Clients of T-Investments can now connect a brokerage account to Alice AI and receive data from the assistant on portfolio status and market conditions. Investors save time on routine analysis, but making trades through artificial intelligence is not yet possible.
The integration gives T-Bank an additional communication channel with clients, and gives investors a reason to decide in advance exactly which data they are willing to entrust to the assistant. The developers plan to add trading through Alice AI, but the user will confirm each operation independently.
What the assistant can do
The account is connected via the T-Investments plugin in the open Alice AI catalog. The assistant and the broker exchange data via the MCP (Model Context Protocol) — a common standard for connecting AI models to external services.
At the current stage, Alice AI responds only to informational and analytical requests: it reports on portfolio status, changes in asset values, and market conditions. This protects the client from an accidental trade due to a misunderstood command. At the same time, the service's functionality is more modest than that of a full-fledged trading assistant.
T-Investments is developing the system for connecting external AI agents to brokerage accounts, launched in August 2026. Clients work with ChatGPT, Claude, and other assistants, independently setting their level of authority — from viewing data to sending trading orders.
T-Bank's statistics are telling: a third of investors who have connected AI agents turn to them almost every trading day, and the average number of requests exceeds 14 million per month. At the same time, only about 5% of users make actual trades through AI — most trust neural networks with analysis, but not with the handling of money.
Competitors are betting on their own ecosystems
SberInvestments launched a similar portfolio breakdown based on GigaChat back on October 8: the service shows which assets and events most strongly affected the value of investments. The difference in approaches is fundamental — the competing platform develops AI within its own app, whereas T-Bank opens access to accounts to external assistants.
Sber went further and announced an AI assistant in Sberbank Online that will consolidate the client's income, obligations, and goals into a single system. The agent will in the future be able, with the user's consent, to connect third-party services and perform actions approved by the user — the same principle of "AI proposes, the human confirms."
Central Bank Deputy Chairman Mikhail Mamuta said that from 2027 the Bank of Russia will embed its own AI assistant into banks' and insurers' apps and websites for selecting deposits, mortgages, credit cards, and CASCO insurance. Commercial assistants select products within a single company, while the regulator's service will compare terms using a unified methodology — financial organizations risk losing some control over which offers the client sees.
Cryptalist analysis: The Russian financial sector is forming a two-tier model of AI integration: banks compete on the convenience of assistants, while the regulator is preparing an independent layer for comparing products, which over time will redistribute traffic and reduce the profitability of sales within ecosystems. The figure of 5% of actual trades through AI reflects the market's maturity, not its weakness: trust in capital automation is formed gradually, through confirmable operations and transparent authority limits. For the crypto segment, this trend sets a benchmark — MCP-compatible agents with human control will become the standard for access to trading accounts within the next 12–18 months.