TradePilot

A supervised agentic trading assistant with a dashboard-first cockpit, paper-session controls, strategy compilation, and a tamper-evident audit layer.

Problem

Most AI-trading interfaces jump straight to execution fantasy instead of building the operator controls, safeguards, and review surfaces that serious use actually requires.

Why it matters

TradePilot is a systems project about supervision, auditability, and controlled planning under risk, not just market prompts or autonomous hype.

System description

TradePilot combines FastAPI, DuckDB analytics, schedulers, paper-session controls, browser planner fallback, and deterministic risk checks into a GUI-first supervised trading environment.

Tools / methods

FastAPIDuckDBScheduler healthPaper tradingAudit ledgerHybrid planningOperator dashboard

Constraints

Paper-only scope is deliberate.
Execution paths must remain deterministic and reviewable.
Credential handling and session state have to stay explicit.

Workspace source

AI Projects In Process/tradepilot-prototype/README.md

TradePilot is now a real Python web app scaffold for a supervised agentic trading assistant with a GUI.