Three ways of working together
Most organisations in rail and public transport are somewhere between AI-naive and AI-curious: the tools are in use for summaries and first drafts, and nobody is quite sure what the next step should be. Which way of working fits depends on where you are on that path. The natural order is an assessment first, then enabling your people, building systems, or both, depending on what you need now and where you want to end up.
- AI-naiveNot in use, or in use without anyone deciding it.
- MOST OF THE INDUSTRY IS HERE
- AI-curiousSummaries and first drafts; no view of the next step.
- AI-enabledUsed well on real work; outputs checked as a matter of routine.
- AI-nativeProcesses designed around the tools and owned in-house.
Assess — the AI Readiness Assessment
Who it is for. Any organisation that wants a clear view before committing: an operator with experiments in several departments and no strategy, a supplier with one process that hurts, a consultancy that wants to know what to offer its clients.
What happens. Over one to two weeks we establish where each department stands on the path from AI-naive to AI-native, look at the processes you nominate and the documents and data behind them, and check your security, privacy, governance and compliance constraints: they set how quickly experiments can run, how representative they are of real use, and what has to be hardened before anything touches a production system. Each process is scored against what works reliably today.
What you get. A map of where you are; the first areas where efficiency gains and lessons can be had with limited effort; a sketch of what the end state could look like for you and what has to change to get closer to it; a recommendation on where to run things — local, hybrid or cloud; and the short list of what has to be true before anything goes live.
Format. Three to five interviews, a document review, a half-day readout with your team.
Book a call →| PROCESS | MATURITY | SCORE | CONSTRAINTS | RECOMM. | DEPLOY |
|---|---|---|---|---|---|
| Depot shift handover notes | curious | 6.8 | DPO review | enable | local |
| Spare-parts requisition | naive | 4.1 | ERP access | leave | — |
| Timetable change notices | curious | 7.2 | IT sign-off | build | hybrid |
| Tender document screening | enabled | 8.4 | none | build | cloud |
| Incident report drafting | naive | 5.6 | works council | enable | local |
Sample readout page. Processes and figures are illustrative.
- 09:00 your tools today
- 09:45 your documents
- 10:30 what the model does with them
- 11:30 checking the answer
- 13:00 your constraints today
- 13:45 sensitive material
- 14:30 your first tool
- 15:45 the next three experiments
Sample agenda for a first day. Timings are indicative.
Enable — moving your people along the path
Who it is for. Operations, maintenance and procurement teams at operators; bid, BD and engineering teams at suppliers; delivery teams at consultancies. Anyone who is using these tools already and would like to use them well.
What happens. We start from where your team actually is. The first session establishes what is being used today and for what, and maps the areas where AI can help with limited effort — usually more than summaries and first drafts. From there: where current models genuinely help with your kind of work and where they do not, how to check an output before acting on it, how to handle sensitive material within your governance and compliance rules, and how to build a first internal tool without a data-science team. As the team gains confidence, the sessions turn to what an AI-enabled, and eventually AI-native, version of their work would look like, and which habits and structures it needs.
What you leave with. A written summary of what worked on your material, the two or three experiments worth running next, a first outline of the end state, and a one-page note on what has to be in place before any of it goes near daily operations.
Format. On site or by video; small groups; from a single day to a series of sessions over several months, depending on scope. Scoped with you after a first call or an assessment.
Book a call →Build — agentic systems around your existing processes
| your process owner | your IT or security contact | our engineer | |
|---|---|---|---|
| Scope | involved | informed | involved |
| Constraints settled | informed | involved | involved |
| Build inside your team | involved | informed | involved |
| Pilot on real material | involved | informed | involved |
| Go-live | sign-off | sign-off | sign-off |
SUPPORT AGREEMENT | |||
| Run and maintain | involved | informed | involved |
The production path, top to bottom; scroll to reveal each stage. Where each of the three roles is involved; constraints settled means verification, access control, logging, data residency, governance approvals and cybersecurity hardening. The support agreement is optional.The production path, top to bottom. Constraints settled: verification, access control, logging, data residency, governance approvals, cybersecurity hardening. The support agreement is optional.
Who it is for. Organisations that have found the process worth automating — a bid pipeline, compliance documentation, an internal assistant over confidential material, a reporting workflow — and need it to run reliably rather than impressively.
What happens. We design and build the system inside your team, on infrastructure that follows your constraints: your own hardware, a hardened European cloud, a European or non-European model provider, or a hybrid of them. Verification, access control, logging and data residency are decided at the start, together with the governance approvals and cybersecurity hardening required before the system touches production, because they are what separates a pilot from something the organisation can rely on. Your people are involved throughout, so that the system fits the working day and someone on your side can own it.
What you get. A system your people use, documented and maintainable, with a named owner on your side. Where wanted, a support agreement afterwards: monitoring, model and prompt maintenance, and further development as the process changes.
Format. Typical engagements run from four weeks to six months, depending on scope, team availability and the constraints of the environment. Scoped after an assessment or a first call; fixed scope wherever the process allows it.
Book a call →Working with consultancies that serve rail
Two forms of cooperation. Directly: we help your own team use these tools in its delivery, so that it advises from experience. Jointly: on client engagements where the advice needs a production system behind it, we work alongside you on the build while you lead the advisory work, under whichever arrangement suits the client.
Talk to us about working together →- we — sessions on your material
- your team — uses the tools in its delivery
- your clients — advised from experience
- you — lead the advisory work
- we — build the system alongside you
- the client — whichever arrangement suits
Not sure which one fits?
That is what the first call is for. Thirty minutes by video, three short questions beforehand.
Book a call