Build with Blumify.

1. The open REST API — no key

Send a link; get the transcript, summary, chapters and notes as JSON, streamed as Server-Sent Events. A link someone already transcribed answers at once, free. Without signing in you can also start 3 new transcripts a day from one network; each becomes a public page. For more, and for questions and translations, connect the signed-in MCP server below.

curl -N "https://blumify.io/api/media/v1/transcribe?url=https://youtu.be/dQw4w9WgXcQ&summarize=1"

event: meta
data: {"key": "youtube:dQw4w9WgXcQ", "cached": true, ...}

event: done
data: {"permalink": "...", "summary": "...", "chapters": [...], "segments": [...]}

Every endpoint, field, error code and limit: /AGENTS.md (written for scripts and AI agents). A short summary of the site: /llms.txt.

2. Your workspace from a script — API tokens

Private transcripts, exports, questions, translations, speaker labels, playlists and the creators you follow, from curl, Python or n8n. Work is paid with your AI credits at the normal rates.

1. In Connected apps, create a token: “Read my transcripts”, or “Read + spend credits” to start work.
2. Copy it (it is shown once) and send it as Authorization: Bearer blm_….
3. Start a transcript, wait for it, and export it:

curl -X POST https://blumify.io/api/media/v1/me/transcripts -H "Authorization: Bearer $BLUMIFY_TOKEN" \
  -H "Content-Type: application/json" -d '{"url": "https://youtu.be/dQw4w9WgXcQ"}'
curl "https://blumify.io/api/media/v1/me/jobs/JOB_ID?wait=50" -H "Authorization: Bearer $BLUMIFY_TOKEN"
curl "https://blumify.io/api/media/v1/me/transcripts/KEY/export?format=md" -H "Authorization: Bearer $BLUMIFY_TOKEN"

Every endpoint, limit and error code is in /AGENTS.md. Follow creators with POST /me/follows and read what is new with GET /me/feed.

3. MCP servers for AI apps

In Claude or ChatGPT, add it as a custom connector. In Cursor, VS Code or Claude Desktop, add it to the MCP settings:

{
  "mcpServers": {
    "blumify": { "url": "https://blumify.io/api/media/v1/mcp" }
  }
}

Tools on the signed-in server

transcribe_link

Start a private transcript of a link, paid with credits.

get_job

Follow it until it is done.

get_transcript

The transcript, summary, key points and chapters.

list_transcripts

What is in the workspace, newest first.

ask

Answer a question from one transcript, citing moments.

translate_notes

The title and notes in another language.

add_speaker_labels

Label who speaks when.

quote_playlist

Price a YouTube playlist before paying.

confirm_playlist

Hold the credits and run it.

get_batch

A playlist's quote or progress.

follow_creator

Follow a channel or podcast; whats_new then lists its new transcripts.

whats_new

New transcripts from the creators you follow, since the last call.

get_usage

Credits available and held, and the plan.

Apps you approved, and the address to copy, are in Connected apps. There you can also revoke any of them.

4. n8n workflows

Three nodes: a trigger, an HTTP request to the REST API, and a Code node that returns the title, link, summary, key points, chapters and the transcript text. Swap the trigger for a form, a schedule or a webhook, and send the result anywhere n8n reaches.

1. Download the workflow and import it in n8n (Workflows → Import from file).
2. In the “Blumify: transcribe” node, put your link in the url query parameter.
3. Run it. A link with captions answers in seconds; audio takes about a minute for every 15 to 20 minutes.

Without signing in, the workflow can start 3 new transcripts a day; links that are already transcribed are always free. The API token workflow transcribes privately, waits for the job and exports Markdown: create a Header Auth credential with Authorization = Bearer and your token. An AI Agent in n8n can also use the signed-in MCP server above through its MCP Client Tool.