Last Wednesday, I attended the Build with Gemini 2026 conference at Expo Tel Aviv. Unlike the usual high-level tech keynotes, this event got straight to hands-on execution.
The setup was divided into two distinct tracks: a Code track for developers, and a No-code track designed for business leaders, product teams, and enterprise innovators. I opted for the No-code track to see firsthand how Google bridges the gap between frontier models and enterprise-grade operational workflows.
Within moments of taking our seats in the main hall, every participant was provisioned with individual credentials to a live sandbox environment featuring full Gemini Enterprise capabilities, diving straight into Google's enterprise ecosystem.
The immediate standout was the speed and friction-free process of orchestrating organizational workflows and deploying custom Agents. The native integration into enterprise systems via the Model Context Protocol (MCP) was particularly impressive, providing a solid foundation for enterprise data connectivity.
There are still areas that need polish. In one instance, the agent struggled with a relatively basic task - generating an email template in clean HTML instead of raw text, an area where competing models currently deliver smoother results. However, several other labs pushed Gemini's context handling and multi-step reasoning to the edge, demonstrating impressive enterprise depth.
The highlight of the day was the custom agent build exercise.
Within minutes, I built an operational Agent focused on Identity Security for enterprise environments:
- Data Ingestion: The agent processes access event logs (CSV format) mapping entity behaviors across the network - distinguishing between human users, service accounts, and autonomous agents, alongside source IPs and target endpoints.
- Anomaly Detection: Powered by Gemini's reasoning layer and targeted security rules, the agent flags anomalies and unauthorized access patterns in real time.
- Executive Escalation: It automatically generates routine audits and immediate trigger-based alerts to the CISO whenever enterprise Identity Security policies are breached.
Being able to prototype a functional, domain-specific security workflow connected to enterprise data in just a few minutes - without writing a single line of code - highlights how quickly enterprise AI tooling is maturing into daily operations.
