Barlow

Barlow puts an AI panel beside any YouTube video or arXiv paper and lets you ask about what you are watching or reading, using your own API keys. WHAT IT DOES ON YOUTUBE - Opens a panel next to the video. Ask anything about it in plain language. - Answers are grounded in the video's transcript and cite timestamps. Click a timestamp to seek the video to that moment. - Capture the current frame with one click, or paste and drag in screenshots, and ask about what is on screen. - Open the transcript, select a line or a range, and ask about exactly that passage. - Insert the current playback time into your question with one click. - Chats are saved per video, so you can come back and pick up where you left off. WHAT IT DOES ON ARXIV - Opens a side panel for the paper you have open and sends the PDF to the model with your question. MODELS - Choose Claude, GPT, or Gemini per question. Gemini can watch the video itself; Claude and GPT read the transcript and your captured frames. - Barlow is free. Requests run on your own Google, OpenAI, or Anthropic key (or a Claude Pro/Max subscription), which you add once at barlow.ai. Keys are encrypted on the server and never reach the extension. PRIVACY - The extension runs only on youtube.com, arxiv.org, and barlow.ai. It does not read other sites or your browsing history, and has no analytics of its own. - Full policy: https://www.barlow.ai/privacy