An AI Roadmap
You Can Build From
All the AI ideas, none of the AI plan
Find repeated work
Compare value and risk
Choose what to test
You get a short, costed list of projects.
All the ideas, none of the plan
Everyone in your organisation wants to be using AI, but the initiatives don’t add up. There’s lots of great ideas from workshops, but no plan to implement any of them. No clarity on cost and return to get finance sign-off. And the working group don’t know what to do next.
We deliver a pragmatic AI roadmap that identifies what to build first, and builds it.
What’s wrong with typical corporate AI programmes?
- Big picture assessments provide maturity scores and aspirational sticky notes from workshops. But they’re too abstract to build anything from them and don’t measure time or £ savings.
- Use case discovery identifies use cases with high-level business value. But they’re not prioritised on return, don’t estimate implementation cost, and don’t specify the impact on roles and responsibilities. So you can’t take them to the steering group and get a ‘yes’.
- Individual projects each build an AI capability for a part of the business. But the projects compete for the same data, budget, and IT resource. So they create friction instead of helping the whole business.
How do we fix it? Five steps to an AI roadmap
1. AI audit of current state. Who has access to what, what they’re using it for, and what data and models they’re using. We’ll see how people actually use AI, sanctioned and unsanctioned.
2. Estimate the value of AI use cases. Estimate the time and £ savings from all your AI ideas.
3. Build the first AI capability that gives return. Prioritised for return, and sign-off from the users who do the day-to-day job.
4. Governance and tech foundations baked-in from day one. Embedding your AI policy, data protection requirements, ICO guidance and the EU AI Act. And tech foundations for maximum benefit from AI.
5. Maximise the return from future AI. Maximise the benefit of future projects by re-using data pipelines, evaluation datasets and deployment playbooks.
What do we deliver?
AI audit. We interview leadership, shadow 2-3 teams in operations, review current tools, and investigate your data. Deliverable: a picture of the status quo: accounts, tools, data, processes, risks, and hidden toil.
AI roadmap. We interview everyone who has AI ideas and estimate time and £ savings. Deliverable: a prioritised list that even non-technical stakeholders can read, specifying who will benefit, £ saved, implementation cost, business owner, and the impact on roles. We kill losers here, not in the steering group.
First use case. We design, build, evaluate, deploy, monitor, and gatekeep your first AI system. Deliverable: working, monitored system with guardrails. We provide a test and validation set drawn from real-world examples before deployment. And handover code and models to your team.
AI implementation plan for the coming year, prioritised for £ saved.
AI policy for your organisation. Includes what staff can and can’t do in ChatGPT.
Want to implement more use cases? We offer an on-going retainer for this.
Questions we get asked
How is this different to an off-the-shelf LLM tool?
Off-the-shelf LLM tools provide immediate benefit from pre-trained capabilities. But their scope is limited to those capabilities. We’ll investigate off-the-shelf solutions where appropriate. But ultimately deliver software to fit the shape of your processes and data.
What if AI models evolve? Will this be a wasted investment?
We can minimise risk in two ways:
- We will call underlying models through an API, so it’s simple to swap the model.
- We’ll prioritise the parts of AI capability that will remain relevant as models evolve: data pipelines, evaluation, deployment, monitoring, gatekeeping, etc.
How much will it cost?
AI audit and roadmap: Fixed price
First use case: Fixed price per phase, specified for your chosen use case before we write a line of code.
What do you need from us to make it happen?
Sponsor: An empowered decision-maker looking for return on AI
Access to business users in operations
Access to data and examples of your processes
Can we book a roadmap call?
Tell us about the return you’re looking for from AI. What your teams are asking for. And any AI you’ve already tried. Then we’ll explain how we help organisations like yours implement AI that delivers ROI.