Netomi’s lessons for scaling agentic systems into the enterprise
Netomi shares three key lessons for deploying reliable, scalable agentic AI systems in the enterprise: build for complexity, parallelize for low latency, and embed governance into the runtime.Summary
Netomi, which builds agentic systems for Fortune 500 clients like United Airlines and DraftKings, outlines three essential lessons for scaling AI agents reliably in enterprise environments. First, systems must be built for real-world complexity, handling messy, multi-system workflows by using a governed orchestration pipeline that leverages GPT-4.1 for fast tool use and GPT-5.2 for deeper planning, guided by agentic prompting patterns like persistence reminders and structured planning.
Second, to meet strict enterprise latency expectations, Netomi emphasizes parallelization over sequential execution, utilizing GPT-4.1's low-latency streaming and stable tool-calling to ensure the entire system remains responsive under extreme load, such as during traffic spikes at DraftKings.
Third, governance must be intrinsic to the runtime, not an afterthought. Netomi's architecture integrates mechanisms for schema validation, policy enforcement, PII protection, and deterministic fallbacks directly into the execution layer, ensuring trustworthy and compliant operations, especially in highly regulated industries.
(Source：OpenAI)