Indexing a year of video locally on a 2021 MacBook with Gemma4-31B (50GB swap)

The author, a photographer/videographer shuttling between the Maasai Mara and Silicon Valley, built a local AI indexing system to solve the problem of managing years of unlabeled video footage. Rather than paying $140/month for AI video editing SaaS tools that assume pre-labeled content, they used Claude Code with DaVinci Resolve's open-source MCP and Gemma4-31B running locally on a 2021 MacBook to create a searchable semantic index of their archive. The key insight was that AI video editors solve the wrong problem—the real bottleneck isn't editing but indexing unlabeled footage, making the archive queryable in plain English.