Chrome access lets Codex browse sites and apply knowledge stored in the wiki to practical marketing work. The hosts highlight AEO audits, ad testing, website reviews, and product-marketing collateral audits, while acknowledging that browser operation remains slow to watch in real time.

Use browser access to inspect live websites
Apply stored best practices during audits
Run repeated marketing checks and iterations
Treat current computer-use speed as a practical limitation

“You can use Codex for AEO audits. You can use it for ad testing and optimization, you can use it for website audits, for product…”
22:00

“the computer use and the Chrome use, they're um almost agonizingly slow.”
Matt Wolfe · 23:00

#browser agents#marketing audits#codex#computer use

Hot Take25:00

Personal AI Wikis Are Single-Player Today but Could Transform Teams

The current setup is local and centered on one person's files, making it primarily a single-player system. The hosts expect collaborative context repositories to become valuable for small teams once shared AI wikis mature.

Current vault files reside on the user's computer
Today's workflow primarily serves one individual
Shared context could reduce information gaps across teams
Collaborative AI wikis may become a new team operating layer

“AI is still very single player mode, right?”
25:00

“imagine you're on a small team and all of you you have this collaborative wiki of context, is going to be really valuable, and I…”
25:30

#team knowledge#collaboration#future of work#local first

Explainer· 2

Explainer01:00

How an AI Second Brain Turns Saved Content Into Connected Knowledge

Matt Wolfe describes a personal wiki in which an LLM reads saved transcripts, articles, and notes, then identifies relationships among them. Instead of leaving information in isolated folders, the system creates an interconnected knowledge network that can support future questions and work.

Save useful source material as markdown files
Let an LLM identify related concepts and entities
Cross-link sources into a navigable personal knowledge network
Use the network as context for questions and projects

“It looks at all of these and sort of cross-links them.”
Matt Wolfe · 02:30

“It's like your own little personal curated internet.”
03:00

#second brain#knowledge management#llm#personal wiki

Explainer08:30

The AGENTS.md File Tells Codex How to Process Every Source

A local AGENTS.md file supplies persistent instructions for processing raw material. It can direct Codex to read the source, detect duplicate URLs, generate source records, and update topic, entity, synthesis, or comparison pages consistently.

Store persistent processing instructions in AGENTS.md
Validate source URLs before ingestion
Create or update source and topic pages
Use one short command to invoke a detailed workflow

“This literally tells Codex or whatever IDE you're using, it tells it how to handle specific prompts.”
Matt Wolfe · 08:30

“Read the source from the raw before processing, validate that the same source URL hasn't already been processed.”
Matt Wolfe · 09:00

#agents.md#codex#automation#ingestion

Story· 3

Story11:00

A Personal Wiki Produces Grounded AEO Recommendations for Future Tools

Wolfe asks Codex how Future Tools should appear more often in ChatGPT and Claude responses. The answer draws on sources already stored in his wiki, producing concise recommendations about positioning, crawler access, and the distinction between search and training bots.

Ask questions against the accumulated wiki
Ground recommendations in saved sources
Use citations to inspect supporting material
Apply the system to a live marketing problem

“Based on what I've saved into the wiki, tell me the best strategy to get my website future tools to show up within ChatGPT and…”
Matt Wolfe · 11:00

“You can see exactly what it referenced from within my wiki here.”
Matt Wolfe · 11:00

#aeo#future tools#grounded answers#marketing

Story19:00

Use the Wiki as a Memory Aid for Professional Relationships

Wolfe records where he met someone, what they discussed, and other memorable details in a personal networking CRM. Before a later encounter, he can retrieve that history on his phone and quickly restore the context of a conversation from months earlier.

Record a contact immediately after meeting
Save the location and discussion topics
Preserve memorable personal details
Retrieve the context before a future encounter

“I created a sort of personal networking CRM inside of it as well.”
Matt Wolfe · 19:00

“And I can quickly remember what conversation I had with that person six months ago, and I've actively used that and it works really, really…”
Matt Wolfe · 19:30

#personal crm#networking#memory#relationship management

Story23:30

Journal Into Codex and Get Advice Grounded in What You Have Saved

Wolfe journals his accomplishments, conversations, unresolved problems, and thoughts directly into Codex. The system cross-references each entry against his wiki, combining relevant saved material with the model's general knowledge to produce highly personalized feedback.

Brain-dump the day's events and unresolved issues
Cross-reference journal entries with saved knowledge
Surface relevant advice from previously consumed material
Ground general model knowledge in personal context first

“And what it actually does is it cross-references my journal with everything that's in my wiki.”
Matt Wolfe · 24:00

“for me, that's probably been the absolute most useful, helpful thing that I've done with all of this”
Matt Wolfe · 24:00

#journaling#personalized ai#reflection#knowledge base

Q&A· 2

Q&A15:30

How Codex Gets Access to an Obsidian Knowledge Base

The integration does not require a bespoke connector. Because the Obsidian vault is an ordinary folder on the computer, the user can create a Codex project from that existing folder and grant Codex access to its contents.

