People working together

adesso Blog

In a nutshell, the answer is yes, since banks and insurers can run analytics and AI entirely on-premises, next to their existing data warehouse, by adding an in-memory analytics engine such as Exasol instead of moving data to a public cloud AI service, with the help of adesso Türkiye.

Through their partnership, adesso Türkiye and Exasol deliver this as a sovereign-by-design architecture. Hence, your data stays in your data center, heavy compute moves off the core database, and license costs stop tracking the exchange rate as aggressively.

Most AI conversations in financial services start with a cloud platform and end with a compliance question. For a Turkish bank, that question usually arrives first: where does the data go, and who can see it? In this article, we're here to explain how, where, and with whom to start.

Why Data and AI Workloads Cannot Leave the Data Center?

Data location cannot be a preference since it's a design constraint for all regulated Turkish financial institutions. Two frameworks shape most architecture decisions: BDDK (Banking Regulation and Supervision Agency): its information systems rules expect banks to keep primary and secondary systems and critical data under tight control in Türkiye. Outsourcing core banking data to public cloud services is heavily restricted. KVKK (Personal Data Protection Law): personal data of customers is subject to strict conditions, including for cross-border transfers. Training or scoring a model on customer data in a foreign region can trigger exactly those conditions.

The practical result is quite simple. A generic cloud AI service is often not an option for customer, transaction, or claims data. As a result, the AI has to come to the data. Each bank should confirm the exact scope with its own compliance and legal teams. The architecture explained below is built to make that conversation easier, not to replace it.

Four Pressures on Bank and Insurer Data Teams

Data leaders in Turkish banks and insurers tell a consistent story, and eventually, four pressures stack on top of each other.

  • Data and AI cannot leave the data center. Every proof of concept that needs an external platform stalls in risk review.
  • FX-priced licenses eat TL budgets. Database licenses priced in foreign currency grow in lira terms every year, while the business budget does not.
  • Heavy reporting and new AI workloads strain the core database. Regulatory reports, risk runs and data science queries compete with transactions and daily operations for the same resources.
  • AI governance is now on the regulator's agenda. The BDDK and KKB AI sandbox signals that supervisors want to see how models are built, tested, and controlled.

These four don't represent themselves as four separate projects. They share one root cause: analytics and AI are being forced onto infrastructure that was never designed for them. Solving them together is what makes a sovereign AI architecture worthwhile, and that's exactly when you might need Exasol and adesso Türkiye.

Sovereign by Design: Exasol Next to Your Existing Database

The key behind this idea is to add a dedicated analytics layer inside your own data center, not to replace the warehouse you already trust. Exasol is an in-memory, columnar, massively parallel (MPP) analytics database that can be deployed on-premises, in the cloud, or in hybrid setups, without a rip-and-replace migration.

In a typical adesso Türkiye design, the flow looks like this:

  • Source systems and the existing warehouse stay where they are, inside the bank's perimeter.
  • Exasol runs on-prem as the analytics and AI engine, loading the data it needs through parallel import or querying external sources through virtual schemas.
  • BI tools, risk engines and data science environments connect to Exasol over standard SQL, so reporting and model workloads no longer touch the core database.
  • Models are trained and scored inside the same environment, so no customer data is sent to an external AI service.

Because this flow allows every component to sit in your data center, your data never moves as far as BDDK and KVKK reviewers are concerned.

A Fraction of the License Cost, and Less FX Exposure

When analytics and AI run on the core database, every extra core, tier, or option adds to a foreign-currency license bill. Moving those workloads to a purpose-built engine changes the equation in three ways:

  • Smaller footprint for the same work: in-memory, columnar compression and MPP processing mean fewer resources are needed for heavy queries.
  • Stop scaling core database licenses for analytics. Your existing database is sized for daily transactions, not to absorb the load of every new report or data model.
  • Predictable planning: a right-sized analytics layer gives finance a clearer, more stable line to budget in TL.

Knowing that the savings depend on your current estate, volumes, and contract terms, adesso Türkiye starts with a total-cost review instead of a generic promise that may or may not fit your needs. The goal is a comparison you can take to your CFO: current license and hardware spend against the proposed on-prem analytics layer.


Frequently Asked Questions

See What Exasol Can Do for You

Your data. Your use case.

You don't have to commit to a licence to see the difference. Try Exasol on your own data with support from adesso Türkiye's engineering team. Discover how much faster your analytics and AI workloads can run in your own environment.

Get Your Free Consultation

Author Melek Pervan

Content Specialist