Month 4 at the AI Lab: New Faces, Mind Maps, and AI in the Trades
Month 4 - new open-discussion format kicked off with AI headlines, a Gemini doc-processing demo, real talk on running local models, and a crowd-favorite conversation about AI showing up in traditional trades.
New Faces around the Lab
Jake joined us as our newest intern, fresh off graduating from Washington University in St. Louis. Originally from NYC, he's spending his time at the lab building with AI. His favorite thing about the AI Lab? Meeting so many cool people... and the open garage door!
---Meetup Notes---
New Meetup Format: AI News Discussion to Kick Things Off
We debuted an open-discussion format that opens with the latest AI headlines, giving everyone a chance to compare notes on what they've been reading. The "scoop" round is a useful warm-up before diving into the deeper demos, since most people are already skimming the same feeds.
Mind Maps Are Having a Moment
Mind maps have become a surprisingly hot topic, likely because LLMs like Claude output structured markdown that maps cleanly onto nested visual hierarchies — making AI-generated outlines trivial to convert into diagrams. Tools like Ayoa are riding this wave by pairing classic mind mapping with AI-generated branches.
Wiring Gemini into a Document Processing App
We walked through a live demo of plugging Gemini into a document ingestion pipeline, from upload through structured field extraction. We also talked about how far OCR has come — modern vision models now handle messy handwritten images nearly flawlessly, something that felt out of reach even a year ago.
Running Local Models: Reality Check
Darin's take on running state-of-the-art open models like Qwen locally: you need way more RAM than most of us can justify. That said, smaller open-weight models are already practical for real sysadmin work — think batch-logging into groups of servers to push updates without hand-holding each box.
AI in the Service Industry (Crowd Favorite)
The most engaged conversation was about AI showing up in traditional trades — fence installation, small contractors, local service businesses — rather than the usual SaaS framing. People are quietly using LLMs for quoting, scheduling, and customer comms in industries nobody would have predicted, and this "less theoretical, more applicable" thread is clearly what the group wants more of.
Links that came up
- Nate Jones — Highly recommended AI commentary
- Mo Bitar — This guy's content is designed to bring you joy in the hellscape that AI often leaves in its wake
- Seedance 2.0 — AI video generation model worth a look
- HeyGen Hyperframes — open-source framework from the HeyGen team
- Google Stitch — AI-assisted UI/design tool
- Mike's Sketchmap project
- Ayoa — AI-enhanced mind mapping
Thanks to everyone who showed up. Bring a topic or a link next time!