Discovery, build, operate. Each priced separately, each fixed-fee. You decide whether to keep going at every gate.
We sit with the human currently doing the work. We map inputs, decisions, exceptions, and what done looks like. We deliver a scoping doc: what's automatable, what's not, what it costs, what it takes. Fixed fee. Two weeks.
We ship the system. Python, Playwright, LLM tools, your APIs. We build against an eval set drawn from real cases. Production-grade code, instrumented, alerted, documented. You own the repo.
Monitoring, model upgrades, new capabilities. Monthly retainer, cancel anytime. Or take it in-house and run it yourself. Either is fine.
If we can't write a scoping doc in two weeks, we shouldn't be in the room. No 90-day "AI strategy" engagements.
Real operations need real code. Zapier and Make are fine for trivial flows. Beyond that you need Python.
You own the repo, the prompts, the eval set, the runbook. We don't bury logic in proprietary platforms.
If your workflow doesn't need an LLM, we'll write the script without one. Tools fit problems, not the other way around.
Every prompt, every line of Python, every eval, written in-house in Indianapolis.
Fixed fee per phase. You see the number before we start. No T&E surprise invoices.