Technology
AI systems that run the internal work of a software startup
Answers from your own docs and policies, your tools kept in step, and the metrics you review each week assembled, built into the stack you already run.
- Sources
- Billing, product data, CRM
- Definitions
- Written and versioned
- Checks
- Totals match across sources
- Review
- Monday, founders
A small team runs on a dozen tools that do not talk to each other. Support asks engineering the same questions because the answer is in a doc nobody can find. Customer records differ between the CRM, the billing system and the product database. The weekly metrics review starts with an hour of copying numbers into a sheet, and the numbers are not always the same as last week’s.
- 01
Answers from your own docs
Engineers, support and sales ask in plain language and get answers from your internal docs, runbooks and policies, with the source cited and permissions respected.
Built on Knowledge lookup - 02
Tools kept in step
Customer, account and usage details passed between your CRM, billing, support and product systems, so a change is made once and the record agrees everywhere.
Built on System connectors - 03
Weekly metrics assembled
The figures you review each week pulled from your systems with the definitions written down, so the number means the same thing every time.
Built on Reporting pipelines
How it fits what you already run
We work with the CRM, billing, support, analytics and documentation tools your team already uses, through their APIs. You own the code, it lives in your repository, and your engineers can read every line.
Works with tools such as
- HubSpot
- Stripe
- Intercom
- Notion
- GitHub
- PostgreSQL
What stays with your people
- Any change to a customer’s plan, bill or access.
- Anything sent to a customer, an investor or a board.
- The definition of every metric, and the decision made on it.
Your obligations, handled in the open
How we work, not legal advice- 01
Customer data
Your customers’ data is governed by your own terms and privacy policy, and by the privacy law where they live. We process it only where you approve, scope access to the minimum, and never use it to train models for anyone else.
- 02
Security commitments
The promises you have made to customers and partners about where data is stored and who can reach it are the constraints we build within. Every automated action is logged, and you get a data map before anything is built.
- 03
Ownership
All code, configuration and documentation are yours from the first commit. This is how we work, not legal advice. Your organisation and your counsel decide how your obligations apply.
What we do not do
- Change a customer’s plan, bill or access without a person’s approval.
- Build your product for you; that stays with your engineers.
- Process customer data anywhere you have not approved.
- Replace your CRM, billing or support tools.
Questions we are asked
We could build this ourselves. Why would we not?
You could. We do it so your engineers stay on the product, and we hand over code your team can read, run and change.
Does it live in our infrastructure?
Where you approve. Private deployment inside your own cloud account is available, and you own the code either way.
What does the hand-off look like?
A repository, a runbook, a data map, and monitoring that reports to you. We keep running it after launch for as long as you want us to.
Can it use the model provider we already pay for?
Usually, yes. We choose the model with you, within the data commitments you have made to your customers.
Can you build the heavier things: a knowledge base over everything we have, or a model of our own?
Yes, as quoted work under Engineering: knowledge systems, custom models, private inference in your own cloud account, evaluation and agent systems. Each starts with a diagnostic and a fixed written quote, and a model trained on your data is yours alone.
Tell us about one workflow that eats your team’s week.
Fifteen minutes on a call. We ask about the work, and we tell you plainly whether a system fits.