Cloud software with AI inside, shaped on the work of the people who use it and shared across organisations with separate data.
Organisations: Several organisations on the same installation. Each one reads and writes only its own rows.
The platform: The work of the people who use it, step by step, with AI inside.
The shared base: Written once, under every platform.
The method: The rules under which code agents work and account for their work.
The layers of a platform, from the organisations using it down to the method.
The definition
What the lab means by a platform
Shaped on the work: Screens follow the steps of the people doing the work, in the order they take them. Requirements are written in the words of whoever asks for them and carry a key, so every change to the code traces back to a request.
AI inside: AI works inside the screens, with a spending ceiling and a reserve model written into the configuration. It can be switched off, and the work carries on without it. The decisions that matter stay with a person.
In the cloud, shared: One installation serves several organisations. The database keeps their data apart row by row, and each one sees only its own. An update reaches all of them at once.
Kept with the method: Requirements, tests, probes and verdicts, like every project of the lab. An incident closes only with a regression test that reproduces it, and that test stays.
How it is made
The shared base
The parts every platform repeats are written once. The lab’s apps already run on them, and their tests hold them in place.
Access: Accounts, roles and permissions for each resource. A person grants them, from the admin panel or with an invitation.
Notifications: An inbox inside the app and push notifications on the phone. Each one opens the page it refers to.
Telemetry: Usage events are collected in-house, with no IP address and no third-party trackers. Calls to the models leave a row with the app, the model and the tokens used.
AI spending ceilings: Before every call the database reserves its maximum cost under a daily ceiling in dollars. When a model’s price is unknown, the estimate stays high.
Row-level security: Access rules live in the database, table by table. A test reads the migrations in order and checks the rules that remain in force.
Release: On every push CI runs the typecheck and the tests; if one is red, the release stops. Migrations and server functions go through dedicated scripts.
A new project starts from a template with these pieces already wired together: sign-in, a table protected row by row, the ceiling test, a probe, CI, and a release that waits for a person’s verdict.
Questions
For questions about the platforms, one line to the lab is enough.