hydra953/invoice-processing-env

Invoice Processing Environment

An OpenEnv environment where an AI agent learns to process invoices — from basic field extraction to complex multi-step reconciliation involving purchase order matching, currency conversions, partial deliveries, and duplicate detection.

Environment Description

Invoice processing is a critical real-world accounts payable workflow. AP departments handle varied invoice formats, match them against purchase orders, convert currencies, detect duplicates, and flag discrepancies — all manually. This environment simulates that workflow across three difficulty tiers.

Tasks

Easy — Invoice Field Extraction

Extract vendor_name, invoice_number, date, and total_amount from structured invoice text.

Scoring (0.25 each):

• —vendor_name: Fuzzy string match (exact = 0.25, partial = 0.15)
• —invoice_number: Exact match
• —date: Parsed date comparison (any format → YYYY-MM-DD)
• —total_amount: Numeric comparison (1% tolerance)

Medium — Invoice-PO Matching

Match an invoice to the correct purchase order and flag quantity/price mismatches.

Scoring:

• —Correct PO identification: 0.40
• —Mismatch detection: 0.60 (proportional to found vs. ground truth, minus false positives)

Hard — Complex Invoice Processing

Handle partial deliveries, currency conversions, and duplicate detection simultaneously.

Scoring:

• —PO match: 0.15
• —Currency conversion: 0.20 (within 1% = full, within 5% = half)
• —Duplicate detection: 0.25 (flag + ID)
• —Partial delivery: 0.25 (flag + delivered lines)
• —Mismatches: 0.15

Action Space

{
  "answer": "JSON string with extracted/processed invoice data",
  "explanation": "Optional reasoning text"
}

Observation Space

Reward Function

• —Continuous 0.0–1.0 with partial credit per component
• —Feedback loop: Detailed feedback after each attempt helps agents improve
• —Hint system: Hints provided after 2 low-scoring attempts
• —Exhaustion penalty: 0.85x multiplier if max attempts reached without perfect score

Setup & Usage

Prerequisites

• —Python 3.10+
• —Docker (for containerized deployment)

Local Development

pip install -r requirements.txt
python -m server.app

The server starts at http://localhost:7860.

Docker

docker build -t invoice-processing-env .
docker run -p 7860:7860 invoice-processing-env

API Endpoints

Running Inference

export API_BASE_URL="https://router.huggingface.co/v1"
export MODEL_NAME="deepseek-ai/DeepSeek-V3-0324"
export HF_TOKEN="your-token"
export ENV_URL="http://localhost:7860"

python inference.py

Baseline Scores