AI Agent Use Cases
Views —

AI Agent Use Cases

Added today

10All 231

1–20 of 231

Work & Research

An agent team that reviews pre-trade materials and proposals

A case of dividing roles among 11 agents and using adversarial review to handle trading-preparation materials.

Grok Bot

How it works

  1. The author explains that trading preparation was moved to Grok Bot, with 11 agents running daily.
  2. Each agent was assigned one job, with MAP.md and CONTRACT.md defining its scope.
  3. Proposals must pass review by a Red Team agent before proceeding, and no agent was given permission to place orders.

Results

The author reported that clearly defining roles and the review process helped with operation.

Notes

Investment returns and decision accuracy were not published. This is a case of pre-trade preparation and review, not a system that executes trades.

Source

Reddit / Glad-Ranger1879 Original post ↗

Example prompt · Grok Bot

Please design an 11-agent setup for organizing pre-trade review materials for [research subjects and data scope]. Do not grant permission to place real orders or access accounts; use only public materials or [documents provided by the user]. First divide the necessary work so each agent has one responsibility, record role relationships in MAP.md, and document inputs, outputs, and prohibited actions in CONTRACT.md. Have the Red Team role check the proposal's supporting evidence, contrary evidence, missing information, and uncertainty, and record whether it passes and why. Provide role-specific instructions, the two configuration documents, and a review report template with sources. Use [simulated proposal] to test that a failed review stops the next stage, no numbers absent from the materials are invented, and no path can execute an order.