Descriptions:
Asaf Bord, a technology leader at Northwestern Mutual, presents a candid and detailed case study at AI Engineer on building GenBI — a generative AI business intelligence agent — inside a 160-year-old Fortune 100 financial services company managing life insurance and wealth management. The talk directly addresses the tension between enterprise risk-aversion and the speed of AI innovation, under the company’s guiding motto of “generational responsibility.”
GenBI allows employees to ask natural language questions about business data and receive answers as a BI analyst would — routing to the right reports, generating query results, and surfacing insights on demand. A key early insight shaped the architecture: Northwestern Mutual’s BI teams reported that roughly 80% of their work was simply routing colleagues to existing reports. This meant GenBI could deliver immediate value without hallucinating new information — just delivering trusted assets faster and more interactively.
Bord outlines a deliberate four-phase incremental delivery strategy: Phase 1 tackled natural language to SQL translation; Phase 2 defined what high-quality metadata and context look like for an LLM in a BI context (insights that also fed a parallel enterprise semantic layer initiative); Phase 3 introduced multi-context semantic search for varied incoming questions. Each phase had standalone business deliverables, giving leadership visibility and the ability to stop funding at any point. The talk is especially valuable for enterprise AI practitioners navigating budget justification, trust-building with executives, and the challenge of working with genuinely messy legacy data rather than clean synthetic datasets.
📺 Source: AI Engineer · Published December 23, 2025
🏷️ Format: Workflow Case Study
