Exploring Moltbot and Beyond: What if we built AI automation from our own tools?

Recently, I came across a fascinating GitHub account with 168 repositories dating back to 2009. The creator built an AI assistant—originally called Clawdbot, now Moltbot—by stitching together their own tools like VibeTunnel, Peekaboo, and AXorcist. These tools automate everything from macOS interfaces to messaging apps through terminal commands. What struck me was not just the scale but the DIY spirit behind assembling small projects into a larger AI system.

For Norwegian SMBs with 10-50 employees, this kind of automation sparks ideas. Managing daily workflows often costs 400-600 NOK per hour if done manually or through standard software. Then, specialized AI or integration work can easily exceed 1000 NOK per hour. Many of our systems—Tripletex accounting, Fiken invoicing, Vipps payments, and government APIs like Altinn or NAV—still require repetitive manual input or fragmented tools that don’t talk to each other.

What if you could harness a modular approach like Moltbot’s creator, but tailored to your specific needs? Imagine having a system that automates your accounting syncs, customer communications, and data reporting—built from components you can control and extend, rather than waiting for a perfect “all-in-one” solution. It feels more accessible and realistic, especially when budgets and time are tight.

This CAN be done with the tools and workflows I use daily. Using n8n, I create API integrations and webhooks that connect Norwegian business systems seamlessly. For example, linking Tripletex entries to Vipps payments or automating reports for Skatteetaten using cron jobs. Telegram bots can handle notifications and data collection, freeing up admin time. On the AI side, I use Azure OpenAI and Claude APIs to build intelligent assistants that understand your business context without heavy custom training. All these pieces come together quickly with modern low-code platforms—no need for deep programming skills.

This approach fits small to medium companies looking to optimize without massive investments in custom software. It’s not for enterprises needing complex, large-scale systems for hundreds of users or companies demanding bespoke ML models from scratch. Instead, it’s for those who want pragmatic, “good enough” solutions that evolve over time and respect GDPR and local regulations.

So, what if you stopped waiting for the perfect software and started building your own intelligent automation from trusted building blocks? How could that change your daily work and costs? I’ve tested similar setups on my own projects, and the results are promising. Now, it’s your turn to imagine the possibilities.

Resources: github.com