Install the Obsidian vault in a known local folder
Create a new Codex project
Choose the existing vault folder
Allow Codex to read the files in that folder

“When you select this, use an existing folder, you would just select whatever folder you installed your obsidian vault on, and then it can just…”
Matt Wolfe · 16:00

“That's as simple as it is.”
Matt Wolfe · 16:00

#codex#obsidian#setup#local files

Q&A27:30

A Basic Codex and Obsidian Second Brain Can Take About 15 Minutes

Wolfe estimates that a basic implementation takes roughly 15 minutes because Andrej Karpathy published the method on GitHub. A user can point Codex at those instructions and ask it to create the initial structure, although adding useful source material remains an ongoing process.

Start from the published GitHub instructions
Open Codex against the Obsidian vault
Ask the AI to build the initial structure
Expect knowledge capture to continue indefinitely

“Um, probably 15 minutes or so.”
Matt Wolfe · 27:30

“you can literally open up your codex, open up your obsidian vault, copy and paste his GitHub page in and say, build this for me”
Matt Wolfe · 27:30

#setup#codex#obsidian#andrej karpathy

Tool· 3

Tool05:00

Why Obsidian Works as the Storage Layer for an AI Wiki

Obsidian stores and organizes ordinary markdown files, making the underlying knowledge base portable and readable by AI coding tools. Its pages can link topics, entities, and original sources without requiring a conventional database or custom application.

Obsidian organizes local markdown files
Markdown is text with lightweight formatting
Concept pages can link back to supporting sources
The files remain accessible outside Obsidian

“Now, obsidian is just a markdown file organizer.”
Matt Wolfe · 05:00

“But again, this is just a text file with some extra formatting on it.”
Matt Wolfe · 05:30

#obsidian#markdown#knowledge base#local files

Tool06:30

Capture Articles, Posts, and YouTube Transcripts in One Click

The Obsidian Web Clipper provides a low-friction way to add online material to the wiki. Wolfe demonstrates saving a post and a complete YouTube transcript into a raw inbox that can later be processed by the LLM.

Use the browser extension while consuming content
Capture complete posts and articles
Extract and save YouTube transcripts
Route new material into a raw inbox

“I just click add to obsidian, and it's now inside of my knowledge base.”
Matt Wolfe · 07:00

“I just press this one button and it saves it into my obsidian vault.”
Matt Wolfe · 07:30

#web clipper#obsidian#youtube transcripts#content capture

Tool20:30

Monitor Sitemaps When Important Websites Do Not Offer RSS

Wolfe monitors website sitemaps to detect newly published pages on AI company sites that lack RSS feeds. The automation alerts him when a URL appears, giving him near-immediate notice of new blog posts and occasional administrative updates.

Monitor a site's sitemap for new URLs
Use the method when no RSS feed is available
Track multiple industry sources automatically
Expect occasional alerts for non-content pages

“I actually set up an automation that looks at the website's site map and alerts me whenever there's a new page added to the website's…”
Matt Wolfe · 21:00

“95% of the time it's a new blog post that I'm getting almost like instantly the moment it goes up.”
Matt Wolfe · 21:30

#competitive intelligence#sitemaps#monitoring#ai news

Takeaway· 2

Takeaway09:30

Let Codex Organize Your Saved Material While You Sleep

Codex automations can periodically check the raw directory and process anything waiting there. This separates quick daytime capture from slower classification and cross-linking, allowing the knowledge base to maintain itself in the background.

Capture material throughout the day
Schedule a nightly raw-folder check
Process only unhandled files
Remove the need to trigger every ingestion manually

“If there are any unprocessed files in the raw directory, please process them now.”
Matt Wolfe · 10:00

“And then at night when I'm asleep, it processes everything that was in the raw folder.”
Matt Wolfe · 10:00

#codex#scheduled automation#productivity#inbox processing

Takeaway17:30

Turn Saved Knowledge Into a Proactive Morning Business Brief

Rather than waiting for a user to search the wiki, an automation can proactively recommend actions based on recently saved information. Wolfe sends such a brief to Slack each morning, shifting the system from passive storage toward active decision support.

Schedule recommendations for a fixed time
Focus the brief on recently saved knowledge
Deliver it inside an existing workflow such as Slack
Use proactive prompts to resurface forgotten ideas

“Based on the stuff that you've saved inside of your wiki over the last couple of weeks, here's what I recommend you try next.”
Matt Wolfe · 18:00

“And it just proactively brings me things.”
Matt Wolfe · 18:30

#proactive ai#morning brief#slack#business productivity

Frameworks in this episode

Context-Triangulated Inbox Response Rule

Contextual Relationship Memory

LLM Wiki Ingestion Loop

Proactive Knowledge Resurfacing Loop

Sitemap Change Intelligence Monitor

Wiki-Grounded Reflection Loop

Insights & moments

◆Hot Take· 2

Codex Browser Access Could Speed-Run Repetitive Marketing Audits