Build what the audit finds
Builds are sold one at a time, only for what the audit points to, each with its own one-page agreement and a fixed price. We start with whatever the audit ranked first.
Book a 15-minute callWhat we build
Workflow automation
Connecting the systems you already run (project boards, ticketing, CRM, billing, spreadsheets) so data moves between them without anyone retyping it. Includes scheduled summaries, like a weekly digest of what each project actually did.
AI agents and assistants
Assistants that do a defined job well: answering questions from your own documents, scoring and routing new leads, drafting first replies, or reading a week of tickets and commits and telling the owner what changed.
Custom AI software and predictive models
For software companies and teams with their own data: AI features inside your product, internal web tools, and predictive models (which customers are likely to leave, for example) paired with plain-English explanations a manager can act on.
Ongoing support
An optional monthly retainer once something is live: keeping builds running, fixing what breaks when another tool changes, and adjusting things as your team finds out what they actually want from them.
How a build runs
One-page agreement
Scope, delivery date, price, what data the build can read, where it's hosted and who owns what. Signed before we start.
Build
A first build is scoped tightly enough to ship in a week or two. Read-only access to your systems wherever that's possible.
Adjust
You use it with real work, tell us what's off, and we make a round of changes included in the price.
Hand over or keep us on
Take it from there yourself, or add a monthly retainer for support and the next build on the roadmap.
What moves the price
A fixed price per build, quoted after the audit and written into the agreement. Running costs for AI usage and hosting are estimated up front. All of these get talked through before you see a number:
How many systems it touches
Reading from one tool is simpler than keeping three in sync.
Read only, or read and write
A build that only reports on your data is smaller and safer than one that changes it.
Platform or custom code
Some jobs fit an existing automation platform. Others need software written from scratch.
Running costs
AI usage and hosting have monthly costs. We estimate them up front and agree who holds those accounts.
Who uses it
A digest for one owner is different from a tool your whole team or your customers rely on.
How messy the data is
If records are inconsistent, some cleanup has to happen first, and we'll say so in the quote.
What we build with
Platforms when they're enough, code when they aren't. The right choice is whatever you can keep running after we're done.
- Claude API and Agent SDK
- Python
- FastAPI
- pandas and scikit-learn
- Supabase
- Make.com
- Voiceflow
- Relevance AI
- Render
