AI-powered PO import cut order entry time ~50% and errors ~75%.
Athleticknit is a sporting apparel manufacturer serving B2B and B2B2C accounts across Canada and the United States. Custom team uniforms, performance wear, and branded merchandise, sold through resellers and direct corporate accounts. Named public reference — Daniel Sliwin has reviewed FYB publicly on Clutch.
THE PROBLEM
Customers loved placing orders online — but many still sent orders via email in whatever format was fastest: spreadsheets, PDFs, plain-text emails, even photos of handwritten forms. For repeat orders and complex team rosters, attaching a file beat clicking through a web form.
The customer success team was drowning:
- Hours per day transcribing orders from emails and attachments.
- High error rates: typos in SKUs, wrong sizes, missed customisations.
- Every customer used their own spreadsheet layout — no consistent format.
- Team orders with player names, numbers, and sizes across multiple products made every roster a manual translation exercise.
OUR APPROACH
We built an AI-powered order import tool that reads any incoming format — email, Excel, PDF, plain text — and converts it into a valid Magento order.
The CSR drops in an email, spreadsheet, or PDF. An LLM extracts products, SKUs, sizes, colours, quantities, and customisations. Fuzzy matching connects the customer's natural language ("12 red youth medium jerseys") to the actual catalogue. The CSR reviews the parsed order, makes any needed corrections, and confirms — the validated order then flows into Magento as a standard order via the REST API.
Traditional rule-based parsing failed because every customer used a different format. LLM-based document understanding handles the format variation while structured output validation keeps the extraction reliable.
KEY TECHNICAL DECISIONS
- Structured output (JSON mode with Pydantic schema). Consistent extraction across wildly varied input formats.
- Hybrid product matching. Exact SKU match first, then semantic search on product names, then configurable variant matching — balances accuracy with flexibility for natural-language descriptions.
- Human-in-the-loop review. Low-confidence matches flagged for CSR review rather than auto-rejected — preserves throughput while maintaining accuracy.
- Custom schema for team rosters. Per-item customisations (player names, numbers) mapped to Magento configurable-product custom options.
- Native Magento REST API integration. Orders enter the same fulfilment workflow as web orders — no parallel process, no back-office confusion.
RESULTS
- Order entry time cut ~50% on the same volume.
- Error rates dropped ~75% (SKU typos, wrong sizes, missed customisations).
- Fewer reships and returns as a downstream effect of accuracy improvement.
Got unstructured documents that need to flow into your systems?
— A few seconds to fill in a form and we'll do the rest!
Book a 45-minute discovery call. Bring your ticket volume and top ticket types. We'll tell you honestly which parts an AI Support Console can deflect and what an honest pilot would look like.