AI-assisted product management
Three engines on one runtime.
Krine decides what to build, Tholo builds it, Basano proves it. They are independent tools that compose. Some tasks use one engine, some use all three.
At a glance
Decide, build, prove.
Each engine is its own tool with its own page. Open one to read how it works on its own.
What AI answer surfaces return changes over time. The SERP event registry records model releases, search feature changes, and confirmed algorithm updates, each row carrying the confidence tier of its date.
The substrate
What the three share.
The engines are separate tools, so the real unity is underneath them, in what they stand on. Three things are shared, and that is where the system is one.
One runtime.
The three engines run on the same Cogent runtime. It is the shared ground they stand on, not three separate stacks that happen to share a name.
One catalog.
Each engine selects from the same open Claude Skills catalog. Krine, Tholo, and Basano pick the few skills a task needs from one library, so the same skill means the same thing across all three.
Consistent artifacts.
The engines emit artifacts with published shapes you can build against: Krine's stop package, Tholo's plan, and Basano's verdict. One engine's output is the next one's input because the shapes are stated.
One idea, through all three and back.
The engines compose; there is no single automated run that does all three. A person moves the work from one engine to the next, carrying each engine's artifact across. This is the shape across a full lifecycle. What each engine does on its own lives on its own page.
From an idea to a live, verified product, the path one founder runs.
The lifecycle
An idea, traced through all three engines and back. Hover, focus, or tap a stage to read what it does.
One idea, traced through all three engines and back. The operator carries the work across each seam.
An idea. A new idea or request enters the lifecycle. It has no order and no plan yet.
Inside a team
Running them against your own work.
A sample page is one thing and a team's repository is another. Two references cover what changes: where the engines sit in the pipeline, and what the people around them run.
- Reference
Deploying the engines
Where the three engines sit inside an existing CI/CD pipeline, and the read-only data access they run under. Written for a team that already has a warehouse and a deploy process.
Read the reference - Procedures
Forward-deployed workflows
Operating procedures for the engineers, marketers, and analysts around the engines. Each one carries a prove gate and an honest status, and the verification ladder renders from the repository manifest.
Read the workflows
The do-it-yourself entry point is the starter kit: paste one instruction and your own Claude builds your workflow library from the same method, no engines required.
Run one today.
Basano is the engine you can watch work right now. See a real verdict on a sample page, holds and fails and all.
A full case study, one real site taken through all three, is on the way.