How We Work

    We deploy into your firm, not into a ticket queue

    Your firm has engineering capacity and no software team. We become that team, and we do it from inside: our engineers work on your projects, with your people, on your real files. Then we prove the hard part on your own material, build to a fixed scope, and hand you something you own.

    What an engagement looks like

    01

    Discovery

    We sit with your team before we write a line of code.

    Our engineers work alongside yours, on your real files and your real projects, and turn what they learn into a confirmed technical and functional plan. Discovery is a paid, scoped phase with defined deliverables. It isn’t a free sales call.

    • Technical development plan
    • Domain knowledge memorandum
    • Workflow charts and functional descriptions
    • Data governance, security and usage plan
    • Integration and dependency checklist
    • Prioritised build backlog and confirmed scope
    • Risk, assumptions and open questions register

    02

    Proof of concept gate

    We prove the hard part works on your material before you fund the build.

    Some engagements hang on AI doing one specific thing: extracting requirements, reading drawings, drafting to your standard. We evaluate that against your real documents and verified human ground truth, measure it, write down where it is weak, and confirm the acceptance criteria still make sense. If it does not clear the bar, you find out here rather than three months in.

    • Evaluation against representative project material
    • Measured accuracy, with limitations documented in writing
    • Acceptance criteria confirmed or revised before build
    • You approve Discovery in writing before development begins

    03

    Build

    Fixed scope, fixed fee, milestone-based.

    We build to the backlog you approved, against defined acceptance tests covering function, workflow, performance and security. You are not buying an open-ended hourly engagement, and the scope does not drift without a written change.

    • Fixed build fee payable across defined milestones
    • Acceptance testing against criteria agreed up front
    • Written change control for anything outside scope

    04

    Pilot and revision

    Real users, real projects, before anyone commits further.

    You run it inside your firm on live work and give consolidated feedback through structured revision periods. It is accepted when the defined tests pass and you approve it in writing, not when we declare it finished.

    • Internal pilot on live project work
    • Structured revision periods with consolidated feedback
    • Acceptance on your written approval

    05

    Handover and ongoing capability

    You own what we built. We stay as the team behind it.

    On acceptance you own the delivered platform and its commercial rights: the application codebase, your workflows, prompts, architecture, configuration and documentation. We keep only the background tooling we brought with us, licensed to you as embedded in what we delivered. Hosting, support and continued development run under a separate agreement, because keeping us should be your choice.

    • You own the delivered codebase, workflows and documentation
    • Ongoing hosting, support and development under a separate agreement
    • No lock-in dressed up as a partnership

    The rule we build to

    AI drafts and accelerates. A licensed professional decides.

    This is written into how we scope work, not added as a disclaimer. Everything we build assumes a qualified engineer reviews and signs what goes out the door. In a profession where somebody stamps the deliverable and carries the liability, a system that cannot be checked is not an asset.

    Never an unchecked decision-maker

    AI drafts and accelerates. It doesn’t make engineering decisions on its own.

    Traceable to source

    Outputs tie back to the field data, prior work or standard they came from, so review is verification rather than rework.

    Auditable by design

    User validation, auditability and transparency around processing and data handling are core design principles in every build.

    What we are, and what we are not

    What we are

    • The AI engineering capability your firm has not hired
    • Civil engineers who also write the software
    • Embedded in your projects rather than briefed from a distance
    • Accountable for a working outcome, not for hours logged
    • Working inside the security boundary your IT team sets

    What we are not

    • A software product you buy seats of
    • A strategy consultancy that delivers a slide deck
    • An AI training or upskilling provider
    • A staffing agency billing bodies by the hour

    Why we work this way

    The second firm pays for less of it than the first

    Working embedded is slower to start and much better at the finish. It is also how we build things we can use again. A code lookup, a document extraction pipeline, a review workflow: each one comes out of a real problem at a real firm, and each one arrives at the next engagement already working. You get the benefit of every project before yours, and you still own everything we build for you.

    We already know the industry

    We’re civil engineers. You aren’t funding six weeks of us working out what a submittal is, or why the stamp matters.

    We bring working parts, not slides

    Where something we have already built fits your problem, it starts the engagement rather than getting quoted into it.

    You own your side of it

    The platform we deliver, and its commercial rights, are yours. Our background tooling stays ours and comes licensed inside what you own.

    Start with the problem, not the solution

    Bring us the workflow that costs your firm the most and we’ll tell you honestly whether it’s worth building software for. Sometimes the answer is no.