Putting AI inside a product that already has users
Adding a model to a live product is mostly not a model problem. It is data you do not have in the right shape, a latency budget you cannot exceed, and a bill that scales with usage. We do that part: search that finds the right document, features that ship behind a flag, and costs you can predict.
What the work actually involves
Search over your own content
Retrieval that cites the source document, so an answer can be checked rather than believed.
Inside the existing product
A feature in the app your users already open, not a separate chatbot nobody visits twice.
Latency and cost budgets
A target response time and a per-user ceiling agreed before the work starts, then held.
Evaluation before rollout
A test set built from your real questions, run on every change, so quality is a number and not an impression.
How we work
We start by reading, not by quoting
Before a number exists we go through the brief, the existing code and whatever is already in production. You get a written plan with dates and the risks named — usually within a week. If the honest answer is that you do not need us, you get that instead.
A small senior team with one owner
Two to five engineers, one lead who is accountable for the outcome and answers your messages. No account manager relaying questions, no rotating bench, no juniors learning on your budget.
Everything is yours from day one
Repositories, cloud accounts, domains and pipelines are in your name from the first commit, not handed over at the end. When the engagement stops, nothing stops working.
Questions clients ask first
Our data is messy. Is that a blocker?+
It is the normal starting point, and a large part of the work. Preparing, chunking and cleaning the content is what determines quality — a better model on bad data loses to a modest model on well-prepared data every time.
How much will it cost to run?+
We model it before building: cost per request, per active user, per month, at your real volume. If that number does not work, we say so while it is still cheap to change the design.
Can it run without sending data to a third party?+
Yes, on open-weight models in your own infrastructure. It is more expensive per request and usually a little less capable — we will show you both numbers rather than push the option that suits us.
Are you the engineer, not the client?
We staff this work partly from independent developers and studios verified on our own platform, and the same platform is how contractors outside the US invoice American clients and get paid.
How that worksOther things we do
Custom software development for companies in the US, Canada and the EU
Websites and SaaS products, built to be measured
Systems architecture and cloud infrastructure
AI agents and automated workflows that run in production
Demand generation for companies that sell something complicated
Tell us what you are building
A short description of the problem is enough to start. We reply with what we would do, what it would take, and what we would not touch.
Discuss a project