Approach
We work in three steps. Each has a fixed scope and a result you can check before the next one. Small senior teams use AI in their work and stay with you from the business case to production.
- 01 · Assess
- 2–4 weeks
- 02 · Build
- 4–12 weeks per workflow
- 03 · Run
- Ongoing, with a planned exit
- Pricing
- Fixed scope, tied to results

01 · The model
Most clients enter at Assess or with the 30-day sprint. Some arrive with a pilot that needs hardening and start at Run. Each step can stand alone.
Find where AI is worth it
A fixed-fee diagnostic that ends with a decision.
We map where time, cost and risk sit in your workflows and measure baselines. We build business cases with your finance team and write the AI policy and roadmap. In the final week we scope and price the first build.
- Pricing
- Fixed fee
- Duration
- 2–4 weeks
Includes
- AI readiness across data, systems, people and governance
- Three to five business cases with ROI and payback
- AI acceptable-use and governance policy
- Twelve-month roadmap with owners and dependencies
- Board-ready recommendation
Exit criteria
A signed-off roadmap and a first build scoped and priced.
Put systems into production
Systems in your environment, priced on what goes live.
A senior pod with forward-deployed engineers builds the workflow in your environment, on your data. The eval set is written before the first prompt and becomes the go-live gate. People stay in the loop where judgement matters.
- Pricing
- Per milestone or per workflow live
- Duration
- 4–12 weeks per workflow
Includes
- Workflow design and eval set from real cases
- Agents, integrations and review interface
- Security, access control and audit trail
- Monitoring, runbook and rollback plan
- Training for the team that will operate it
Exit criteria
The workflow passes its eval gate and runs in production with your team operating it.
Operate, improve, hand over
We keep it accurate, affordable and auditable, then hand it over.
We monitor quality and cost and handle model upgrades safely. We extend the system to new entities and teams. We hand it over when your team is ready, as planned from day one.
- Pricing
- Monthly retainer
- Duration
- Ongoing, with a planned exit
Includes
- Continuous evaluation and production monitoring
- Model and prompt release management
- Roll-out across subsidiaries and entities
- Monthly quality, cost and adoption review
- Planned knowledge transfer and handover
Exit criteria
Your team runs the system on its own, with documentation and evals in place.
02 · Pricing
We prefer to be paid for what changes in your firm. This is how it works.
Fixed fee for diagnosis
Assessments have a fixed scope and price agreed before we start. You know the cost of a decision before you make it.
Milestones you can verify
Builds are priced per milestone or per workflow live. Each milestone has a test you can run. The eval gate passes, the system is in production, the team is trained.
Outcome share where the baseline is clean
When the baseline is measurable, part of our fee can be tied to the result, such as hours saved, cycle time or error rate. We only propose it when both sides trust the number.
No lock-in, no licence fees
You own the code, prompts, evals and documentation. Model costs are passed through at cost or billed to your own accounts.
We use time and materials only for exploratory work, and only when you ask.
03 · The 30-day sprint
A good way to see what AI is worth in your firm is to put one workflow into production. The sprint combines a short assessment with a first build, at a fixed price.
01Week 1
Scope
We map candidate workflows, measure the baseline for the strongest one and agree the success metric.
Outputs
- Opportunity shortlist
- Baseline measurement
- Success metric and eval plan
02Week 2
Build
A senior pod builds the workflow in your environment, on your data, against an eval set drawn from real cases.
Outputs
- Eval set from real cases
- Working workflow in a pilot environment
- Integration with one or two systems
03Week 3
Harden
Your people test it on live work. We handle edge cases, access control, logging and the review interface.
Outputs
- Pilot with real users
- Access control and audit trail
- Eval pass rate reported
04Week 4
Live
The workflow runs in production. Your team is trained to operate it, and you get a roadmap for what to build next.
Outputs
- Production release
- Runbook and training
- Measured result against baseline
- Roadmap for the next workflows
Our commitment
If the workflow does not pass its agreed eval gate by day 30, we keep working at no additional fee until it does.
04 · Engagement models
30-day sprint
Fixed price, one workflow live, team trained. The most common starting point.
Best for
Firms that want proof in production before a larger commitment.
Assessment
Fixed fee, two to four weeks, business cases, policy and roadmap.
Best for
Leadership teams that need a board-level decision across the firm.
Forward-deployed pod
A senior pod embedded with your teams, priced per milestone or workflow live.
Best for
Programmes with several workflows and a clear roadmap.
Run retainer
Monthly retainer for operation, improvement, roll-out and transfer.
Best for
Systems already in production that need care, scale or a safety net.
05 · Who you work with
Each engagement is led by people who have done this work before. They use AI to do more of it.
Engagement lead
Accountable for the outcome. Owns the business case, the relationship with your leadership and the decisions on scope.
Forward-deployed engineers
Senior engineers in Geneva or on site, embedded with your team, building in your environment.
AI and evaluation specialists
Design agents, retrieval and the eval suites that decide when a system is ready.
Global delivery network
Additional engineering capacity across Asia-Pacific and the Americas, working on what is safe to share.
06 · FAQ
Next step
Tell us which process takes your team the most time. In a thirty-minute call, we can tell you if the sprint is a good start.