Alpha Bank AI Assistant

In 2024, Alfa-Bank held a contest to rethink ALF 1.0 — the assistant in the mobile bank. Together with a colleague, we developed the ALF 2.0 concept in five days, which won 1st place.

Role: Product Designer
Scope: Research • Visual Design • UX/UI • Prototyping • AI Interaction Design • Product Vision
Team: Amgel Dorje, designer ↗
Date: August 2025

Conducted research: analysis of Alf 1.0, 5 corridor tests, surveys of 18 users of similar AI (Alisa, ChatGPT, Oleg), interviews with 10 bank clients.

Defined key functions: financial advisor (budget, subscriptions), banking consultant (products).

Designed interface: new entry point (chat tab), text/voice input, scenario prototypes.

Developed visionary plans: AI ethics, inclusivity, generational adaptation, B2E.

We started by studying Alf v 1.0 from 2021:

An unclear entry point: associated with broadcasts rather than an assistant, resulting in low discoverability.

Voice input only: limits use in noisy places and situations that require privacy.

Low usage: mostly used for navigation, leaving its broader potential unexplored.

Trust in AI: fear of data leaks and reluctance to delegate financial transactions.

No personalization: no recommendations based on user data.

Research methods and comparable AI assistants

And conducted a study on bank customers' attitudes towards AI

Customers are not ready to trust AI assistants with financial transactions.

At the same time, users highly value AI’s ability to personalize offers.

People turn to assistants to check balances and transactions, ask about products and terms, resolve technical issues, and contact support.

Bank customers are willing to ask an AI assistant questions if it reduces the wait for a support agent.

The research shifted the project's focus from automation to decision support. Therefore, we built ALF 2.0 around two key tasks: helping users better understand their finances and selecting suitable banking products.

Users don’t want AI to manage their money. They want it to help them understand their spending

Alf can be a financial advisor

AI can recommend ways to optimize spending and manage a budget.

Alf can be a bank consultant

It can help customers choose products based on their needs. For example, it can find a card with the best cashback terms.

Bank consultant: personalised financial advice

All this is available with voice use as well:

Results

The main outcome of the project is a shift in the interaction model between the user and the financial product, moving from a transactional interface to intelligent financial support, with an increased impact on business metrics. Although the project remained a competition concept, we identified its potential impact on product metrics and defined how the proposed solutions could be evaluated.

MAU

Higher retention among assistant users

CSAT

Higher cross-sell conversion through the assistant

CTR

Higher CSAT for consultation workflows

The project won the Alfa-Bank competition 🏆

Within the concept, I outlined directions in which ALF 2.0 could develop after launch:

Custom scenarios for Zoomers and Gen Alpha

Inclusivity — adaptation for people with disabilities (like Sber's Salut)

Ethics and trust: transparency in data use to reduce user fears

B2E version of the assistant — a service assistant that would advise on how to help the client

Alfa-Bank competition winner certificate