Waste invoices, audited: from a hauler's PDF to structured data, and into the books.

IntellaWaste gives businesses visibility into what they spend on waste: invoice auditing, contract and billing workflows, and the reports that come out of them. The invoices themselves arrive from the haulers as PDFs.

That a stack of invoice PDFs can become data an audit can trust, and that the system knows when not to trust itself.

Most of the work is in the data rather than the screens: a GIS-enabled Django schema of around 60 models, held up by 1,386 backend tests.

1. 01

   Invoice intelligence

   extract(pdf) → json

   PDF to images, then multimodal extraction with OpenAI and Gemini, run as background jobs with retries, status tracking, and sync into the platform's import workflow.

2. 02

   Reconciliation guard

   totals ≠ lines → human

   When the extracted total doesn't match the line items, the invoice is flagged for a person instead of imported.

3. 03

   One name per hauler

   hauler.normalize()

   Fuzzy matching on hauler names, so the same company never becomes two records.

4. 04

   The backend & the books

   /api/v2 · export.iif

   A v2 REST API of 15 Django REST Framework modules tuned for heavy queries, an audit trail on invoice changes, and QuickBooks IIF export that turns billing reports into AP transactions.