Build AI That Answers Right, Remembers, and Stays Governed
Large models are everywhere, but a counterintuitive truth keeps surfacing: most enterprise AI projects fail not at the model, but at the data foundation — knowledge isn't curated, answers can't be traced, agents don't remember context, and data can't safely leave the domain. For financial institutions, "answering right, remembering, and staying governed" isn't a nice-to-have; it's the precondition for going to production.
This is an application-only, closed-door workshop for technology and business decision-makers at FSI, Fintech, and the ISVs that serve them. In one afternoon, we go deep on two things along a single spine — and you build something with your own hands:
Answering right: how an enterprise RAG knowledge base achieves sentence-level source tracing, refusal-without-evidence, and measurable quality — illustrated by AI insurance underwriting.
Remembering & governing: how an agent's long-term memory and context foundation lets financial assistants understand customers across sessions and agents — without data leaving the domain, fully auditable and compliant.
Hands-on lab (60 min): with a dedicated coach and pre-loaded, de-identified financial documents, build a traceable, memory-enabled enterprise knowledge agent and see how it plugs into team collaboration and internal systems.
You'll leave with: a framework to pinpoint where your AI project is stuck in the data loop, a reproducible hands-on result, and a 1:1 architecture clinic with Alibaba Cloud database solution architects.
Format: limited seats, registration reviewed; enterprise attendees (decision-makers in the room), individual learners and those interested purely from a technical perspective are not within the scope of this session.
Date: Friday, 27 November 2026, 13:30-17:00 (GMT+8)
Venue: Hong Kong
Event Organizers: Alibaba Cloud × AMD
(Notes: Venue details will be shared with confirmed attendees.)