Drop a bug recording, a meeting, a demo, a tutorial. Ask anything. VideoLens returns a structured analysis with citations to specific moments in the video — for humans or AI agents.
Free. Bring your own OpenAI API key — we never store it, never see it. You stay in control of cost and data.
Most video tools wrap Whisper and stop. VideoLens combines transcription, frame-level vision, and prompt-directed analysis into one cached pipeline.
Local files, YouTube, Loom, Vimeo, TikTok, Twitter/X, Twitch, Reddit, Google Drive, direct URLs. Roughly 1,500 platforms via yt-dlp.
Three modes built in — General, Bug, Meeting — plus your free-form prompt. Each mode tunes the analyst for what matters: repro steps, decisions, frictions, claims.
Every finding cites a specific timestamp. Click any citation in the UI to jump the player to that moment. Output as PDF, Markdown, or JSON.
Resolve the source. Extract everything. Build a timeline. Synthesize a report. Each step is cached at .videolens/cache/<hash>/ so re-runs are cheap and follow-up questions cost cents.
Detects local file, YouTube, Loom, Vimeo, direct URL, generic page. Reports limitations clearly when a site isn't fully supported.
yt-dlp fetches remote video, ffmpeg samples frames and chunks audio, OpenAI transcribes each chunk, GPT-5.4-mini describes every frame and reads any visible text.
Frame summaries and transcript segments are merged into time-windowed segments — each with visual, OCR, transcript, scene type, and confidence.
One GPT-5.5 call against the timeline, your prompt, and the active mode → structured findings, evidence citations, recommendations, and ticket-ready tasks.
--mode general
Broad review: what's happening, what stands out, what's worth knowing.
Use for: tutorials, demos, content reviews, any video you just want explained.
--mode bug
Bug recordings → reproduction steps, severity hint, ticket-ready summary, possible root-cause areas.
Use for: screen recordings of broken UIs, crashes, session replays of failed flows.
--mode meeting
Decisions, objections, commitments, follow-ups. Uses diarized transcription when available.
Use for: Zoom / Teams / Meet recordings, standups, briefings, sales calls.
Python 3.12+, ffmpeg, and an OpenAI API key. That's the whole list.
# macOS
brew install ffmpeg
git clone https://github.com/shadoprizm/videolens.git
cd videolens
uv sync --extra ui
export OPENAI_API_KEY=sk-... # Web UI uv run videolens ui # Or CLI uv run videolens analyze ./bug.mov \ --mode bug --prompt "What broke?"
The hosted app is up. Drop a video, paste a prompt, get a timestamped report. You bring your own OpenAI API key — we never store it and we never pay for your calls. It's the same code as the open-source repo, just hosted.
Launch app.videolens.ioPaste your OpenAI API key in the sidebar. Calls go straight to OpenAI; you only pay them, not us.
The key lives only in your browser session. Close the tab, it's gone. We don't write it to a database.
Need an OpenAI API key? Get one at platform.openai.com/api-keys · A $5–10 prepaid balance is plenty to test.
VideoLens Pro lives in your browser's side panel. Open any YouTube video or HTML5 player, pick a mode, and get the same timestamped, evidence-grounded report — without leaving the tab. Local files work too.
Heads-up: DRM-protected players (Netflix, Disney+, …) can't be captured. Live streams aren't supported.
No subscription. No usage markup — analysis costs go straight to your own OpenAI account (typically $0.05–$1.50 per video).
Your license key arrives by email right after checkout.
14-day refund policy, no questions asked.
"Analyze once, ask many times" — follow-up questions reuse the cached timeline for cents instead of dollars.
Native Model Context Protocol server so Claude Code, Cursor, and other agents can analyze video as a first-class tool.
Embeddings over processed timelines: "find where they talked about pricing", "find every error message", across your whole library.
PostHog, Clarity, Hotjar, FullStory, LogRocket, OpenReplay — read event exports, not just rendered video.
Everything you need to know about how VideoLens works, what it costs, and how your data is handled.
VideoLens is an open-source video intelligence tool that turns any video into a timestamped, evidence-grounded report. You drop in a video — a local file or a link from roughly 1,500 supported platforms — ask a question, and it returns structured findings that cite specific moments in the video. It's MIT licensed and built for both humans and AI agents.
VideoLens runs a four-stage cached pipeline. It resolves the video source, extracts audio, sampled frames, and on-screen text (OCR), merges them into a time-windowed timeline, then runs one analysis pass against your prompt and chosen mode to produce findings with timestamp citations. Each step is cached, so re-runs are cheap and follow-up questions cost only cents.
Yes. The core tool is free and open source under the MIT license — self-host it or use the hosted app with your own OpenAI API key. You pay only OpenAI's usage, roughly $0.05–$1.50 per video. A one-time $29 VideoLens Pro Chrome extension adds an in-browser side-panel workflow.
Local files plus roughly 1,500 platforms via yt-dlp, including YouTube, Loom, Vimeo, TikTok, Twitch, Twitter/X, Reddit, and Google Drive. DRM-protected players such as Netflix and Disney+, and live streams, are not supported.
No. Videos are processed on your own machine, in your browser (the extension), or in an ephemeral session (the hosted app). Your OpenAI API key stays on your device or in session memory only and is sent only to OpenAI. VideoLens runs no analytics on your video content and never stores your reports.
Analysis costs go straight to your own OpenAI account, typically $0.05–$1.50 per video. A 5-minute clip with 20 frames is about $0.20; a 30-minute meeting is roughly $0.50–$1.50. Cached timelines make follow-up questions on the same video cost only cents.
Yes. VideoLens outputs structured JSON with evidence citations and ships a native Model Context Protocol (MCP) server, so agents such as Claude Code and Cursor can analyze video as a first-class tool.