01Approach
We are a new firm, and we do not invent case studies. In their place, this page sets out the whole engagement: the five phases, what you receive at each, how we bill, what you own, and the work we decline.
Set out in full, so you can hold us to it.
02The phases
We sit with the people who own the problem and the data, and establish what the system must do, what it must never touch, what a wrong answer would cost, and whether the organisation is ready. If it is not ready yet, the assessment states that and sets out what to fix first.
You receive
The staged plan with its reasons attached: sequence, a build-or-buy decision for each use case, the governance requirements, and the architecture for whatever comes first. Each phase is priced in the written proposal before any work starts.
You receive
One use case built end to end on your data and inside your constraints, then demonstrated running. A pilot that fails at this stage costs little; the same failure at scale does not. If the pilot shows the approach is wrong, we change course and explain why.
You receive
The systems the roadmap committed to, delivered in slices you can run as they arrive, with evaluations and guardrails written alongside each feature. Progress is shown in a weekly demonstration of working software.
You receive
Deployment onto your infrastructure, monitoring in place before go-live, then a walkthrough with your engineers that ends when they can change the system on their own. Ongoing support is an agreed retainer with named hours; where it is not agreed, it is not promised.
You receive
03Engagement models
Assessments, audits, roadmaps, pilots and defined builds. Scope and price are agreed in the proposal before work starts, and any change is re-quoted before it is done.
For transformation at scale: an agreed monthly block of consulting and engineering. You direct it, you can stop at any phase boundary, and you keep everything produced.
Reviews, second opinions and governance assurance, with no build attached. A written opinion you can act on, whether or not we do the work that follows.
04What you own
05Work we decline
AI is the right answer less often than the attention around it suggests. We make these judgements in the first conversation.
If a process always runs the same way, it needs a script or a queue, and we will recommend the script even where an agent would be the larger contract.
AI cannot make up for data an organisation does not hold. That is a data project first, and we will scope it as one.
If every output must be provably correct with no human review, current models cannot give that guarantee. We design the human checkpoint and state where the residual risk sits.
If your timeline needs more people than we can assign to it properly, we will decline rather than staff it thinly.
06Start here
A few paragraphs on where the organisation stands and what it wants to change are enough. The first reply gives an honest view of whether AI belongs in the problem, and where.
Start a conversation