Four ways we do this
differently than an agency.
The process below is the mechanical part. These are the non-negotiables that shape how it actually feels to work with us.
Five phases,
then ongoing.
Every engagement follows this shape. The content inside each phase changes per project, the shape doesn't.
We figure out if this is even the right thing to build.
The first call is 30 minutes. We're trying to answer three questions: what are you actually trying to change, what would "done" look like, and is the thing you think we should build really the thing we should build. A fair number of discovery calls end with us sending you somewhere else; that's by design.
- A 30-minute discovery call, no deck, no salespeople
- An honest read on whether we're the right team for it
- A written follow-up within two business days
- 30 minutes of your time
- A rough idea of the problem (we refine from there)
- The budget range you can spend
You see what we'd build before you commit to building it.
We come back with a written scope and a clickable prototype, the actual screens and flow of what we'd build, not a deck about it. On the call we walk through it together, you push back, we adjust. By the end you know the problem we're solving, the shape of the solution, the timeline, and the fixed price. No decision on the call; we send everything through for you to sit with.
- A clickable prototype of the proposed solution
- A written scope: problem, solution, timeline, fixed price
- A walk-through of two or three possible shapes before picking one
- A clear go/no-go, we say no to ~25% of projects
- 60 minutes for the review call
- The right people in the room to say yes or no
- Pushback, the prototype exists so you can redirect us before we build it
We put a working version in your hands.
Whatever the shape of the engagement, the first milestone is the same: something you can click on, try with your team, and push back on. For a Custom Build this is typically the core workflow end-to-end with placeholder edges. For an AI Agent it's the agent doing the real job on sample data, inside your real runtime.
- A working pilot deployed to a staging environment you can log into
- Weekly check-ins with demos (we record them if you can't attend live)
- An open Slack or Teams channel with the whole team
- A nominated product owner on your side, one person, not a committee
- Access to the systems we're integrating with
- 30 minutes a week for a demo call
We take the pilot from 'works on my laptop' to 'works for your whole team.'
This is where the 80% that people underestimate happens: edge cases, error states, permissions, activity logging, onboarding flows, training materials, migration from whatever you were using before. We resist the temptation to add features in this phase, we finish the thing.
- Production deployment on your infrastructure with SSO
- Migration plan and rollout schedule
- Training materials, short videos, and written runbooks
- Handover session with your internal team
- Rollout windows agreed with your team
- Final copy/content signed off
- Any data migration decisions made (we advise, you decide)
We don't walk away.
Software that ships but doesn't get cared for slowly rots. Most of our clients fold ongoing care into a working-with-AI retainer, a predictable monthly rhythm for bug fixes, small improvements, and the next round of features. Not required, but usually the right call.
- Monthly retainer (three tiers)
- Uptime monitoring and alerting if we host
- Quarterly roadmap review
- First right of refusal on any future builds
- Decision on retainer tier
- A named contact who owns the relationship
What makes a project
actually move.
Most stalls aren't technical. They're decisions that nobody on your side is empowered to make. We'd rather have this conversation up front than in week seven.