PipSync can use large-language-model providers to interpret unstructured signal messages and optional analytics to surface risk context. You remain responsible for trading decisions.
When enabled for a channel or provider, message text may be sent to configured model endpoints (for example via OpenRouter) to extract structured fields such as symbol, direction, entry, stop, and targets. Regex and rule-based parsers may run before or instead of AI.
Only the content required for parsing and audit is processed. Retention of message bodies and parsed outputs follows the privacy policy and product tier limits. Disable AI parsing in channel or AI settings where the product allows it.
Default and selectable models are identified in product settings. Model vendors appear as subprocessors where they process personal data; see the subprocessor register for transfer mechanisms offered by each vendor.
Configurable confidence thresholds can require explicit human approval before trades are executed from parsed signals. Thresholds alone do not remove market or broker risk.
Optional analytics may summarise risk-relevant context (for example volatility or distance to stop). Outputs are informational and not investment advice.