Problem-first AI adoption

Know what AI will change before you spend on it.

Keep people accountable for the work. AI reveals your system of work, and we fix what it reveals. It can magnify strengths and weaknesses in how your organisation works: good practice can go further; unclear decisions, unreliable information and delays can grow too. We help your leaders and workforce build the baseline, investigate where AI could help, and engineer systems where the evidence supports them.

Find where it hurts

Aligning leadership first? Start with leadership alignment. Building the workforce baseline? AI Literacy & Foundations. Already have a problem to investigate? Go straight to Discovery.

What follows from that

We treat AI as an engineering and organisational system, not a novelty or a collection of tools.

Twenty-five years of software delivery: DevOps, platform engineering, distributed systems, and engineering leadership, building real products on .NET, Azure, GitHub, and Azure DevOps. Independent of every AI tool vendor, so tools are judged on evidence and outcomes, not allegiance. And we run our own consulting practice on governed AI agents: live client work, not a demo.

Traceable

Trace changes and the evidence behind them so people can inspect what an agent did.

Honest

Check the result against the requirement. An agent's confidence is not proof that its work is correct.

Accountable

Define who can approve, stop and recover the work. Responsibility stays with people.

Where you are

Start where it hurts.

For leaders who already suspect the tool isn't the problem.

You Rolled Out AI Tooling. Delivery Didn't Change.

Copilot adoption is up but delivery is unchanged. Compare coding, waiting, review and rework before deciding what to change.

Find the real constraint

AI work needs too much checking

AI output is creating rework, or you are unsure how people use AI. Establish what is happening, who is responsible and what better work would look like.

Make sense of everyday AI use

Your AI Pilot Stalled

The demo impressed everyone. Production never happened. That is evidence, and here is how to read it.

Get the pilot honest

The Board Is Asking for an AI Strategy

You need an answer that survives a sceptical CFO. Real problems, prioritised on evidence, not a vendor's slide deck.

Build a defensible answer

EU AI Act Article 4: Your AI-Literacy Obligation

Understand the current Article 4 literacy duty and plan practical, role-appropriate learning without treating a certificate as proof of compliance.

Meet it properly

Output Went Up. Stability Went Down.

More code, faster, and more incidents, longer reviews, less trust in what ships. That trade is not inevitable.

Fix the discipline first

None of these? Tell us what's actually happening and we'll work it out together.

Official Kendall Partner

Start with the problem. Build the capability to act on it.

We are an official Kendall Partner. The Kendall Framework brings problems validated before tools, the knowledge your AI needs written down and owned, and named accountability for keeping it true. We add the investigation of what is holding the work back, and the engineering to act on what the evidence supports. The Kendall Framework and its marks are The Kendall Project's.

What our Kendall partnership means for your organisation

Start free

The Engineering AI Adoption Guide

Our guide applies the Kendall Framework to engineering and technical leadership, with a separately labelled NKD Agility operating layer.

A problem worth AI investment is an operating gap you can evidence: a named owner, a reproduced failure, a measurable cost - stated in the language of operations (throughput, cycle time, error rate, cost, customer outcome), not the language of tool capability.

The discipline is not to make language rigid. It is to make the language that matters stable enough to reuse: canonical definitions, certified terminology with named owners and review dates.

Excerpts from the current guide's problem and language principles.

You can ask for help before you have that evidence. Discovery starts with what you can observe now and makes the gaps clear before you decide what to invest in.

Openly licensed. Read it, run it with your own people, never talk to us. If that is all your organisation needs, that is a good outcome and we mean it.

Read the guide

When you want us in it

What changes when we work together.

Come in at the level your situation demands, not the level a salesperson suggests. Published rates, no time tracking, and if an engagement fails to meet the agreed standard we refund the fee.

A shared baseline for your organisation

Leadership training helps people examine AI decisions across the organisation. Workforce training gives everyone a baseline for using AI, checking its output and recognising when human judgement is needed. Fixed price, agreed before work starts.

For your leaders
For your workforce

A defensible next decision

Investigate one important problem, establish the available evidence and compare AI with other changes. You keep the findings and a recommendation for the next test, or a reason to stop. Priced per problem investigated.

How discovery works

Agent systems with explicit controls

Build with your engineers, on your estate: limited permissions, traceable changes, verification and recovery, with responsibility assigned to people. Ongoing support is capacity, not hours.

How we build

The thinking behind this

AI adoption and agentic engineering insights

Artificial Intelligence
Article

Written rules influence an AI agent, but do not enforce compliance. My agent records and build failures show why independent checks matter, and where …

DevOps
Article

AI exposes that coding was never the main bottleneck in software delivery; real constraints are in system flow, team practices, and organisational …

Artificial Intelligence
Engineering note

Explains how generative AI and PowerShell scripts automate and improve blog post tagging and categorisation in Hugo, with human oversight and …

All insights

The next step

Before you buy anything, it's worth a conversation.

Tell us what needs to improve, what you know about AI use today and what remains unclear. You can start with informal use, unreliable work or a problem you cannot yet measure. We can work out the next step, including whether you need help at all.

Start the conversation

Frequently Asked Questions

Questions we hear before every AI conversation.

If your question isn't here, ask us directly.

Adoption alone does not identify the constraint. Coding may be faster while work waits for review or release; rework may have increased, or the tool may not suit the task. Compare coding, waiting, review and correction before deciding what to change.

No. We are vendor-neutral and problem-first. We investigate what you need before recommending an intervention. That may involve AI, or it may mean changing how the work gets done without adding an AI system.

Examine ownership, context, technical feasibility and operating costs against the value of the problem. Restart when there is a credible next test with success and stopping criteria. Stop when the evidence does not justify further investment. A real problem alone does not guarantee success.

Yes. We use AI agents in our consulting practice and can show a sanitised walkthrough. Written instructions set expectations; permissions, checks and human review address particular risks. Instructions alone do not enforce obedience, and an agent’s claim of completion is not verification.

As checked on 14 September 2026, Article 4 requires providers and deployers within scope to support AI literacy development in context. Training can contribute; a certificate alone does not prove compliance. See our Article 4 explanation for the current legal source, and have your legal or compliance lead assess your organisation’s scope.

Training and workshops are fixed-price and agreed before work starts. Discovery is priced per problem investigated, so you control the scope. Ongoing support is a capacity retainer, not billed hours. Our rates are published, expenses are actuals with no markup, and if an engagement fails to meet the agreed standard we refund the fee.

That depends on the work and the team’s existing capability. We examine skills alongside tests, permissions, delivery checks and recovery. The first step may be training, an engineering improvement or a limited experiment. Workforce literacy is a separate baseline for everyone across the organisation.

Not sure what your AI adoption actually needs?

Tell us the problem you want to solve, what you've tried and what happened. We can examine the next step, whether that is training, investigation or technical work. If the guide and your own people are enough, we'll say so. If we work together and the engagement fails to meet the agreed standard, we refund your fee.

Start a conversation