How it works
Add a source, extract, understand, create, automate
Every GistCite tool follows the same five steps, and every step is one call in the API. This page explains the flow, the engines that do the work and how usage is metered.
Published · Last updated · Reviewed
The five steps
Step 1
Add a source
Paste a YouTube link, a TikTok or Instagram post, a web address, or drop a PDF, Office file, recording or video. Nothing is stored except the text that comes out.
Step 2
Extract
The transcript or text comes back with its structure — timestamps, chapters, page anchors, headings and tables — ready to copy, download as Markdown, Word or PDF, or send to your AI.
Step 3
Understand
Ask for a summary, key points and chapters, or ask a question and get an answer with citations back to the exact page or second.
Step 4
Create
Turn the source into a post, a brief, a script, captions, subtitles or clips, and download the file.
Step 5
Automate
Wire the tools into one run on the Workflow canvas, with a record of every step. Everything in the web app is also one API call with the same account and quota, so a chain that works by hand runs on a schedule.
What goes in and what comes out
| Tool | Takes | Gives back | Cost |
|---|---|---|---|
| YouTube Transcripts | A watch URL, a short link, or a bare video id all work; playlists and batches on Pro. | Click a line to play it, search, copy with a link, download Markdown, Word, PDF, TXT or SRT, or summarize with AI. | 1 unit per transcript; cached repeats also 1 unit; anonymous visitors get a small per-session allowance without an account. |
| Document Extractor | PDF, DOCX, PPTX, XLSX, TXT, MD, EPUB or HTML, from your computer or a URL. | Save in nine formats, or summarize, ask, translate and create from it. | 1 unit per document; OCR 3 units per page; translation 2 units per page; batch items 1 unit each. |
| PDF Editor | Drop it into the editor; password-protected files ask for the password. | Download the flattened PDF. | Browser editing is free; server-side page operations (merge, split, protect, unlock) are 1 unit each. |
| Workflow | A YouTube link, a link to a document, or a source already in your library. | A file, a project entry, an email or a webhook post, plus a record of every step's input, output and cost. | The workflow itself is free; each step costs what its tool costs (a transcript 1 unit, an AI summary 10 or 30). |
| AI Summary, Ask and Create | A transcript, document, web page or recording lands in your history. | Copy, download, or open the cited point in the source. | 10 units at Standard, 30 at Deep, 1 for a cached repeat. Ask and clip ideas run at Standard. |
| Speech to Text | Dictate live, record a browser tab, or upload audio or video. | Fix names, then copy or download TXT or SRT. | 1 unit per started minute, minimum 3; 2 units per minute at high quality; cached repeats 1 unit. |
| TTS Reader | Or open an extraction from your history. | Optional MP3 download. | Playback free; MP3 export 5 units. |
| Markdown Editor | Start blank, or open an extraction. | .md, .docx or .html. | None. |
| Web Extractor | Or a list, or a start page for a crawl. | Markdown, text, HTML, JSON, or a ZIP for batches and crawls. | 1 unit per page extracted, including each batch item and crawled page. |
| Media Tools | A file up to 500 MB, or a public video link. | Download each result; the transcript joins your history. | 1 unit per transform or Audio Kit export; downloads 1 unit per MB; clip ideas 10 units. |
| Sentiment Analysis | One video, or up to ten. | CSV, XLSX, JSON or HTML table. | 1 unit per 100 comments collected; 10 units per 100 comments scored with AI. |
| Open Library | Title, author, subject, ISBN or DOI. | Extract, summarize and cite. | 1 unit per download, then normal extraction costs. |
Working with a source
The features around every transcript and summary: what each does and what it costs.
A transcript you can click
Click any line to play that moment in the video above it; the line being spoken stays highlighted. Search inside the transcript, step through the matches, and copy a selection with a link to the second it starts.
Cost: Included with the transcript: 1 unit per fetch, nothing extra.
Open a YouTube transcriptMarkdown, Word and PDF downloads
Save a transcript as Markdown with Obsidian-ready frontmatter and linked timestamps, as Word, or as a page to print to PDF, with caption fragments merged into paragraphs. Summaries download the same three ways.
Cost: Free: downloads use no units.
Download a transcript as Markdown or WordChapters and video details
A video's own chapters show as chips that play in the page's player and become headings in the exports. The description and publish date come along too, and the API returns them as metadata.
Cost: Free.
See a video's chaptersYour caption languages, in order
Set up to five preferred caption languages in Settings (for example “fr, en”). When a video has several tracks, your first match is used before the video's default.
Cost: Free, for signed-in accounts.
Fetch a transcript in your language
What it runs on
The engines are named so the behaviour is checkable, not a black box.
youtube-transcript-api and yt-dlp
YouTube caption retrieval
Docling and pypdf
Document extraction and layout-aware Markdown
OpenAI Whisper
Audio and video transcription
Claude by Anthropic
Summaries, Ask This Source, Create From Source
Playwright
Browser rendering for web extraction
Stripe
Checkout and billing portal
How usage is metered
Every plan is a monthly allowance of units. An extraction is 1 unit, whether it is a transcript, a document, a web page or a media transform. AI actions are 10 units at Standard quality or 30 at Deep, and a cached repeat is 1; with your own Anthropic or OpenAI key an AI action is 1 unit. Transcription is priced per started minute, OCR and translation per page. Downloads in every format are free. The cost is shown before an action runs, and every metered API response carries quota headers that report what it cost and what is left.
Two rules hold everywhere: a request that fails is never charged, and a retried API call with the same Idempotency-Key is charged once. The monthly allowance resets each month; top-up units never expire.
The free plan has 100 units a month and covers extraction, PDF editing, Markdown, text to speech and dictation. Pro has 2,000 units and every AI, media and research tool; API has 10,000. Top-up packs work on every plan and never expire: on the free plan a pack unlocks the paid tools, pay as you go, with no subscription. See the pricing page for the current prices.
What is kept and what is not
Uploaded files are processed in memory and discarded; media streams through a temporary file for the duration of the job. What is kept is the extracted text, in your history, so that summaries, questions and drafts can be run on it later. You can export or delete everything from the Account page. The security page lists the controls in detail and says plainly which audits have not been done.
See it on a real source
Paste a YouTube link and follow the transcript through summary, question and draft.