Search hours of your video footage by description
Uses Gemini's native video embeddings to search raw footage with plain text queries such as a car cutting you off, with no transcription.
Source
sohamrj on Hacker News · 2026-03-24 · Project link
Written in our own words. Credit to the original author. Link checked 2026-10-14.
What they did
Uses Gemini's native video embeddings to search raw footage with plain text queries such as a car cutting you off, with no transcription.
Result (as claimed)
Open-sourced as SentrySearch with sub-second search.
Tools they used
Cost: Pricing depends on the tools you pick; check each official pricing page.
Set it up
Collect your inputs
List what goes in (raw footage, script, topic list or audio) and the exact output you want, for example 5 vertical clips under 45 seconds. Put inputs in one folder.
Set up the tools
Open Gemini, Qwen. Connect accounts only where needed and note any per-minute or per-credit limits on the free plan before you start.
Run one piece end to end
Make a single output first. Check pacing, captions, voice and framing by watching it on a phone.
Turn it into a repeatable recipe
Save the prompts and settings that worked as a template so the next run takes minutes, then batch.
Review before you post
Watch every output before publishing. Check rights for music, likeness and footage, and label AI content where the platform requires it.
Copy the full setup prompt
Paste into your agent and it will walk you through the setup, asking before anything that sends, posts or spends.
You are helping me set up this AI use case end to end in Claude Code. Goal: Search hours of your video footage by description What the original author did: Uses Gemini's native video embeddings to search raw footage with plain text queries such as a car cutting you off, with no transcription. Reported result (as claimed by the author): Open-sourced as SentrySearch with sub-second search. Source: https://news.ycombinator.com/item?id=47503617 Tools they used: Gemini, Qwen Category: Creators & video Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
You are helping me set up this AI use case end to end in Codex. Goal: Search hours of your video footage by description What the original author did: Uses Gemini's native video embeddings to search raw footage with plain text queries such as a car cutting you off, with no transcription. Reported result (as claimed by the author): Open-sourced as SentrySearch with sub-second search. Source: https://news.ycombinator.com/item?id=47503617 Tools they used: Gemini, Qwen Category: Creators & video Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
You are helping me set up this AI use case end to end in Grok Bot. Goal: Search hours of your video footage by description What the original author did: Uses Gemini's native video embeddings to search raw footage with plain text queries such as a car cutting you off, with no transcription. Reported result (as claimed by the author): Open-sourced as SentrySearch with sub-second search. Source: https://news.ycombinator.com/item?id=47503617 Tools they used: Gemini, Qwen Category: Creators & video Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
Related use cases
Generate AI UGC ad clips with a Claude skill
A custom Claude skill runs the pipeline: Gemini for talking heads, Kling for B-roll, Higgsfield and Nano Banana for images.
Search every photo and video frame on your Mac
Built an AI search over every photo and every frame of video stored on a Mac.
Get an explainer video for any article
Scrimba plugged an LLM into its interactive video format and made HN.watch, which turns every HN post into a video.
Turn your own lyrics into a full music video
A musician writes the lyrics himself, produces the song in Suno and designs the characters in Midjourney.