Codeprints Logocodeprints
Back to Work
🤖AI Integration

Sentient Core

Agentic Document Reconciliation Engine

PythonFastAPIOpenAI Embeddingspgvector
Outcome

99.4% Ingestion Reconciliation Accuracy

Outcome

95% Less Manual Invoicing Work

Outcome

Under 5 Seconds Processing Speed

Project Overview

Sentient Core is an agentic AI workflow developed to ingest invoices, parse varying layout formats, and reconcile payments against corporate ledgers.

The Business Challenge

Manual accounting reconciliation was slow and prone to errors. Diverse vendor invoice layouts prevented standard template parsers from capturing values reliably.

Our Solution & Engineering

We developed a FastAPI-based ingestion pipeline. Unstructured invoice PDFs are parsed, converted into vector representations using OpenAI Embeddings, and searched semantically against pgvector records in a PostgreSQL database. Verified items automatically trigger payouts through secure APIs.

Key Integrations

Semantic Invoice Matching

Extracts billing elements by context and meaning rather than relying on absolute layout positions.

Automated Ledger Sync

Matches transaction IDs directly with accounts receivable tables to flag duplicates.

Instant FastAPI Webhook

Triggers third-party payouts immediately upon positive semantic confirmation.

Want similar results for your product?

Let’s discuss your technical bottlenecks and build a scalable solution tailored to your operational metrics.

Work With Us