Knowing by reasoning, not vectors.
Deep and reliable. Vectorless plays nicely with your documents. Ask questions in plain language; Get answers by reasoning with Vectorless.
Warning
Experimental project — not production-ready. Vectorless is under active, rapid development. Releases are cut on an irregular schedule and may introduce breaking API changes without notice. It is currently positioned as a research/experimental project and makes no stability or production-readiness guarantees. Pin a version and expect churn.
Install using pip install -U vectorless. For more details, see the Installation section in the documentation.
importasynciofromvectorlessimportEngineasyncdefmain():
asyncwithEngine(api_key="sk-...", model="gpt-4o") asengine:
# Compile a documentdoc=awaitengine.compile(path="./report.pdf")
# Ask a questionresponse=awaitengine.ask("What is the total revenue?", doc_ids=[doc.doc_id])
print(response.answer)
asyncio.run(main())Installing vectorless also provides a vectorless CLI for indexing and querying from your terminal:
vectorless init # create a .vectorless/ workspace
vectorless add ./report.pdf # index a file (or a directory with -r)
vectorless query "total revenue?"# ask a one-off question
vectorless ask # interactive query REPL
vectorless list # list indexed documents
vectorless info <doc_id># show a document's details
vectorless tree <doc_id># visualize the document tree
vectorless stats # workspace statistics
vectorless remove <doc_id># remove a document from the index
vectorless config llm.model gpt-4o # view / set configurationRun vectorless --help (or vectorless <command> --help) for all commands and options.
See documentation for more details.
Contributions welcome! See Contributing for setup and guidelines.
Apache License 2.0