How do we take the incredible momentum of autonomous AI agents and scale it into reliable, enterprise-grade production? As businesses rush to deploy agentic workloads, the focus is rapidly shifting from just making them work to making them seamless, efficient, and standardized.
In this Agentic AI Foundation (AAIF) Community Spotlight, Yaron Schneider—CTO of Diagrid Inc. and Chair of the AAIF Workflows Working Group, draws on his experience as a co-creator of Cloud Native Computing Foundation (CNCF) projects like Dapr and KEDA. Yaron shares how building on established open source standards can help developers manage LLM costs, optimize day-to-day operations, and streamline the architecture behind Agentic AI.
Building on the Success of Open Source
During the early growth of the CNCF, the industry naturally prioritized immediate, practical implementations. Everyone wanted tools to deploy and monitor applications right away, which led to an incredibly vibrant and expansive ecosystem of independent projects.
In the era of Agentic AI, Yaron believes we have a golden opportunity to build on those lessons by prioritizing architectural collaboration early on.
“In the era of Agentic AI, standardization is extremely important to make sure that we actually take a lot of our learnings from the previous eras and that we don’t create a whole stack of tools that can go unused or confuse users.” — Yaron Schneider
Instead of building fragmented tools from scratch, the industry benefits most from universal specifications, protocols, and reference architectures. Initiatives like the Model Context Protocol (MCP) are excellent examples of this mindset. By aligning companies around open standards, the AAIF aims to establish clear guidelines that help projects deliver cohesive, long-term value from day one.
Balancing the Operators and the Builders
When building infrastructure, it is vital to support every persona involved in the software lifecycle. While cloud-native infrastructure traditionally focused on DevOps, projects like Dapr succeeded because they deliberately bridged that world with the developer experience.
For Agentic AI to succeed in production, the AAIF is actively balancing two critical personas:
- The Builders (Developers): The engineers who write agent logic and design user interactions day in and day out.
- The Operators (DevOps/Infrastructure): The teams responsible for “Day 2 operations”—monitoring, securing, and maintaining those agents once they are live.
By catering equally to both sides, the AAIF helps organizations confidently transition experimental AI agents into reliable, production-grade environments.
Real-World Value: The Critical Role of Agentic Workflows
To explain the value of agentic workflows to non-technical stakeholders, Yaron uses a simple, relatable analogy: Imagine your morning routine. You wake up, take a shower, brush your teeth, and brew a cup of coffee. If the coffee machine suddenly glitches, you don’t reset your entire morning and restart from the shower. You simply address the machine and move forward.
Right now, many basic AI agents don’t inherently have that capability. If an agent encounters a minor hiccup on step 18 of a 20-step enterprise workflow, it often restarts from step one.
Why Reliable Workflows Matter to Business Leaders
- Optimizing LLM Token Costs: Resetting a complex workflow means paying for the same Large Language Model (LLM) API calls all over again. Ensuring agents can resume where they left off keeps API costs predictable.
- User Experience (UX): End-users and employees need fast, seamless, and predictable execution loops without unnecessary repetition.
- Resiliency: Robust infrastructure design acknowledges that temporary network drops or API timeouts happen. The goal of the AAIF Workflows Working Group is to provide frameworks that let agents gracefully pause, state-save, and recover exactly where they stumbled.
A Pragmatic Take: Agentic AI Built on Proven Foundations
While some perspectives frame agentic architecture as a complete rewrite of computer science, Yaron offers a grounded, reassuring view for developers and enterprise leaders alike.
“My take is, look, it’s not rocket science. It’s a loop in a code that talks to a black box that talks back to you in clear text. That’s essentially what it is… We have the building blocks. There’s not a lot of whole new stuff to invent, really.”
— Yaron Schneider
Rather than abandoning existing modern infrastructure, the industry can apply the robust architectural principles mastered over the last decade to this new paradigm. Success in Agentic AI relies heavily on three core, foundational pillars:
| Pillar | Focus Area |
|---|---|
| Reliability | Ensuring agentic workflows can pause, state-save, and recover gracefully without costly execution resets. |
| Security | Protecting data integrity, managing permissions, and safeguarding the interaction loops with external models. |
| Observability | Providing operators with clear visibility into agent states, decisions, performance metrics, and Day 2 maintenance. |
Get Involved with the AAIF Community
The Agentic AI Foundation is entirely open, public, and built on community collaboration. Whether you are an enterprise strategist looking for architectural blueprints or a developer ready to jump into GitHub issues, there are several ways to get started:
- Stay Informed: Follow the AAIF social pages (LI, X, BlueSky) and subscribe to the community newsletter for highly condensed, curated updates on global AI innovation.
- Read the Curated Content: Dive into The Daily Agentic for quick, highly digestible insights on current trends.
- Contribute Code: Visit the AAIF site and navigate to the repositories to check out open issues, collaborate with maintainers, and help shape open source agent projects.