Local demo · Optional live AI

Evidence-grounded study pack

Turn lectures into notes you can study.

LectureWeaver audits what is missing, rebuilds your notes into a clearer learning guide, and creates optional Anki cards—each tied back to the source that supports it.

Judge-ready sample

Build a complete study pack in one click.

Real local PDF, TXT, and Markdown parsing
Reorganized notes with trusted evidence
Anki-ready cards with no API key needed

No API key · no model request · about 3 seconds

Your materials

Build a trusted source map.

Add at least one lecture source: a lecture PDF/TXT or pasted text, a transcript, or a transcribed recording. Existing Markdown notes are optional—include them for coverage and contradiction checks, or omit them to build from scratch.

01 / Lecture source

Lecture material

Choose or drop a PDFText-based PDF · up to 10 MiB

02 / Spoken context

Transcript file, pasted text, or audio

Choose or drop a transcriptUTF-8 TXT · Up to 1 MiB

03 / Optional baseline

Existing notes

Choose or drop a fileMarkdown · Up to 1 MiB

Optional live analysis

AI providerAI model

Flagship quality through the stable GPT-5.6 alias.

Local source map only

Use a temporary key in this tab

Enter separate OpenAI, DeepSeek, or Kimi credentials. A valid temporary key takes priority over that provider's deployment key and disappears when you clear, reset, close, or reload this page.

Temporary OpenAI API key

Not entered

Temporary DeepSeek API key

Not entered

Temporary Kimi API key

Not entered

Kimi API region

LectureWeaver keeps these keys only in this page's memory. It does not save them to browser storage, cookies, URLs, logs, or a database.

For a live request, the selected key crosses this site's HTTPS Vercel function and is forwarded only to the selected provider. Your browser, developer tools, extensions, and the provider can still see it; use a trusted device.

Server configuration

Connect an AI provider

For a quick personal run, use the masked temporary-key panel above; LectureWeaver keeps that value only for the current tab and does not save it. Server configuration remains the recommended deployment setup.

Where do I add an API key?

Local development

Create an untracked .env.local file in the repository root.
Add the key for each provider you want to enable. Model variables are optional and use the defaults shown below.
Restart npm run dev after changing environment variables.

Vercel deployment

Open your Vercel project, then go to Settings → Environment Variables.
Add the provider keys you need to the appropriate Production, Preview, or Development environments.
Redeploy after saving the variables. Never commit a key to Git or use a NEXT_PUBLIC_ prefix.

Supported environment variables

Blank key lines mean that provider is disabled. The model values below are the application defaults.

OPENAI_API_KEY=
OPENAI_MODEL=gpt-5.6
OPENAI_TRANSCRIBE_MODEL=gpt-4o-transcribe-diarize
OPENAI_TTS_MODEL=gpt-4o-mini-tts

DEEPSEEK_API_KEY=
DEEPSEEK_MODEL=deepseek-v4-flash

KIMI_API_KEY=
KIMI_MODEL=kimi-k3
KIMI_REGION=cn

KIMI_REGION accepts cn (default) or global. Audio transcription and speech require either a temporary OpenAI key or deployment OPENAI_API_KEY.

API billing is separate

ChatGPT and Codex subscriptions do not include API credits. A temporary key bills its owner; a server key bills the deployment owner.

Before enabling keys on a public site

Add authentication or access control, rate limits, quotas, provider budget alerts, monitoring, and an emergency disable path. Otherwise keep the deployment private or leave provider keys unset and use the no-key demo.

Study pack outputs

Choose the language for the next live analysis. Existing results are not translated automatically.

Output language

Enhanced notes

Always included

Always rebuild a complete, logically ordered Markdown study guide.

Create Anki cardsOptionalAdd evidence-grounded Basic cards that cover every core assessment.

Output choices apply to the next analysis. An existing study pack stays available until you run it again or replace a source file.

The deterministic included demo remains a fixed English sample.

PDF, TXT, Markdown, and pasted text stay local · audio is sent to OpenAI only for explicit transcription · normalized chunks are sent only after you explicitly start live analysis · no silent truncation

Add lecture material, a transcript, or a completed audio transcription. Only one is required.

Enter a valid temporary OpenAI key above, or configure its deployment key.

Extract

PDF pages and numbered paragraphs

Analyze

No-key demo or an explicitly keyed live model

Build

Enhanced notes, trusted evidence, and study cards