Working together
Three ways to engage.
Most agencies push whichever model bills most. These are described plainly so you can pick the one that actually fits, including the cases where the answer is none of them.
Fixed-scope project
A defined outcome, a fixed price, and explicit exit criteria. Best when you know what you need built.
We scope in phases with a decision point between each. You can stop after any phase and keep everything produced so far — code, documentation, evaluation harness, infrastructure definitions. This is the right model for a first engagement because it caps your risk while we prove we are worth continuing with.
Embedded pod
A small senior team working inside your process, on your board, for a defined period. Best when requirements are still moving.
Typically two to four engineers with a technical lead, working in your sprints and reporting to your people. Suited to AI product work where the specification genuinely cannot be pinned down in advance because you are still learning what the model can do. Priced monthly, minimum three months, because anything shorter is mostly onboarding.
Ongoing retainer
Continuous work on systems already in production: evaluation, monitoring, cost tuning and incremental features.
AI systems drift. Models get deprecated, indexes go stale, providers change pricing, and the accuracy you measured in March is not the accuracy you have in September. A retainer covers that maintenance plus a defined amount of new work. We will not sell you one for a system that does not need it.
When none of these fit
If your problem is solved by an off-the-shelf tool, we will tell you which one and take no fee.
This happens more often than the industry admits. It costs us a project and earns us a referral, and over a decade that trade has been strongly positive.
What the phases add up to
The model decides how work is structured; what it costs is decided by integration surface, data quality and the accuracy bar you have to clear.
None of these models has a price on this page, because the same phase is a different amount of work depending on how many systems have to be read from and written to, what state the data is in when we get there, and whether a regulator sets the threshold for being wrong. What AI development costs in India goes through those variables one at a time, including the ones that make an hourly rate the least useful number in the conversation. The FAQ covers ownership, timelines and what happens if you stop after a phase.
Work out which model fits
Tell us what you are trying to build. We will tell you honestly whether we are the right team for it, and what it would realistically take.