Building Persistent Memory and Content Indexes

Jay explains uploading podcast transcripts and blog JSON so Claudebot can reference his past work and generate ideas.

Play episode from 09:08

chevron_right

Transcript

chevron_right

Transcript

Episode notes

Over the last 48-72 hours, I completely fell down the rabbit hole with a new AI tool called Clawdbot (rebranded TODAY to Moltbot). Instead of my planned episode about what's on my mind in January 2026, I decided to share my raw, unfiltered experience setting up this AI assistant that runs 24/7 and integrates with all my tools. This isn't your typical AI chat interface—it's an always-on assistant I can text through Telegram that proactively handles research, automates workflows, and maintains institutional memory of all my content.

I'll walk you through exactly how I discovered it, my security-first setup approach using a virtual private server, the learning curve (spoiler: it took me until 1:30 AM), and the specific ways I'm using it now. From automated guest research and fitness tracking to content ideation from my 300+ podcast transcripts, this tool is changing how I think about AI assistance. But I'm also wrestling with bigger questions about what this means for content creation, human creativity, and where we draw the line on AI-generated work.

→ Alex Finn video → Learn about Clawdbot → Setup video from Neil Stephenson → Setup video from VelvetShark

Full transcript and show notes⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

***

TIMESTAMPS

(00:00) Discovery Through Alex Finn's Video

(02:50) What Actually Is Clawdbot?

(05:46) Security-First Setup on Virtual Private Server

(09:48) Current Integrations: Notion, Oura, Fathom, and More

(14:15) Guest Research and Automated Workflows

(16:48) Writing Style Analysis Exercise

(19:13) Privacy Controls and Fathom Integration

(22:54) Security Threats and Ongoing Protection