Divakar Kumar on AI Agents for Event-Sourced Fraud Review

AI Engineer21m 37s
Divakar Kumar in a blue shirt with one hand at his chin beside an AGENTS FOR FRAUD REVIEW headline in white and blue on black.
0 comments · 0 votesOpen discussionClose discussion
Sign in to join the discussion

    Kumar starts with a declined laptop purchase to explain why a conventional rule or machine-learning score may block a legitimate transaction without giving customer support a clear reason. His proposed design retains existing checks as tier one and sends ambiguous cases to a second tier of agents. It is presented as an addition to established business systems, not a replacement for every fraud decision.

    The architecture separates transaction, account, device and payment data into bounded contexts. Domain eventsEvent sourcing stores changes as a sequence of events from which the current state can be derived. and change feeds or message brokers feed read-optimized projections, creating a contextual view for the agent layer. A risk agent and a behavior agent inspect that view through defined tools; a third verdict agent combines their findingsAgent orchestration coordinates AI agents, tools, people, tasks, state, and control flow so a larger workflow reaches a verified outcome. and emits an eventAn event-driven AI agent reacts to typed events instead of following only one fixed workflow or schedule. back into the wider payment workflow.

    Kumar notes a sub-500-millisecond transaction requirementLow-latency AI inference produces a model result quickly enough for an interactive or time-sensitive application. and the need to bound an agent's reasoning loop. A closing proof of concept uses synthetic eventsSynthetic data is information generated to represent examples or situations rather than collected directly from the events it describes., but the database starts slowly and the captions do not show a completed end-to-end verdict. The transcript therefore supports the architecture and intended workflow, not measured fraud accuracy, latency or production readiness.

    Original YouTube thumbnailWatch on YouTube

    Video topic coverage

    • Agents & workflows100
    • Coding & tools25
    • Models & learning10
    • Infrastructure & compute50
    • Research & evaluation25
    • Safety & security50
    • Business & productivity75
    • Creative & media0
    • Robotics & perception0
    • Society & policy0

    Share this video

    Divakar Kumar in a blue shirt with one hand at his chin beside an AGENTS FOR FRAUD REVIEW headline in white and blue on black. With WWW.ARTIFICIAL-INTELLIGENCE.VIDEO, 30 July 2026 and duration 21m 37s.

    Divakar Kumar proposes sending uncertain fraud cases from existing rules and machine-learning checks to agents that use event-sourced projections for context, while the synthetic demo leaves performance unproven.