AlphaSense cuts weeks-long data rebuild to hours with dbt and Euno's AI context

AI demos can look flawless on stage, but real production data is scattered across systems and full of conflicting definitions. Sarah Levy, co-founder and CEO of Euno, and Brad Levy, director of analytics at AlphaSense, used a case study to show what it takes to run AI agents on that messier reality: context that is automated, governed and scoped, rather than assembled by an agent on the fly.

Euno continuously crawls metadata across a company's data platforms, ETL and ELT tools, BI, security and observability systems, rebuilding it into a live context graph that captures how data assets relate to one another. AlphaSense layers dbt on top as its governance layer, and the pair walked through three use cases built on that combination.

The first was rebuilding a legacy dashboard model: a 287-field table with 68 dbt models upstream, powering 12 dashboards for roughly 300 users, with the real business logic sitting in the BI layer instead of dbt. Using Euno for context and Cursor as the IDE, Brad rebuilt it as a governed star schema of 13 models in about 15 prompts in a single session, reconciling the new models against the legacy ones for parity before shipping. Work that would normally take three to five weeks came down to a few hours, and roughly half the resulting diff was tests and documentation rather than net-new logic.

The second use case redesigned a billing architecture that had outgrown a single fact table as a second product line arrived. No code was generated: Euno traced the current state to the column level, and the proposed redesign, including two Mermaid diagrams, went to AlphaSense's data architect as a Notion document for review and comment. The proposal took about two focused hours with an agent, against an estimated day and a half by hand.

The third use case, still in progress, scales conversational analytics so that many agents across the business pull from the same governed context instead of each reconciling metadata independently and disagreeing with the others. AlphaSense summed up the payoff in three points: models were more accurate because they came out governed and tested, not just generated; the work was faster, measured in hours instead of weeks; and it was more enjoyable, since time shifted from re-tracing what already existed toward judgment and design.

In the Q&A, the Levys discussed how Euno's live labels, encoded rules such as "certified if it only queries gold-layer dbt models," turn conflicting definitions, like two teams' different definitions of churn, into an explainable choice rather than a guess. They also described how Euno's benchmark-question mechanism continuously checks agent accuracy over time rather than relying on a single point-in-time evaluation.

Source: dbt Labs Blog