One workflow for multiple formats
Process text, image and Excel inputs into a consistent seven-column structure for business users.
Purchase AI Recognition
Turn varied requests into consistent purchasing data.
Process text, images and Excel requests through one parsing workflow, organizing customer requirements into seven fields for purchasing operations.
- input types: text / image / Excel
- 3
- standard business fields
- 7
- support users
- 30
A better way to work
BEFORE
Customers send requirements as messages, screenshots and spreadsheets. Support teams must interpret product details, separate quantities from packaging and prepare usable records.
WITH THE PRODUCT
A consistent output organizes product, brand, catalog number, specification, packaging, quantity and notes for review and downstream purchasing.
Core capabilities
Give each field its business meaning
Parsing constraints distinguish packaging, specifications and order quantities. Output repair and model fallbacks support reviewable results.
Preserve the original request
Developed and validated source-field and link preservation for Excel, aligning parsed output with the original information for review and traceability.
Handle repeated requests efficiently
Developed and validated thread-pool execution, concurrency controls and caching to improve how synchronous calls and duplicate files are handled.
How it works
- 01
Submit text, image or spreadsheet
- 02
Parse seven business fields
- 03
Review content and sources
- 04
Continue purchasing workflow
PRODUCT / ENGINEERING
What I delivered
I mapped the existing recognition workflow and analyzed 307 production logs to identify field alignment, quantity interpretation and source-preservation issues. I independently developed and validated parsing constraints, output recovery, caching and concurrency improvements.
Does your team face a similar challenge?
Start with one workflow. Let’s explore how AI could fit your operations.
Keep exploring
All workEnterprise fulfillment assistant0→1 · Solo build
LD Assist
An enterprise fulfillment agent powered by a locally deployed large language model. Connect orders, purchasing, warehouse records and follow-ups so support teams can check progress, confirm batches and prepare reply drafts.
- support users
- 30
- queries per day
- ~300
跟一下
Personal follow-up app0→1 · Solo build
跟一下 (Genyixia)
A lightweight iOS app for replies, refunds and documents that still need follow-up. I independently designed and built the product, connected its paid features and shipped it on the App Store.
- Available now
- App Store
- Native app
- iOS
- Subscription & lifetime
- Pro