Blog

Asia Alliance Bank and Gorgona AI to jointly advance AI in customer service

The system drafts the reply, a bank employee makes the decision, and the data never leaves the bank's perimeter
Asia Alliance Bank and Gorgona AI have signed a memorandum of cooperation on the use of artificial intelligence to improve the quality of customer service. The parties intend to develop a model in which customer enquiries from different channels are consolidated in a single workspace for the support agent, and the system drafts a suggested reply based on the bank's internal materials — with a reference to the source the reply is based on.
The topic became measurable in 2026: the Central Bank began publishing an index that relates a given bank's share of complaints to its share of customers in the market. In the second quarter of 2026, the regulator received 11,331 enquiries concerning the activities of banks. What used to be an internal metric has become a public figure.
At the same time, the workload on the first line is growing faster than its headcount. Asia Alliance Bank's customer base grew by 25.8% over 2025, primarily in the retail segment. Customers reach out through the mobile app, by phone, in messengers and at branches — and expect the bank to remember the context of their question. Agents have to piece the enquiry history together and search manually for the answer in internal regulations, tariffs and product terms.

What the parties will work through

  • a single window for handling enquiries from all channels, with classification, prioritisation and routing;
  • preparation of suggested replies for the agent based on the bank's internal knowledge base — with a source reference;
  • structuring the knowledge base in a form suitable for use by AI systems, and keeping it up to date;
  • a control layer: masking of personal data, input and output filters for the model, logging and an audit trail;
  • training bank employees to work with artificial intelligence tools.
Two principles are set out separately in the memorandum. First, the decision on the substance of an enquiry is made by a bank employee: the system plays a supporting role and does not make decisions affecting customers' rights and legitimate interests. Second, sovereignty of processing: solutions are designed for deployment within the bank's own infrastructure, and customers' personal data and information constituting bank secrecy are not transferred to external services. Development and prototyping are carried out on anonymised or synthetic data.
"Support is the area where the effect of artificial intelligence is easiest to measure: response time, the share of questions closed on the first line, repeat enquiries. It is also the area where projects most often break down: customer data cannot be handed over to an external service, and an answer whose source cannot be verified is of no use to a bank. That is why we deploy the solution inside the client's infrastructure and show the agent which document the suggested reply is based on," says Alexander Simonenko, Gorgona AI.
The parties have agreed to define a list of measurable service quality indicators — response time to an enquiry, the share of enquiries resolved at first contact, the share of repeat enquiries, and customer satisfaction — and to measure their baseline values before work begins.
The memorandum is a framework agreement. The first stage — forming a working group and agreeing the channels, priority enquiry topics and the parameters of the pilot scenario — is scheduled for 30 days from the date of signing. Prototyping is carried out in an isolated environment; a pilot in the bank's own environment is possible subject to positive results and will be formalised under a separate contract.

Related articles

2026-08-28 14:46 News