Automation
DocETL turns extraction, classification and summarization into reusable pipelines. Use Python, YAML or DocWrangler to specify models, instructions and output fields, then review sample results before running the full batch.
For data teams repeatedly processing tickets, interviews or documents. Setup requires a model and familiarity with the CLI, Python or Docker.
The code is MIT-licensed. Model calls, compute and hosting can incur separate costs.
Specify a prompt and output fields for each record in DocWrangler.
Check pipeline progress and inspect individual records and their outputs in the terminal.
It depends on the model and deployment. The hosted DocWrangler requires your own LLM API keys, and its prompt assistants require an OpenAI key.
Check where your deployment and model send data. The hosted version documents server-side storage and caching. You can deploy locally, then check separately whether the configured model receives inputs.
Check OCR and layout quality first. You can parse scanned files with Docling and pass the reviewed text to the extraction pipeline.
The code is MIT-licensed. Model calls, compute and hosting can incur separate costs.
Turn tickets, interviews and documents into structured records in batches.
Visit website ↗Submit once. Get listed permanently after approval.
