Manual Transcription Bottleneck
Staff opened each scanned document alongside a PowerBuilder DataWindow and typed in reference numbers, dates, and line-item amounts by hand.
Case Study · PowerBuilder Consulting
Data Entry, Automated — AI Outside PB, Value Inside It
An external document-intelligence AI service, connected through a REST/JSON gateway, cut manual re-keying from a live PowerBuilder workflow.
3 Wks
POC duration
Zero
PB code rewritten
1 Screen
POC scope
The Challenge
Manual Re-Keying Was Slow and Error-Prone
A mid-market logistics operator processed a high volume of scanned bills of lading and invoices every day — and every field was typed in by hand.
Transcription Errors Downstream
Mis-keyed values surfaced later in reconciliation and billing, creating rework that was harder to trace back to the original scanned document.
No Path to Automation Inside PB
The PowerBuilder application had no built-in way to read a scanned document — automation meant looking outside PowerBuilder for the capability.
The Approach
An External Document AI Service, Called Through a Gateway
The AI capability was built and hosted entirely outside PowerBuilder. A secure REST/JSON gateway is the only thing the application talks to.
Use Case & Screen Selection
One document type — bills of lading — and the one PowerBuilder DataWindow used to key them in were selected as the POC scope.
Document Intelligence Service Setup
An OCR and document-classification AI service was stood up independently of PowerBuilder, trained and tested against a sample set of real historical scans.
REST/JSON Gateway
A secure gateway was placed in front of the AI service, handling authentication and mapping extracted fields to a structured JSON response.
PowerBuilder Integration
The existing DataWindow was wired to call the gateway with a scanned document reference and populate its fields from the JSON response — no other logic changed.
Accuracy Validation & Rollout
Extracted data was checked against real historical documents before staff began relying on it, then the pattern was scaled to additional document types.
Value Visible From Week One
A three-week POC was enough for staff to go from manual transcription to reviewing AI-extracted data — a fast, low-risk way to prove the concept before wider investment.
No PowerBuilder Rewrite Required
The AI capability lives entirely outside PowerBuilder behind a REST/JSON gateway, so the existing DataWindow and business logic needed zero rework.
Staff Redirected to Higher-Value Work
Automating field-level transcription freed staff to focus on exception handling instead of repetitive data entry, cutting downstream reconciliation errors.
The Impact
What the Client Gained
- Manual data entry time reduced significantly per document
- Extraction accuracy validated against real historical documents before rollout
- Staff redirected from data entry to exception handling
- Zero changes to core PowerBuilder business logic
- AI service and gateway scaled to additional document types after the POC
"The PowerBuilder screen looks exactly the same — it just shows up already filled in. That's the whole story." — OptiSol AI Integration Lead
