In partnership with
You’re brilliant on calls. Why do your emails sound flat?
I built a system that turns one coaching call into weeks of emails using your actual words. No blank page. No AI slop. Just your real thinking, extracted.
How can AI power your income?
Ready to transform artificial intelligence from a buzzword into your personal revenue generator
HubSpot’s groundbreaking guide "200+ AI-Powered Income Ideas" is your gateway to financial innovation in the digital age.
Inside you'll discover:
A curated collection of 200+ profitable opportunities spanning content creation, e-commerce, gaming, and emerging digital markets—each vetted for real-world potential
Step-by-step implementation guides designed for beginners, making AI accessible regardless of your technical background
Cutting-edge strategies aligned with current market trends, ensuring your ventures stay ahead of the curve
Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.
I spent three days setting up Moltbot (formerly Clawdbot), a self-hosted AI agent that promises persistent memory and proactive check-ins.
I'm not a developer. Just a guy who likes to play with AI.
So I had Perplexity teach me how to install it step by step through terminal on a VPS. Got it working in Telegram, then moved to Slack.
Set up GitHub integration to give it memory from my Capacities notes. Created cron jobs for proactive check-ins.
I set up 8 Slack channels for different projects and then hit the wall.
My mental model was wrong. I thought it would work like Claude projects - different conversations, one brain that remembers everything. Instead, each Slack channel was a silo. Making them talk to each other would burn through tokens fast.
So I simplified down to 2 channels.
Result?
Three-ish days of work. A system that's partially working.
And a lot of learning about what it actually takes to run an AI agent. My verdict is still out on whether it's worth it.
How I Did It
Here's the setup process I went through:
Day 1: Research and VPS Setup
Used Perplexity to research Moltbot and learn if it was worth trying
Had Claude analyze my use case (7 inboxes, 4 Google accounts, 2 newsletters, multiple clients)
Chose VPS over local install (security risk) or Mac Mini ($600 I didn't want to spend)
Decided on Hostinger VPS ($5/month)
Had Perplexity walk me through terminal commands step by step to set it up and using my Claude Pro account.
Got Moltbot working in Telegram first.
Then had to set up MiniMax credits when my Claude Pro credits ran out.
Called it Data (shoutout to my nerdy teenage self who loved Star Trek: TNG so much he had a book on the schematics of the Enterprise)
Day 2: Telegram and Slack Integration
Created a free Slack workspace and moved there.
Set up GitHub repository to sync with Capacities backup (hundreds of notes and newsletters)
Had Moltbot create 8 channels for different projects (daily emails, YouTube, client work, etc.)
Discovered each channel was an independent conversation - no cross-channel memory
Days 3-4: Simplification and Memory Setup
Realized token costs of cross-channel memory would be high
Simplified down to 2 channels (daily workflow and general)
Created cron jobs for:
9:30am: Ask for daily todos
10:30am: Follow-up check-in
5:30pm: End of day review
Midnight: Create daily summary note
Started using it to share my daily to dos to see if if it keeps me accountable.
Tools Used:
Moltbot (open source, free)
Perplexity → Step-by-step setup guidance
Claude → Research and analysis / API credits
Hostinger VPS → $5/month hosting
MiniMax → API credits
Telegram → Initial testing
Slack → Main interface
GitHub → Memory/document storage
Capacities → Personal knowledge base backup
My Take
If you've been following along with these How I AI issues, you know I usually share examples of people doing smart things with AI that work.
This isn't that.
This is me showing you what happens when you try to set up one of those hyped AI agent tools everyone's talking about.
The promise: An AI that remembers everything, works across platforms, and proactively helps you get things done.
The reality: 3+ days of terminal commands, mental model mismatches, and figuring out token economics.
I'm still not sure if this will be worth it.
But here's what I learned that might help you:
Your mental model matters. I assumed Moltbot would work like Claude (different projects, shared memory). It doesn't. Each channel is separate. That assumption cost me a full day of setup I had to redo.
"Open source" and "free" doesn't mean easy (or free). Moltbot is free to use, but you need technical chops to set it up. If you're still figuring out Claude projects or ChatGPT custom instructions, this isn't for you yet.
The learning has value even if the tool doesn't. Even if I abandon Moltbot next week, I learned a ton about VPS setup, terminal commands, GitHub integration, and how AI agents actually work. That knowledge transfers to whatever tool comes next.
If you're at the edge of "I feel like there's got to be a better way to get AI to do stuff for me," maybe this is worth exploring.
But go in knowing it's going to take time, patience, and willingness to troubleshoot.
For everyone else?
Stick with Claude or ChatGPT.
They work. They're simple.
And you can focus on getting max value out of them instead of fighting with setup of a new tool.
Dig Deeper
👉 Watch my full YouTube video: Molt Bot Review: Is This Open Source AI Agent Worth the Setup?
What did you think of today's edition?
Were you forwarded this email? Click here to subscribe.