Coordinate a swarm for equity research and coding

They used Claude to manage multiple Codex agents for coding and dozens of agents for equity analysis.

Done with ClaudeCodex· Advanced· Nov 2026

What they did
For coding, they set up a communication channel where Claude managed a set of Codex sub-agents. Claude also checked the work as a second verifier, while the main Codex session only passed messages to Claude. For equity analysis, they used ctlflow.com to launch many agents against a large data store.
How it went
They found that one coordinating agent was faster to work with than juggling a whole group directly. The equity-analysis swarm also turned up unexpected information from the data store. They did not report specific outputs or measurements.
Tips
This approach is expensive to copy, requiring 2x ChatGPT Pro 200 and a Claude Max 200 subscription. If you try it, keep the coordinator separate from the workers and make sure it can pause before any file changes or other irreversible steps.

Try it yourself with Claude

I have [files/data] at [location]. Break the work into parallel sub-agents, with one agent per [company/ticker/topic], and have a single coordinator collect their findings, compare disagreements, and ask a second agent to verify anything uncertain. Summarize the results in [output file/format] using [criteria]. Do not change my source files. Before you send, submit, delete, or pay for anything, stop and ask for my approval.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.