A real-world multi-model workflow for faster coding
OldChippy · reddit · 2026-07-20
A C++ developer with 25 years of experience shares a multi-model workflow for working faster on a 170k LOC codebase.
Workflow
- Use one model to research a topic and write a detailed .md spec.
- Feed that spec into another model for validation and possible algorithm/API changes.
- Pass the docs to Codex/ClaudeCode to generate a workplan, track dependencies, and maintain a progress.md file so work survives restarts or hung terminals.
- Take fast snapshot backups before major runs.
- Iterate through implementation gates with human verification.
- Use the other model to review the workmanship and diagnose stuck loops.
Main takeaways
- The author runs multiple Codex and ClaudeCode terminals in parallel on the same codebase.
- They say this setup makes them more than 10x faster than before.
- In their experience, one model tends to produce a safer but less elegant solution, while the other more often suggests better, more creative approaches.
- They plan to keep both model types around and use the cheaper one for most work, with the other as a deadlock breaker.
Related event: Veteran Developer Shares Multi-Model AI Programming Workflow(2 posts)→
More from coding & agent
- Claude Code + Monid MCP builds a $12K/year B2B prospecting engine on live data — aryanXmahajan · 2026-10-26
- Karpathy shares CLAUDE.md rulebook to stop AI coding agents from piling up tech debt — mdancho84 · 2026-10-26
- Rethinking dependencies in the agent era: drop libs that save 20 lines of code — gethackteam · 2026-10-26
- RustGIS hits 2,000 GitHub stars in two weeks with agent-first architecture — giswqs · 2026-10-26
- Vercel Ships 12 New Products at Its 2026 Conference as It Races Rivals for the Agent Entry Point — FinanceYF5 · 2026-10-26
- Build an AI agent from scratch in under 50 lines of pure Python: ReAct loop tutorial — adnan_hashmi · 2026-10-26