Agents
FiClaw Desktop presets different roles across research, quant, risk control, analysis and execution, so tasks flow in order along responsibility boundaries.
Roles
Breaks directional questions into research goals, judgment frameworks, review paths and follow-up tasks.
Drives quant research tasks around strategy ideas, factor research, experiment records, backtest validation and review.
Handles rule review, exception checks, risk-metric interpretation and risk-process advice, keeping the team's conventions consistent.
Good for single-stock information triage, fundamental analysis, market-event grouping and first-draft research material.
Assists around pre-execution checks, position and order thinking, slippage analysis and live-trading feedback.
Takes the user's default input, routes tasks to the right role and keeps overall context.
Collaboration
Do not throw every task at the same AI. First judge whether a task is research, risk, execution or general judgment, then pick the matching role.
FiClaw's goal is not multi-round chat, but keeping research, review, execution and capture moving forward.
The rules, methods and review material in each role workspace gradually become the team's own working assets.
If you plan to connect these roles into research, risk control, trade execution or internal data systems, first define the pilot scenario and process boundaries.