Two AI readers agreed—and both were wrong

Description and use

In TransferDue’s workflow, a single AI reading of Thai property documents often misreads digits. Adding a second reading and a cross‑family check with Gemini reduced unflagged wrong answers by 38.6 % but increased warnings, making errors more visible but adding manual review work.

Problem

A single AI reading of Thai property documents can misinterpret digits, producing wrong but plausible values that go unchecked.

Outcome

The new rule set reduced unflagged wrong answers by 22 (from 57 to 35) and increased warnings by 30, making errors more visible but adding manual review work. Adding Mistral OCR increased correct answers by 19 but also added 3 more wrong answers without warnings.

Workflow

• Run first Claude reader to extract values
• Run second Claude reader independently
• Run Gemini reader from a different model family
• If first two readers agree, skip third Claude read; otherwise request third read
• Apply business rules: if majority vote from same family, accept; if family conflict, flag as warning; if Gemini disagrees with majority, flag as warning; if Mistral OCR available, compare and possibly correct

Evidence

Wrong answers accepted without a warning fell from 57 to 35—a 38.6% reduction

Source

The new rules produced 22 fewer wrong answers without warnings

Source

The system can now flag the value for checking. In the code, that status is called family_conflict

Source