity gains to process level efficiency, and from off the shelf tools to purpose-built ones.

What we do: map processes to find genuine automation candidates such as approvals, reporting, intake and triage. Connect AI into the systems you already run on, whether that is a CRM, finance software, an internal database or a line of business application. Where existing tools and simple integrations are not enough, design and build something bespoke. Then test it, document it and hand it over.

What you end up with: one or more processes meaningfully automated or connected, and where the business case justifies it, a custom application embedded in how you operate. This stage is iterative rather than a one off. Each piece of work tends to surface the next opportunity.

Delivered as: integration and application building courses, plus opportunity mapping, integration builds, custom application development and ongoing support

The Same Framework, Across Three Platforms

The stages and the course sequence stay the same whichever platform you use. Only the delivery detail changes.

Microsoft Copilot

The practical default if you already run Microsoft 365, because it works inside Word, Excel, Outlook and Teams where your people already spend their day. Integration work runs through Copilot Studio and the Power Platform.

See the Copilot journey

Claude

Often preferred for writing, document analysis and technical work. Shared Projects give teams a common setup, and integration runs through the Model Context Protocol, an open standard for connecting AI to business systems.

See the Claude journey

ChatGPT

The widest name recognition, which matters when you are getting a whole team on board. Custom GPTs and Workspace Agents give you configurable assistants for repeat tasks without writing code.

See the ChatGPT journey

In practice most businesses already have a preference or existing licences, and that usually settles it. If you do not, we will help you work out which fits the way your organisation works before you commit to anything.

Two Ways We Deliver It

Every stage is available as training, as consulting, or as a combination of both.

Training courses

Standardised and repeatable. Each course covers the same ground for any organisation, has a fixed price and a defined format, and builds capability across a group of people at once. The sequence runs from Fundamentals through Advanced and Champions, then into integration and application building for the more technical members of your team.

Browse the course catalogue

Consulting

Scoped to your organisation. Readiness assessments, rollout planning, integration opportunity mapping, integration builds and custom application development. This is the work that depends entirely on your systems and processes, and could not sensibly be delivered as a standard course.

Discuss a consulting engagement

Why Work With Us

We have spent 25 years putting technology into businesses and getting people to use it. That second part is the hard bit.

25+ Years

Document management, process automation and Microsoft platforms delivered across more than 1,250 projects.

Adoption First

We have watched enough software go unused to know that training and change management decide the outcome.

Platform Neutral

We work across Copilot, Claude and ChatGPT. The right answer depends on your business, not on what we prefer to sell.

Build Capability

Champions training and documented handover, so you depend on us less over time rather than more.

Frequently asked questions

Do we have to start at Foundation?

Not always. If you already have a rollout plan, agreed use cases and governance in place, we can start at Adoption. What we will not do is skip straight to Automation, because automating a process your people do not yet understand well enough to describe tends to produce something nobody uses. Most organisations that come to us believing they are ready for Stage 3 turn out to have gaps in Stage 1 that would undermine the work.

Our staff are already using AI on their own accounts. Where does that leave us?

It is the most common thing we find, and it is not a disciplinary matter. People with deadlines used a free tool that helped them, and most had no idea it was a problem. What it leaves you with is company information sitting in services you have no agreement with, no visibility of and no way to retrieve, governed by consumer terms rather than business ones. There is also no audit trail, so if a client or regulator asks what was shared you have no answer. The fix is not a ban, which usually just moves the behaviour somewhere you cannot see it. It is to give people a managed alternative that is genuinely better than what they found on their own, set out plainly what can and cannot go into it, and make switching easy. We build that into the Foundation stage, and it is often the single most valuable thing a business gets out of the first month.

Which platform should we choose?

Usually your existing setup decides it. If you already run Microsoft 365, Copilot is the practical default because it works inside the applications your people already have open. Where there is no strong existing preference, the choice tends to come down to the work: Claude is often preferred for writing, document analysis and technical work, and ChatGPT has the widest name recognition. The framework and the course sequence are the same whichever you pick, and there is nothing wrong with using more than one.

How long does each stage take?

Foundation is typically a few weeks and ends with a 90 day rollout plan. Adoption runs across that 90 day period and beyond, because usage takes time to embed. Automation is not a fixed engagement at all; it is iterative, and each piece of work tends to reveal the next opportunity. Most organisations run Foundation once, treat Adoption as an ongoing programme, and return to Automation repeatedly.

Is this only for small businesses?

The framework was built for organisations without a dedicated AI team, which is most businesses below a few hundred staff. Larger organisations use the same three stages, but the work at each one tends to be heavier: more stakeholders in Foundation, more complex governance, and integration work that touches more systems. We deliver both. The sequence does not change with headcount.

What is the difference between the training courses and the consulting?

Training is standardised and repeatable. A course covers the same ground for any organisation, has a fixed price and a defined format, and is the efficient way to build capability across a group of people. Consulting is scoped to you: assessments, integration builds, and application development where the work depends entirely on your systems and processes. Most engagements use both, with training carrying the volume and consulting handling what will not fit a standard course.

Start With a Conversation

Book a free 30 minute discovery call. No obligation, and no pitch. Just an honest view of where you are and what a sensible first step looks like.

Book a Discovery Call General Enquiry

Or call us directly: 0207 947 4041

A Staged Approach to Making AI Work in Your Business