You follow a dozen YouTube channels and there are never enough hours to watch them. Connect TranscriptHook to Claude or ChatGPT, and your AI reads the new uploads for you. You get a summary per video, with timestamps, and write the issue from there.
Say you write about AI tools, markets or coffee. The best material is often in a video: a founder interview, a product walkthrough, a 90-minute conference talk. Watching all of it is impossible. Skimming means you miss the one good minute. So you write about what you happened to catch.
A transcript is the same video as text. You can search it, skim it, and hand it to an AI to summarize. The hard part is getting the transcripts of the right videos in front of that AI each week. That is the job TranscriptHook does.
If you are searching for a way to go from YouTube to newsletter without the manual copying, this is the short version of how it works.
Set it up once. After that, the only manual part is the writing.
Add the YouTube channels you rely on. Starter takes 15, which is about the size of a typical source list. You don't need a YouTube API key.
TranscriptHook checks your channels about every 10 minutes and keeps the transcripts of new uploads. Your AI can pull them when you ask, or a webhook can send them to a doc.
Ask for a draft with a short summary and timestamps per video. Then do the part only you can do: pick what matters, add your view, link to the source.
Most writers start with the first one.
Use the one-click connect to link TranscriptHook to your AI app. It works with Claude, ChatGPT, Cursor and VS Code. There is nothing to install and no API key to paste. The connection gives your AI 11 tools, including fetching a transcript and listing what your channels posted recently.
Then you just ask. Here is the kind of prompt writers use:
Setup steps are on the MCP connect page.
If you want transcripts to land somewhere on their own, add a webhook. When a new video is picked up, TranscriptHook posts the transcript to your workflow. From there it can go into a Google Doc, a Notion page, a spreadsheet, or an AI step that writes a first summary.
Starter includes 3 webhooks. Guides: Zapier, Make and n8n. For a fuller look at turning videos into posts, see content automation from YouTube.
Most videos aren't relevant to this week's issue. Set a keyword on a channel and TranscriptHook checks the title and the full transcript for it. Matching is by whole word, and it ignores case.
Say you write a markets newsletter. Put "interest rates" on the channels you follow, and you hear about the videos where someone actually says it. You can also ask your AI for the videos that matched your keywords since last Friday.
This is the same idea we describe for monitoring YouTube for news, pointed at your weekly deadline instead of a live feed.
Product demos, founder interviews and release walkthroughs.
Analyst commentary, earnings calls and central bank talks.
Roaster interviews, gear reviews and brewing experiments.
The creators you follow made those videos. A good curated newsletter points readers to them. It doesn't reprint their words.
Use the transcript as research. Write your own summary, keep any direct quotes short, name the creator and link to the video. Readers get a reason to click through, and the creator gets the view. If you plan to reuse more than a line or two, ask permission or check with a lawyer, since the rules vary by country.
Everything above works without code. If you do write code, the pull API returns a week of transcripts in one call, and this script turns them into a digest. The response below is an example, not real output.
# Pull everything from the past 7 days for a tag curl "https://transcripthook.com/api/videos?tag=AI+Research&since=7d" \ -H "Authorization: Bearer tr_live_xxxxxxxxxxxx" # Response structure { "videos": [ { "channel_name": "Example AI Channel", "video_title": "The New Reasoning Model That Changes Everything", "published_at": "2026-04-28T14:00:00Z", "description": "Paper and code links below...", "transcript_text": "Welcome back. Today we look at the new results..." } ], "query": { "tag": "AI Research", "since": "7d", "count": 23 } }
// weekly-digest.js — run every Sunday via cron const TRANSCRIPTHOOK_KEY = process.env.TRANSCRIPTHOOK_API_KEY; const TAGS = ['AI Research', 'Dev Tools', 'Market Analysis']; async function buildWeeklyDigest() { // Pull all tags in parallel const results = await Promise.all( TAGS.map(tag => fetch(`https://transcripthook.com/api/videos?tag=${encodeURIComponent(tag)}&since=7d`, { headers: { Authorization: `Bearer ${TRANSCRIPTHOOK_KEY}` } }).then(r => r.json()) ) ); // Flatten and sort by publish date const videos = results .flatMap(r => r.videos) .sort((a, b) => new Date(b.published_at) - new Date(a.published_at)); // Build context for your LLM from the transcripts const context = videos.map(v => `**${v.video_title}** (${v.channel_name})\n${v.transcript_text.slice(0, 800)}` ).join('\n\n---\n\n'); // Generate digest const digest = await openai.chat.completions.create({ model: 'gpt-4o', messages: [{ role: 'user', content: `You are a research assistant for a tech newsletter. Below are summaries of YouTube videos published this week. Identify the 5 most interesting ideas worth featuring. For each, write a 2-3 sentence explanation of why it matters to a technical audience.\n\n${context}` }] }); // Send to Notion writing doc await pushToNotionDoc(digest.choices[0].message.content, videos.length); } buildWeeklyDigest();
Start on the free plan with one channel and 10 lookups a month. Try the one-click connect with a channel you already follow.
No credit card · Free plan, no time limit