Why You Care
Ever wondered if AI could truly handle the messy, unpredictable tasks of a large business without breaking down? What if AI agents could reliably manage complex customer service, like rebooking flights during a storm? Netomi is showing how this is possible, offering crucial insights for anyone building or deploying AI in the enterprise. This news directly impacts how your business can adopt AI safely and effectively.
What Actually Happened
Netomi, a company serving Fortune 500 customers such as United Airlines and DraftKings, has developed a blueprint for scaling agentic systems into the enterprise, according to the announcement. They built their system using OpenAI’s GPT-4.1 and GPT-5.2 models. This system is designed to manage complex workflows reliably and predictably, even under heavy load. The company reports that their system pairs GPT-4.1 for low-latency, reliable tool use. Meanwhile, GPT-5.2 handles deeper, multi-step planning. Both operate within a governed execution layer, ensuring predictable model-driven actions in production conditions, as detailed in the blog post.
Why This Matters to You
Running agentic systems at an enterprise scale presents unique challenges. Netomi’s experience provides valuable lessons for anyone looking to implement similar solutions. Their approach emphasizes building for real-world complexity, not just idealized scenarios. This means considering workflows that span many systems like booking engines, loyalty databases, and CRM systems. The data in these systems can often be incomplete or conflicting, as mentioned in the release. Netomi designed its Agentic OS to place OpenAI models at the core of an orchestration pipeline. This pipeline is specifically built to handle such ambiguity.
Imagine you’re a customer service manager. Your team deals with inquiries that require checking multiple databases and applying complex policy logic. A system that can reliably navigate these tasks frees your human agents for more nuanced interactions. This is where Netomi’s lessons become incredibly relevant for your operations.
“Our goal was to orchestrate the many systems a human agent would normally juggle and do it safely at machine speed.” —Puneet Mehta, CEO, Netomi
How might your business benefit from AI agents that can consistently manage multi-step, complex tasks with speed and accuracy?
| Netomi’s Agentic Prompting Patterns |
| Persistence Reminders: Helps GPT-5.2 maintain context across long workflows. |
| Explicit Tool-Use Expectations: Guides GPT-4.1 to use tools for authoritative data. |
| Structured Planning: Leverages GPT-5.2 for outlining and executing multi-step tasks. |
| Agent-Driven Rich Media Decisions: Enables GPT-5.2 to signal when to return images or videos. |
The Surprising Finding
One surprising aspect of Netomi’s success lies in their focus on “situational awareness” over mere workflow execution. You might assume that automating a business process primarily involves mapping out a workflow. However, the team revealed that in dynamic environments like airlines, context changes by the minute. “That’s why situational awareness matters way more than just workflows, and why a context-led ensemble architecture is essential,” said Mehta. This challenges the common assumption that simply digitizing existing workflows is enough for AI success. Instead, the AI must understand the customer’s current situation to provide relevant assistance. This deep contextual understanding, facilitated by GPT-5.2’s reasoning capabilities, is crucial for handling complex, evolving scenarios.
What Happens Next
Netomi plans to continue extending these patterns into richer multi-step automations, using GPT-4.1 and GPT-5.2. This means the models will not just answer questions. They will also plan tasks, sequence actions, and coordinate backend systems, as the company reports. For example, imagine an AI agent not only helping a customer rebook a flight but also proactively checking for loyalty point adjustments and notifying them of potential upgrades. This level of comprehensive automation could become standard within the next 12-18 months for early adopters.
Businesses should consider how their existing workflows could benefit from AI agents capable of this , multi-step reasoning. Start by identifying complex, high-volume tasks that currently strain your human resources. The industry implications are significant, pointing towards a future where AI handles increasingly operational roles. This will free up human employees for more strategic and empathetic customer interactions. If you’re considering agentic systems, focus on how they can truly understand and adapt to dynamic situations.