From signup to first answer
in about 5 minutes.

Step-by-Step

1. 1

   Create an account and claim starter credits.

   Sign up at /app — no credit card required to start. Starter credits cover indexing and querying a handful of documents.

2. 2

   Open the TurboDex console and upload a PDF.

   Navigate to the Documents tab and drag-drop or select a PDF. For best first results, use a structured document — a technical spec, report, or policy manual.

3. 3

   Queue indexing and monitor job state in queue activity.

   Indexing runs asynchronously. The job dashboard shows parse progress, OCR status, and hierarchy-tree construction in real time.

4. 4

   Run your first query and inspect hierarchy-aware results.

   Ask a question scoped to a section — e.g. "What does Section 4.2 say about risk?" — and observe how the answer preserves structural lineage.

5. 5

   Repeat with real documents and compare answer quality.

   Upload your actual working documents and run the same queries you would use with a flat-vector approach. Note the difference in context fidelity.

API Example — Upload & Queue

Use the REST API to integrate document indexing into your own pipeline.

curl -X POST "https://app.turbodex.ai/api/documents" \
  -H "X-API-Key: YOUR_API_KEY" \
  -F "file=@/path/to/document.pdf" \
  -F "model=synthesis"

Use model overrides sparingly for large or complex documents. TurboDex applies adaptive routing and timeout controls for complexity-aware indexing automatically.

curl -X POST "https://app.turbodex.ai/api/documents/{doc_id}/query" \
  -H "X-API-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query": "What are the key findings in Section 3?"}'

First-Use Tips

• → Start with a structured PDF. Technical specs, annual reports, and policy documents index better than unstructured scanned text.
• → Let indexing finish before querying. The job panel shows completion status. Querying before tree construction completes will return partial results.
• → Try section-scoped questions. Ask "What does the executive summary say about cost?" rather than generic "summarize this." Hierarchy-aware queries demonstrate TurboDex's accuracy advantage most clearly.
• → Check the job queue for large files. Very large PDFs (>200 pages) use async indexing. Poll /api/jobs/{job_id} for status.