Fluxy · Use cases · AI strategies
AI that writes trading strategies — and proves them before you see them
Most "AI trading" products are a chatbot bolted onto a dashboard: plausible code, no proof. Ours is wired into the engine. The AI is grounded in the live dataset catalog and actual sample rows, writes against the real strategy API (autocomplete is introspected from the engine itself), and every generated strategy is validated by an actual backtest before it reaches you. Code that cannot run cannot reach your screen.
How it works here
The machinery under it
Grounded generation
Your data, not its imagination
The model sees your datasets’ real columns, symbols, and sample values — so it writes against the data you have, not the data it assumes.
Backtest-validated output
Proof before presentation
Every generated strategy runs through the same point-in-time, funding-aware engine as hand-written code. You review results, not promises.
You stay the author
Proposals, not takeovers
For bots, the assistant proposes changes and only you apply them; your code stays yours, with versions and diffs.
Said plainly
What we won't pretend
AI does not conjure alpha; it removes the blank-page tax and the syntax tax. The judgment — out-of-sample discipline, sizing, when not to trade — is still a human job, and the platform is built to keep it one.