How We Work: Build, Iterate, Grow, Run
Most agencies sell hours and let the scope decide the bill. We do the opposite. Each stage of working with us has a fixed price or a per-outcome price, a clear deliverable, and an exit. You can stop after any stage and keep everything we built.
There are four stages: Build, Iterate, Grow, and Run. Not every client goes through all four, and several should stop after the first. Here is what each stage is, what it costs, and the things we will not do at any stage.
Build: an AI MVP in 15 days
The first stage is a working product in 15 days for a fixed $4,500. Day one is scoping with a clickable prototype, so you react to screens, not a spec. Days two through twelve are the build, with scopes tracked on a hill chart in Basecamp. The last days are polish, launch, and handover.
The scope is aimed at your first paying users, not your fifth funding round. Anything not needed for them goes on a list for later. That cut list is why the price and the date are both fixed. The full details are on the MVP sprint package.
Who should not start here: teams that need regulated data handling, deep enterprise integrations, or many user roles in v1. We will tell you that on the first call.
Iterate: weekly sprints at $1,500
After launch, real users show you what to build next. Iteration happens in one-week sprints at $1,500 each. You pause anytime with no notice period and no penalty. Many clients run three or four sprints, pause to sell, and come back a month later.
The roadmap and milestones live in Basecamp, where you can see them. Each Monday we agree on the sprint's shortlist. Each Friday you get a demo video showing what shipped, what did not, and the decisions we need from you for next week.
A week is short on purpose. It forces small, shippable changes and gives you a natural point every Friday to decide whether the next week is worth $1,500.
Grow: AI agents priced per outcome
Once the product has users, the work that grows with them is usually operations: support tickets, phone calls, invoices. This is where we add AI agents, and we price them by what they finish, not by the hour.
The Support Resolution Desk is $1 per resolved conversation after setup, and escalations to a human are not billed. Finance Doc Autopilot is $1 per processed document. An AI Voice Receptionist starts at $750 per month after a $2,500 setup. Each one has a human checkpoint on anything that emails a customer or moves money. Full numbers are on our pricing page.
Per-outcome pricing keeps us honest. If the agent does not resolve the conversation, we do not get paid for it, so we have every reason to make it work and no reason to inflate usage.
Want this built, not just scoped?
A live AI MVP on your domain in 15 days for $4,500. Then $1,500 a week sprints.
Run: Agent Care
Agents in production need someone watching them. Models change, your policies change, and edge cases appear that nobody tested. Agent Care starts at $1,500 per month and covers watching the agents, re-running evals when something changes, fixing prompts and tools, tracking cost per task, and a monthly report on what the agents did and what we changed.
This stage is optional. If you have an engineer who wants to own the agents, we hand over the code, the evals, and the runbooks, and you run them yourself.
What we say no to
Long meetings. You get recorded updates and one weekly call. If the call has no agenda, we cancel it. Hourly billing. Every stage is fixed price or per outcome, so you know the cost before the work starts, and we carry the risk of our own estimates.
A junior bench. We are a small team. The people you talk to are the people building your product, with agents we built and review. We do not sell senior time and staff juniors behind it. Long contracts. Sprints pause anytime, and Agent Care is month to month after setup.
Surprise AI bills. Model and telephony costs are passed through at cost and reported per task, and we set budgets before an agent goes live. If cost per task drifts, you hear it from us before you see it on an invoice.
How to choose where to start
If you have an idea and no product, start with Build. If you have a product and a list of changes, start with Iterate. If you have a working business and a team drowning in repetitive work, start with Grow, usually after a short AI Opportunity Audit to pick the right workflow.
If you are not sure, book a call and describe what you have. We have built 50+ products over 8+ years, and the honest answer is sometimes "you do not need us yet."
Key takeaways
- Build: a working AI MVP in 15 days for a fixed $4,500, scoped to your first paying users.
- Iterate: one-week sprints at $1,500, a Friday demo video, and pause anytime.
- Grow: AI agents priced per outcome, with a human checkpoint on customer emails and money.
- Run: Agent Care from $1,500 per month, or take the code and runbooks and run it yourself.
- No long meetings, hourly billing, junior bench, long contracts, or surprise AI bills.
Frequently asked questions
No. Each stage stands on its own. Plenty of clients do Build, a few Iterate sprints, and stop.
You tell us before the next Monday and we do not start a new sprint. No fee, no notice period. The project and history stay in Basecamp for when you return.
You do. Code, data, and accounts are yours from the start, at every stage.
We model it with your real volumes before setup. The per-outcome agents have a $500 monthly minimum, and you can see cost per outcome in the monthly report. If a flat model suits you better, we will say so.
The 15-Day AI MVP Playbook
Day-by-day plan, a starter CLAUDE.md, the Lovable-to-production checklist, eval and cost worksheets, and 10 prompts. Pure value, no pitch.
Book 20 minutes, or send a brief
A former CTO replies. You leave with a written plan and a fixed estimate.
Prefer LinkedIn? Connect with Harish.