Google has handed over its Agent Substrate initiative to the Cloud Native Computing Foundation (CNCF), marking a key step in fortifying AI systems. Originally built to separate stateful agent sessions and identities from live computing resources, the technology captures inactive agents as snapshots—allowing them to reboot in under 500 milliseconds upon demand.
This is targeted at managing the idle state of those agents so that they do not burn expensive compute resources, a challenge in a Kubernetes-managed environment that might be tasked with managing thousands of agent-infused pods. Drew Bradstock, VP of product management, orchestration, and Kubernetes at Google, explained the concept in a blog post, comparing it to a hotel room rather than a house.
Agent Substrate and OpenShell
Agent Substrate can be partnered with OpenShell to provide a compelling cloud-native package for running and securing AI agents. Jonathan Bryce, CNCF Executive Director, noted that these projects are directly oriented around taking powerful and non-deterministic compute agents and figuring out how to contain them and give them the access they need to make progress.
Bryce added that this challenge was exemplified by a recent CNCF survey that found while coding agents were extremely widely adopted, the number of people who have an agent in their personal life that is doing anything significant is much smaller, and that is because people are concerned about safety and trust. Bryce stated that for most people, the answer to whether they want to give agents the keys to anything is definitely no.
Securing AI Infrastructure
Agent Substrate helps counter the agentic AI efficiency lull that has happened over the past year. The project has garnered interest from more than 100 contributors since its release by Google into public development in May. It has also been injected into the CNCF kagent AI programming framework project, and Google has unveiled Agent Substrate 1.0 as part of its managed Kubernetes Engine (GKE) for public use.
The donation of Agent Substrate to the CNCF “Sandbox” program is the entry point for projects at that open-source organization. Bryce noted that the project is significant and gives a glimpse of where AI infrastructure is developing over the next six to 12 months.
Evolution of Open-Source Projects
The projects also highlight the rapidly changing evolution timeline for open-source projects. Bryce said that one of the things he is most excited to see is how people start to put these together and get a stack that kind of gels. Because that’s when they, historically in open source, have seen the greatest advancement and the greatest adoption, which is when they get those patterns that work.
With 82% of enterprises running Kubernetes right now, and things like Open Telemetry and the set of commonly deployed cloud-native projects, that gives a huge advantage if an enterprise is one of those who has adopted that stack, because now they can plug OpenShell in and start experimenting with agents in a way where it’s using the same security policies, network policies, telemetry that they’ve already set up and that they know is well baked and well defined.
Leave a Reply