Introduction
On its Bank Day in July 2026, SBI launched YONO Ji, an agentic AI virtual assistant available around the clock on the web and mobile versions of YONO Business. Today it answers questions about products, platform features and services for corporate and MSME users. Over time an "agentic" assistant could do more: fetch statements, prepare payment batches, track trade transactions or raise service requests. Measuring an AI assistant in a business banking app is harder than measuring a chatbot in a consumer app, because accuracy errors can have financial and compliance consequences. I'll use a success metrics framework: product context, goals, a North Star Metric and its breakdown, supporting, guardrail, trade-off and counter metrics.
For AI products, add two layers to the usual framework: quality (is the answer correct and grounded?) and safety (did the assistant stay within what it is allowed to do?). Engagement alone can hide a harmful assistant.
Step 1
Product Context
YONO Business is SBI's digital platform for corporate and MSME customers. It replaced the SBI Anywhere Corporate app and covers account management, bulk and supplier payments, approvals, taxes and, since July 2026, mobile trade finance tracking and authorisation.
YONO Ji sits inside YONO Business as a chat assistant on web and mobile.
Key stakeholders:
- Business users: Promoters, finance heads and accounts executives who need answers fast.
- SBI relationship managers and contact centres: Receive fewer routine queries if YONO Ji works.
- SBI product teams: Learn what customers struggle with.
- Risk, compliance and IT security: Need the assistant to be accurate, auditable and within permission limits.
- RBI: Expects fair, transparent and secure use of AI in customer service.
User flow today:
- User opens YONO Ji from YONO Business.
- Asks a question ("How do I add a new supplier for bulk payment?" or "What is the cut-off time for RTGS?").
- YONO Ji answers with steps or links into the relevant screen.
- If unresolved, the user is handed off to the helpdesk or RM.
Lifecycle stage: Introduction. The assistant is new, usage is growing, and the scope will expand from answers to actions.
Practice similar questions
Subscribe to access the full answer
.png){kind=logo}
