← Cases
TürkiyeFinance and insurance2025 – ongoing

Akbank

Bank-specific language model for documents and written requests

Akbank reported processing 1.6 million written requests monthly without human intervention

WorkingCompany's own claimOfficial ↗

Problem

Akbank needed to process written requests and documents, including handwriting.

Approach

Akbank trained a bank-specific large language model for document processing. It reported that the model understood handwriting and supported processing written requests without human intervention.

Results

1.6 millionwritten requests processed monthly without human intervention, according to AkbankCompany's own claim[1]

Takeaway

The reported volume did not establish an error rate or the share of all requests handled automatically.

Categories

Business function
Production and quality
Innovation type
Process innovation
AI technology
Generative AI
Evidence
Company's own claim

Sources

  1. Genel Müdür'ün mesajı
    Akbank2026Company publication

Source summaries

[1]Akbank2026

Akbank described its handwriting-capable language model. It reported 1.6 million written requests processed monthly without human intervention.

Full details are in the original sources.

Added 1 Nov 2026 · Updated 1 Nov 2026
Open case page ↗