Homeowner builds fully local AI pipeline that turns walkthrough videos into Trello repair lists
gregmushen · x · 2026-10-09
A homeowner automated his property-inspection workflow: 25-30 minutes of walkthrough videos now get processed by a fully local AI pipeline into Trello repair tasks, finishing in about 2.5 hours.
Key stack and process:
- Whisper-Large-v3-turbo for transcription, preceded by Silero VAD v5.1.2 to strip silence and prevent hallucinations
- Gemma 4 8B (Q4KM) running in Ollama 0.20.0 extracts repair screenshots from the video
- The tool itself was built by a multi-agent workflow: Astra High wrote the spec and implementation plan, Opus 5.5 reviewed and drove an end-to-end goal with testing, orchestrating Sol-6.1 subagents
- Mid-run he realized the job was in Cmux without remote mode, so Codex used computer use to enable /remote-control in the terminal
No cloud APIs involved — he can load videos and drive away without network dependency, saving hours of manual video review and note-taking.
More from coding & agent
- Swapping harness lifts GPT-5.6 repo migration from 6.5% to 31%, paper finds — omarsar0 · 2026-10-09
- He replaced a 17-step agent prompt checklist with a 1976 Unix Makefile — 140+ tickets without drift — RandalSchwartz · 2026-10-09
- Developer builds wildfire analysis tool with Opus 5.5: fire progression and hot spot tracking — natesiggard · 2026-10-09
- Subagent-as-a-service: betting on 'expert' agents like Harvey as tool calls for vertical AI — nbaschez · 2026-10-09
- WebMCP Debate: LLM Agents Skip Ads, Putting Ad-Based Business Models at Risk — ConnectRub4818 · 2026-10-09
- At first Grok Bot Meetup, audience-voted idea becomes a 139-marker map app in ~10 minutes — pswider · 2026-10-09