Shipping agents in 6 weeks the boutique cadence and where the time actually goes.

Why six weeks?

Six weeks is not a comfortable number. It is a forcing function.

When I started structuring mid-market agentic AI engagements, I tried longer timelines — eight weeks, twelve, six months. The longer the timeline, the more the work expanded to fill it. A twelve-week engagement became a twelve-week discovery phase with a two-week build at the end, shipped under pressure and missing most of the original scope.

Six weeks is short enough that discovery cannot replace the build. It is long enough to ship something real. That combination is harder than it sounds.

The six-week constraint is what stops a team from spending three months mapping processes it already understands.

The four phases

Week 1: Scope lock

One workflow. One owner. One system. That is the entire output of week one. We write it on a single page — the workflow, the input data, the output action, the human-in-the-loop fallback, the production owner’s name and number.

If it does not fit on one page, the scope is not narrow enough yet.

Weeks 2–3: Build the skeleton

We build the agent end-to-end in the narrowest possible form. It runs on real data, in a staging environment that mirrors production as closely as possible. It does not have to be good. It has to be real.

This is where most clients get surprised. They expect the model to be the hard part. It never is. The hard part is data access, authentication, format normalization, and the edge cases that do not show up in the first ten test runs.

Weeks 4–5: Harden and test

This is where we add what makes the agent trustworthy in production: evaluation loops, monitoring, guardrails, the escalation path when the agent hits an edge case it cannot handle.

We also run the first real users through the system here. Real feedback from real users beats any amount of internal testing.

Week 6: Hand over

We document everything. Runbooks. Alert thresholds. The three edge cases we know about. The one the production owner needs to watch for. We do a live handover session, and we set the retainer terms for ongoing support.

Where the time actually goes

In a typical six-week engagement, the breakdown looks roughly like this:

• Data access and normalization — about 30% of total engineering time
• Evaluation setup and testing — about 25%
• Core agent logic — about 20%
• Monitoring and observability — about 15%
• Documentation and handover — about 10%

The model itself — prompts, context window management, output parsing — accounts for roughly 10–15% of total time. Everything else is infrastructure.

What this means for your first pilot

If you are scoping your first agentic AI pilot, plan for six weeks. Not because it is always enough — sometimes it is not — but because it forces the decisions that make shipping possible. Scope lock. Named owner. Production-first mindset. Real data from day one.

And when you estimate engineering time, triple the time you think you will spend on data access. Then double it again.