Models

What a six-month AI roadmap actually looks like

Daniel Kerambrum Bicourt · Founder, CO-ORDI · 13 October 2026 · 5 min read

Ask most agencies for an AI roadmap and you get a slide titled "Phase 1: Discovery" followed by phases that never carry a date, a cost or a number you could check in six months' time. A roadmap that cannot be checked is not a plan. It is a pitch deck wearing a plan's clothes. Here is what six real months look like when the dates and the numbers are both attached from day one.

Month one: the proof, not the plan

We have written about this month in full elsewhere, because it carries the most weight: by day 30, the first automations are live and the investment either clears a value map we wrote down in week one or it doesn't. That detail matters for everything that follows. A six-month roadmap built on a month one that never produced a working system is six months of hope. Read the 30-day proof for the week-by-week account. Everything below assumes that foundation is already in place.

Months two and three: width, not sprawl

This is where most in-house attempts go wrong. The team that shipped one good automation in month one gets confident and starts three more at once, in three different tools, owned by three different people. Six months later nobody can say which system feeds which, and a change in the CRM quietly breaks a workflow nobody remembers building. The alternative is to add width deliberately: each new system wired to feed the ones already running, so the second automation makes the first one more useful instead of competing with it for attention. We call this coordination, and it is the difference between a stack that compounds and a stack that just grows.

By the end of month three, a well-run roadmap should show three to five systems live, each one documented, each one with an owner inside the business, and a monthly number attached to what each is worth. If that list does not exist in writing, the roadmap is still a plan, not a record.

Month four: the question legal asks, usually too late

Data residency, access controls, what happens to a customer record once an automation touches it: easy questions in month one, expensive ones in month four once a pilot has already handled live customer data without a clear answer. The roadmaps that stall here are the ones that treated governance as a later phase instead of a design constraint from the first build. If this hasn't come up by month four, it is worth raising before someone outside the project raises it first.

Month five: ownership gets tested, not assumed

Momentum is easy to mistake for ownership. By month five, the founder or operations lead who sponsored the roadmap is usually stretched thin elsewhere, and the question of who actually keeps these systems running resurfaces, often because something has quietly broken and nobody felt responsible for noticing. A six-month roadmap that survives this point has already answered who owns AI in the business in practice, not just on a slide from month one.

Month six: the number, not the narrative

A real roadmap ends month six with arithmetic, not a story about momentum. Add up what the systems built since month one are actually worth, in hours returned or pounds saved, measured against what the six months cost. If the number clears the cost with room to spare, the case for month seven writes itself. If it doesn't, that is the point to stop, not the point to add a sixth tool and hope the average improves. Either outcome is a legitimate result of a roadmap done properly. What is not legitimate is reaching month six without a number at all, which is the usual outcome of roadmaps built as decks instead of commitments.

Why this needs one owner across all six months

The roadmaps that stall rarely fail at any single month. They fail because the person steering month one is not the person steering month four, and nothing carries over between them except a folder of half-finished documentation. An embedded AI Architect who stays on the account for the full six months is what keeps the value map from month one, the coordination work from month three and the governance answer from month four pointing at the same business outcome, rather than three separate initiatives that happen to share a budget line. See how CO-ORDI structures that if you are weighing whether to build this in-house or bring someone in to run it.