Product Fundamentals
Why YieldBoost AI
Why the project exists, which problem it tackles, and what trust model it assumes.
Problem
Idle DeFi capital and hallucination-prone AI output
Target Users
Retail users, judges, and contributors
Trust Model
Guardrailed output + proof trail
Problem statement
Many DeFi dashboards stop at recommendation cards. They show a higher APY route, but they do not explain how the route was produced, what part is simulated, where proof is stored, or how a reviewer should validate the claim.
YieldBoost AI exists to reduce that trust gap. The interface is opinionated, the optimization flow is guided, Integrity Auditor checks the recommendation, and the proof surfaces are always close to the main action paths.
Why the product is shaped as a dashboard first
- A user should be able to understand the pitch in under a minute.
- A judge should be able to see action, result, and proof without switching tools repeatedly.
- A developer should be able to inspect the actual routes that produce the optimization snapshot, streamed reasoning, storage write, and history ledger.
Who the product is for
Trust model
YieldBoost AI is not a custody layer. Wallet access stays in the browser wallet or a manually entered watch-only address. The app reads wallet state, prepares optimization output, and stores proof records.
The product is strongest when it is explicit about verification. Optimization output is treated as a proposal, checked by Integrity Auditor, persisted through the proof flow, and only then presented as a reviewable strategy artifact.
- Connected wallets can switch networks and broadcast the selected address into the app state.
- Watch mode allows a valid address to be tracked without an injected wallet session.
- Proof records are stored in KV when configured, or in the local runtime artifact file when running locally.
- The live submission path defaults to 0G Mainnet, while 0G Testnet remains available for comparison and fallback testing.