Marketing Against The Grain

This AI Second Brain Remembers Everything I Save (Codex)

84 snips
May 12, 2026
Matt Wolfe, founder of Future Tools and AI creator focused on tooling and workflows, shows how to build a personal AI second brain with Obsidian and Codex. He walks through clipping content, using Codex agents to process and interlink notes, automating nightly processing, and getting proactive briefs and grounded answers from your own wiki.
Ask episode
AI Snips
Chapters
Transcript
Episode notes
ADVICE

Automate Ingest With Codex Agents

  • Run an agent in Codex to process files in your Obsidian raw folder using a scripted prompt (agents.md) that ingests, deduplicates, and creates topic/entity pages.
  • Wolfe runs this on a nightly automation so new clips are processed while he sleeps.
INSIGHT

Grounded Answers Are Concise And Citable

  • Grounding LLM answers in your personal wiki yields concise, actionable recommendations that cite exact saved sources.
  • Wolfe asked his wiki how to show his site in ChatGPT and Codex returned a short strategy with source links from his vault.
ADVICE

Use Codex As Your Agent Runner

  • Use Codex (OpenAI's IDE-style app) to run the agents; it consumes your ChatGPT account credits and can be used free or with paid tiers.
  • Wolfe explains Codex is tuned for concise outputs and integrates with local folders and plugins.
Get the Snipd Podcast app to discover more snips from this episode
Get the app