METHODOLOGY
The Takshak Method: Assess, Prove, Embed, Run.
Four phases, each with defined outputs and a decision gate. You commit to one phase at a time, and every gate is a genuine opportunity to stop.
The framework
-
Phase 01
Assess
Two to three weeks, fixed fee.
Activities
Structured interviews with the people who do the work, not just the people who commission it. Systems and data audit: what exists, where it lives, what has an API. Process mapping for the two or three candidate areas, with time-and-cost baselines measured, not estimated.
Outputs
A written assessment: candidate processes ranked by expected return and delivery risk; a data-readiness verdict; a pilot proposal with scope, success measure and price.
Decision gate
You decide whether to proceed to a pilot, defer, or stop. The assessment is yours either way, and it is written to be usable with any supplier — including not us.
-
Phase 02
Prove
Typically six to ten weeks, fixed scope and fixed outcome.
Activities
Build the pilot on one process, on infrastructure you control. Run it in parallel with the existing manual process. Measure both against the Phase 1 baseline: throughput, error rate, hours returned.
Outputs
A working system in supervised operation; a measured comparison against baseline; documentation and a data protection record for what was built.
Decision gate
The success measure was agreed in Phase 1. If the pilot clears it, we propose what to embed and what to extend. If it does not, we recommend stopping — and put that recommendation in writing.
-
Phase 03
Embed
Varies with scope; planned and priced from Phase 2 evidence.
Activities
Harden the pilot for production: monitoring, alerting, failure handling, access control, secrets management. Train the people who will live with it. Transfer knowledge — runbooks, architecture notes, handover sessions with your team doing the driving. Extend to adjacent processes only where Phase 2 evidence supports it.
Outputs
Production systems with runbooks; a trained internal owner; a governance record covering data flows, sub-processors and review procedures.
Decision gate
Your team demonstrates they can operate and modify the system. Until they can, Embed is not finished.
-
Phase 04
Run
Ongoing, on a monthly retainer — or not at all, if your team takes it fully in-house.
Activities
Monitoring, maintenance, and small continuous improvements. A periodic review of what the systems are saving against what they cost, because that question does not stop mattering after go-live.
Outputs
Systems that stay reliable as your business and your suppliers' APIs change; a standing record of realised value.
Decision gate
Retainers run in short terms with a clean exit. If the numbers stop working, the honest move is to stop, and the contract makes that easy.
Rules we do not bend.
Fixed-scope pilots
Open-ended "transformation programmes" are how budgets disappear. Pilots have a scope, a price and a success measure before they start.
Exit-friendly contracts
Short terms, no lock-in engineering, and deliverables written so another supplier could take over. Clients who can leave easily, and don't, are the only reference worth having.
Knowledge transfer as a deliverable
An engagement is not finished until your team can run what we built. Documentation and handover are line items, not favours.
Security first, not security later
Access control, secrets management and data protection are designed in at the pilot stage. Retrofitting governance costs more than including it.
How engagements are priced.
We use three models, matched to the phase of work:
Discovery sprints
Short, fixed-fee, fixed-duration. You buy an answer: is this worth doing, what would it cost, what would it return. No obligation beyond the sprint.
Fixed-outcome pilots
A defined scope, a defined success measure, a fixed price. The commercial risk of ambiguity sits with us, which is a strong incentive for us to scope honestly.
Managed retainers
Monthly, for running and improving systems in production — whether we built them or not. Short terms, clean exit, and a standing review of value against cost.
A NOTE ON RATE CARDS
We publish no general rate card because honest pricing depends on scope. But every proposal states its price before you commit, and no phase obliges you to the next. The one exception is deliberate: our productised Rapid AI Applications packages carry published fixed prices.
Start at Phase 1. Stop whenever the evidence says to.
A discovery sprint is the smallest honest commitment we can offer. Most clients know within three weeks exactly what is worth doing.