Go or no-go, on page one, with a name signed under it. The first half costs nothing.
The free assessment takes two days. A Production Audit takes up to another five business days.
AI made software cheaper to build. Building the wrong software is still expensive.
The building got easy and the scoping disappeared. AI will write whatever you ask it to write. What it won’t do is stop and ask whether the thing should exist, or what done means. It says yes to everyone.
Most AI pilots never make it to production, and most of the ones that do never show up in the P&L. Nobody scoped them.
So we scope. Twenty-five years of it, now pointed at AI.
The scoping has a name on it: Phil Cowan. 140+ engagements. ShapeShift and SALT Lending launched out of our own office. Creative direction for Disney, Pixar, and Universal film campaigns. Fund rails for WallStreetBets. The rest is on the record.
The Production Audit is one document, eight parts, every time. Your procurement team will recognize the shape and your engineers will recognize the substance.
On the first page, not a maybe with charts behind it. A person’s name goes on it, because someone has to stand behind the answer, and it isn’t a model.
Everything we found, tested against a written, disclosed rubric: architecture under load, security under attack, the workflow with real users in it, costs at real volume.
Bounded in writing. The exclusions matter more than the inclusions, because that is where the surprises are.
Functional, performance, security, accuracy. Done is defined before the iteration starts.
If the verdict is go, this is the path: sequence, milestones, and the build clock. Whoever builds, ours or yours, this is what they follow.
In numbers, agreed before anything gets built, measured against the non-AI alternative. If it doesn’t pay, the scope says so.
What the work requires from your systems and your people, named up front.
What we’re taking on faith, what could move, what it costs, and how the engagement ends. The exit is written in advance.
The scope runs under NDA, signed before any code moves. Nothing you send trains any model, and your artifacts are deleted on request when the work is done.
No-go is always an acceptable answer.
The verdict is yours to take anywhere, including to someone else’s build team. We’d rather you build with us, obviously. But a scope that only works if you hire us isn’t a scope, it’s a pitch.
If we find where it breaks, good. That’s the point of looking. If it holds, you get your go with the evidence behind it. Either way you stop guessing.
You have the idea, the deck, or the roadmap, and money is about to move. The scope gives you direction first: which door to take, what it will cost, what paying off means in numbers, and a written no if the answer is no. Cheaper to hear it now.
You built it. The demo works, people leaned in, and then it stopped moving: probably around seventy percent done, and the last stretch is the wall. The scope is the diagnosis: what’s actually missing, ranked by what decides, with a priced plan through it.
You didn’t build it. You inherited it: the contractor rolled off, the founder moved on, and now a system nobody can fully explain runs under your name. You need to know what you’re holding, and the scope tells you in writing: what it is, what it needs, what it costs to keep alive, and whether it should be.
An audit attached to a build practice is a conflict of interest, and you’re right to treat it like one. So the controls are structural: the scope is priced to stand alone, a no-go pays us exactly what a go does, and the document is built to hand to someone else’s team without us in the room. If the verdict is go, the scope prices and schedules every door below. You pick who opens them:
Agentic systems built into your operation and wired to your data and tools: research, ops, support, sales. Real work, under human judgment.
The demo works. Production must hold. Your AI-built app hardened, scaled, and shipped for real.
Bring the problem. Get the finished system. Greenfield builds with AI and your team, integrated with what you already run.
AI built it. Now secure it. Reviews and hardening for AI-generated systems, before they meet the real world.
Ten minutes about your build. A straight answer in two days. $0. Not ready? Take the Guide: the six rules with checklists you can run yourself. Free. No email required.