Scaling financial customer service with reliable AI agent

LähiTapiola Finance partnered with Siili to develop Bobotin, an artificial intelligence agent in production that automates the preparation of customer service responses at LähiTapiola Finance. The agent retrieves unanswered customer messages, generates response suggestions based on contract terms, previous responses and industry terminology, and automatically categorizes feedback and complaints.

A human checks and sends the response while artificial intelligence does the heavy lifting. The solution runs on Microsoft Azure and is built to meet the requirements of financial industry regulation and GDPR. Siili maintains and further develops it as an ongoing Managed AI service.

Siili’s ISO/IEC 42001 certification — the standard for AI management systems  — means in practice that systematic governance, transparency, and ethical consideration are built into the development process. Siili Solutions was granted the certificate as Finland’s first listed company.

For a regulated industry, this is not just a competitive advantage, but a prerequisite.

In financial sector customer service, a significant share of personnel resources is spent on answering repetitive questions: each response requires reviewing contract terms, mastering correct regulation-compliant terminology, and applying previous answers. The work is time-consuming and does not scale as volumes increase.

In addition, identifying and archiving feedback and formal complaints required separate manual work — particularly critical in a regulated industry where the documentation of complaint processes has legal significance.

The financial sector places particularly high demands on AI: GDPR, sector-specific regulation of the financial sector, and the growing EU AI Act govern the processing of customer data.

Bobotin was designed to meet these demands from the ground up. Microsoft Azure’s security and compliance infrastructure aligned with ISO/IEC 42001 certification provided a secure foundation on which the AI agent could be built with confidence.

The agent identifies feedback and complaints, classifies them into categories and subcategories, and routes them through dedicated response templates that comply with regulatory requirements.

The solution automates the preparation of customer responses by retrieving unanswered messages and generating reply suggestions based on contract terms, response templates, and historical interactions while using consistent, industry-specific terminology.

GDPR requirements; including data handling, audit trails, and access control, are implemented using Azure services.

The solution operates seamlessly in Finnish, Swedish, and English, always matching the response to the customer’s language.

Customer service representatives handle more messages per unit of time when the groundwork for responses is already done — simple messages are handled significantly faster than before.

Bobotin's responses consistently follow the correct terminology and terms of service. The quality of the responses has been perceived as high or even better than when produced by humans in simple cases.

Customer service representatives can focus on complex cases where human consideration and empathy bring genuine added value. The routine work no longer takes up space from more demanding service.

Automatic recognition, classification and archiving of feedback and official complaints has been accelerated and harmonized — critical for both customer experience and legal documentation.

GDPR and financial sector regulatory requirements did not slow down the implementation — they were the starting point of the design.

Maintenance, further development, monitoring and quality management of the solution and its AI functionalities

Bobotin is not a project that was delivered and forgotten.

Siili maintains the solution with its own Managed AI service model, which covers the monitoring and continuous development of the solution and its AI functionalities.

This ensures that the solution remains high-quality and evolves with the needs of the business — without LähiTapiola Finance needing to build its own capabilities to operate AI solutions.

Learn more →