"How many months does rollout take?" is a question we get often. It does not. A week is enough. Here is the timeline from a real project (a mobile app, three web properties, a backend, an ML server, and an admin tool: seven repositories).
Day 0 - connect the repository
Connect GitHub or GitLab, or bring Platty into your on-premise environment. There is no prep work. No documents to organize, no code to tidy up. Platty starts from code, not documents.
Day 1-2 - the code graph gets drawn
The static analysis engine breaks down every repository. Functions, classes, APIs, screens, tables, and batch jobs become nodes, and calls and references become edges. Seven repositories join into one map. From a button in the app, through the backend API, down to the database table, you can follow a single line.
Day 2-4 - rewriting it in business language
The graph is summarized bottom-up into a business layer. Here is what was generated automatically for this project.
| Deliverables | Count | Content |
|---|---|---|
| Epic (work area) | 26 | Business units such as settlement, orders, campaigns, and points |
| Documents | 630 (18,979 items) | Overview, business rules, data dictionary, design, glossary, actors, and use cases for every epic |
| Glossary | 364 | Business terms ↔ code ↔ synonyms |
| Spec | 1,773 | API, screen, event, and schedule specs |
How many people, for how many months, would it take to write this by hand? And on the day they finished, the code would already have changed.
Day 4-6 - grading the existing documents and data
The specs, wikis, and glossaries lying around the company get checked against the code graph. What is correct goes into the SSOT, what is wrong is recorded as wrong, and what cannot be checked against the code stays unverified. Database schemas are connected too. Now the code vouches for what each table means.
Day 7 - the first question arrives
Connect it to the AI your company already uses through MCP. A product manager asks the first question. "When and on what basis do seller settlements get paid out?" The answer comes back with evidence. From this moment on, communication starts turning into lookup.
And this SSOT is not a one-week snapshot. Every time a commit lands, it re-analyzes the affected parts on its own. Once it is born, it stays alive.