[ 01 / 03 ]HOW WE WORK

Three engagement models. One discovery framework.

How we work is the page that closes the deal — or surfaces the misfit before either side commits. The factory setting is delivery: a first production cut within ninety days, not a recommendation deck. We work in agile two-week iterations regardless of engagement model — production-ready code at the end of every cycle, demonstrable to your control function without intermediation. Subsequent cycles widen scope and harden the build. The engagement-model durations below are end-to-end, not first-cut times: the first cut lands inside ninety days regardless of which model you choose. This page exists so the senior buyer can decide whether the engagement we propose is the engagement they actually want.

[ 02 / 03 ]ENGAGEMENT MODELS

We do not impose one model on every client.

[ TURNKEY ]12–20 weeks

Full delivery.

End-to-end accountability.

Nexura takes complete ownership — discovery, build, deploy, and post-go-live support.

The Nexura team scopes the problem, designs the solution, builds the system, and stands the first production cut up. The client side provides domain expertise, stakeholder access, and the authority to make decisions when the design surfaces a tradeoff. Best when you want a turnkey outcome and have limited internal capacity to drive a build of this character. The deliverable is a working system, not a slide deck. We retain our engineering team after handover for an agreed support window — usually four weeks — to absorb the small things that turn up only in production. After that the system is yours, and so is the institutional knowledge.

[ JOINT ]16–24 weeks

Hand-in-hand.

Capability uplift.

Nexura works alongside your existing team. Joint delivery, knowledge transfer woven in.

Your engineers, data scientists, or operations leads pair with ours on the day-to-day work. We bring the discipline, the architecture, and the senior judgment. They bring the institutional context and the credentials with the rest of the firm. Knowledge transfer is woven into the engagement — there is no handover phase because your team owns the system from day one, with our help. Best when you have talent that needs depth, and want capability uplift to remain in-house. The duration is longer than full delivery because the work is also a training engagement. Outcomes are measured by what your team can ship after we leave, not just what shipped during.

[ GOVERNANCE ]6–12 months

Advisory and supervision.

Lightweight engagement.

Nexura provides strategy, architecture, and oversight. Your team executes the build.

Your team executes the build, attends our weekly architecture reviews, and receives written guidance after each. We are accountable for the quality of the architecture and the soundness of the approach; you are accountable for delivery. Best for mature engineering organisations seeking direction, governance, and an outside perspective they can act on. The cadence is a weekly two-hour architecture review, a monthly steering session with sponsors, and on-demand consultation between. Where the work surfaces a problem we cannot solve in advisory mode, we say so — and we offer a transition to one of the other two models if it suits both sides.

[ 03 / 03 ]DISCOVERY FRAMEWORK

How we map the problem before we touch a model.

Before any model is touched, we map the problem. Most AI engagements that fail in high-stakes industries fail in discovery — the team agreed to a scope they did not understand, made assumptions that did not hold, or signed up to a deliverable that compliance was always going to reject. The discovery framework below is what we apply across every engagement, regardless of model. It produces a written engagement memo that both sides sign before the build phase begins.

  1. [ PROBLEM MAPPING ]What the workflow is today
  2. [ ARCHITECTURE ]Technical design, data flow
  3. [ RISK REGISTER ]Validation envelope
  4. [ ENGAGEMENT MEMO ]Both sides sign
  5. [ BUILD ]Begins from a written scope

FIRST 90 DAYS

  1. [ DAYS 1-15 ]
    Problem mapping
    What the workflow actually looks like today, who owns each step, where the compliance constraints bind, and what the success measure is. Output is a written problem statement with three to five named scenarios the build must handle.
  2. [ DAYS 16-30 ]
    Architecture and risk
    The technical design, the data flow, the model approach if any, the validation envelope, the change-control implications, and the risk register. Output is an engagement memo that both sides sign. If the architecture surfaces a problem that changes the scope, we resurface here — not at week ten.
  3. [ DAYS 31-60 ]
    Build, in agile two-week cycles
    Two-week iterations with working software (not status decks) at every cycle boundary. The first cycle delivers a thin slice that exercises the full architecture against real data. Subsequent cycles add scenarios. Each cycle ends with a working build the client can demonstrate to the relevant control function — model risk, compliance, QA — without intermediation. Where production releases are gated by formal change control, internal cycles continue regardless and production releases batch.
  4. [ DAYS 61-90 ]
    First production cut and handover
    A first production cut goes live — the thin slice from the first build cycles, hardened, with the runbook written and the on-call rotation agreed if applicable. The institutional knowledge transfer happens against a checklist the client validates. Subsequent cycles widen scope and add scenarios. The four-week support window begins from the first production cut.
Discuss your first ninety days