In May 2023, JPMorgan filed a trademark application for IndexGPT with the US Patent Office. The official line: a ChatGPT-like tool, but built on JPMorgan's investment-advisory data. Through 2024-2025 the idea expanded — in-house LLMs, customer-interaction AI, fraud-detection AI.
This move affects not just JPMorgan but every major bank in the world. The question: are Turkish banks following the same path? And what does it mean for the financial software industry?
What happened? The banks' insourcing strategy
Traditionally, large banks bought their financial software from outside: SAP, Oracle, FIS, Fiserv, Temenos. Now the architecture is changing:
Old model:
- Core banking: Temenos or similar
- Customer interaction: Salesforce
- AI/analytics: third party
New model:
- Core banking: still third party
- Customer interaction: their own LLM (proprietary)
- AI/analytics: their own model, their own data
JPMorgan's IndexGPT, Goldman Sachs' internal AI, and Morgan Stanley's AsyncMS projects are all examples of this architectural shift.
For a bank, owning the model is strategic: customer-data privacy + competitive advantage + regulatory compliance.
Why are banks building their own AI?
Three main reasons:
1. Data privacy
A bank cannot send customer data to OpenAI or Anthropic — KVKK (Turkish data protection law), GDPR, SOX, and banking regulation forbid it. The answer: on-premise or vendored AI.
2. Competitive advantage
JPMorgan's 50 years of investment-advisory data give IndexGPT a unique edge. With a generic model, that edge would vanish.
3. Regulatory compliance
The EU AI Act, FED guidance on AI, and in Türkiye the BDDK (Turkish banking regulator) AI guidance all point the same way — financial AI must be auditable. Full oversight is only possible on a model you own.
The picture in Türkiye: Garanti BBVA, Akbank, İşbank
Türkiye's largest banks are on the same path:
Garanti BBVA — UGI and Garanti AI
In 2021, Garanti BBVA deepened its integration with the BBVA AI Factory. Through UGI (Universal Generic Interface) and the next-generation "Garanti AI" projects, it is building customer-interaction AI.
Akbank — Akbank Lab
Akbank Lab invests in the startup ecosystem and runs an in-house AI team. In 2024-2025 it launched conversational banking, fraud AI, and credit-decision AI projects.
İşbank — a Turkish LLM
In 2024, İşbank announced a project to train a Turkish financial LLM — adapting general-purpose models to Turkish financial language and banking jargon.
What does this trend mean for the financial software industry?
Three architectural consequences:
1. Bank API ecosystems are opening up
As banks build their own internal AI, they are exposing open APIs at the outer layer. PSD2 (in Europe) and Open Banking in Türkiye support this opening. Customers are gaining more control over their banking data.
More on this: MiCA and PSD3: the EU's new finance rules.
2. Finance software is the **tool**, bank infrastructure is the **data**
Finance software used to mean "ERP + one bank integration." The new model: the software connects automatically to multiple bank APIs and taps into each bank's AI services. The software layer = orchestrator.
iFinances' three-way reconciliation model runs on exactly this logic: bank + e-invoice + ledger, multi-source orchestration.
3. Niche financial AI is gaining value
JPMorgan is building its own LLM — yet for niche use cases (reconciliation, VAT, accounting firms), specialized AI software remains valuable. Banks build their own AI for broad use cases like investment advisory; they don't build it for reconciliation.
iFinances is positioned in precisely that niche: AI specialized for Turkish finance operations.
What it means for Turkish finance teams
If you are a large enterprise:
- Audit your bank APIs — which services can you actually use?
- Find out which metrics you can pull from your banks' AI services (fraud risk, maturity forecasts, FX recommendations)
- Build a bridge between your internal software and your banks' AI
If you are an SME:
- Instead of integrating directly with bank APIs, use orchestrator software (iFinances, Logo Connect, Mikro Cloud)
- Capture value through niche financial AI solutions
Bottom line
Banks are insourcing financial AI software — their own data, their own architecture, their own regulatory compliance. This is not the end of the financial software industry; the industry is shifting into a different position.
The new role: orchestrator + niche specialist. iFinances sits exactly there: it connects to a Turkish finance team's banks, GİB (Turkish Revenue Administration), and ERP, and layers explainable AI on top.
One post a month.
Get new insights straight to your inbox. No spam, just well-crafted reads.
