With Google Cloud in place, Convert Human Digital Business built Verus, an analytical intelligence ecosystem that combines Looker's semantic layer, specialized AI agents, and advanced prompt engineering to turn marketing data into governed, traceable answers.

BigQuery serves as the data warehouse, holding all marketing and transactional data from Convert's clients. Looker sits on top as the foundation of Verus, with its LookML semantic layer acting as what the team calls a business contract written for the AI.

Every metric, every dimension, every synonym, and every client-specific business rule is documented inside that layer. When an agent receives a question, it doesn't guess what the data means. It consults the same source, using the same definitions, every time.

"The agent always looks at the same place, the same metric, and takes decisions based on a single source of truth," adds Gasparetto.

Verus has five specialized agents, each named after a pioneering woman in science and technology.

While Ada handles media performance and time series analysis, Grace manages media investment pacing, telling the team where to reallocate budget based on historical data. On the other hand, Radia covers paid and organic search, Parisa focuses on creative performance at a granular level, and Hedy is the ecommerce specialist, tracking revenue, sales, and product performance.

"Governance is not an isolated layer at the end," Gasparetto explains. "It is part of the entire workflow. Every answer is born from data."

In practice, a client types a business question and receives a structured, sourced answer in minutes instead of days. Over time, as the team learns what the agents can do, the questions get deeper. "They start simple and then begin to understand the potential," affirms D'Elia. "It becomes a snowball."

Verus changed how Convert measures success. Report delivery time dropped by 65% and response times for complex analysis are now 8.4 times faster, meaning clients can define their monthly strategies at the start of the month instead of waiting for reports that arrive after the 15th.

Operational costs fell by 32%, and the classic 80-20 drain in analytics has inverted. Where 80% of the team's time once went to repetitive questions that answered only 20% of the business, that ratio has flipped.

"That 80% now goes to AI," says D'Elia. "The team focuses on the 20% of questions that drive 80% of the business, and they are doing it faster and with more precision than before."

That shift is changing what it means to be an analyst. Work that was previously done by senior analysts is now handled by Verus, raising a question the team is actively exploring. "What is this senior analyst now?" D'Elia reflects. "An AI developer? An AI analytics developer? We are looking for these new directions that the market is bringing."

Looking ahead, Convert is building toward full autonomous deliverable generation, where Verus produces complete reports and presentations, not just answers to individual questions. The next version will also include trigger-based proactive alerts, where the system signals the right people the moment a significant event happens in the data without waiting for anyone to ask. As part of Grupo Dreamers, Convert is also planning a Verus 2.0 built on Gemini Enterprise as the group migrates to Google Workspace.

For other companies around the world facing the same analytical scale problem, the advice from both Felipe and Matheus is the same. "Data foundation is everything," D'Elia emphasizes. "Without it, nothing will work. If the data is wrong at the start, everything that follows will be wrong."