Executive Story Daki: From AI Champions to “Data for All”
Wednesday, 23 September, 2026 / 02:45 PM - 03:15 PM BRT
About the Session
In December 2025, Daki had a solid AI plan: to train AI Champions in every department, using automation tools and a single model as the internal self-service AI infrastructure. The plan already had nine projects delivered and four more underway. In 60 days, the company decided to abandon this approach. The consolidation of open protocols for integrating generative models and technical parity among leading AI providers changed the equation. Instead of training people to operate a tool, the decision was to build a curated semantic layer and connect models directly to company data — no tickets, no dashboards, no queues. In eight weeks, adoption went from 0% to 82% among active users in this flow. The company shut down its BI tool and now runs eight agents in production, with use cases such as fiscal classification of incoming invoices and automated customer service, all on the same semantic substrate, model- and cloud-agnostic. In this session, discover the honest journey of this shift: what survived from the original plan, what had to be eliminated, how the change was justified internally, and lessons learned about culture, governance, and technology choices. Also, get practical recommendations for CIOs at three different levels of AI maturity.
