Weave Intelligence ads

61% of platform teams made FinOps a top priority for 2027. They now shape which cost tools get bought, and they want different things than finance does. Is your company positioned for them? Download the one-pager to see how platform teams group FinOps tools, what they expect from each category, and how we help vendors position to them.

Is your logo on this map? Our FinOps reference architecture sorts the market into five categories, from systems of record to AI cost specialists, and names the vendors in each. Platform teams read it before they build a shortlist. Download the one-pager to see where you sit, who you are grouped with, and how we help vendors position to platform teams.

So what is an AI Platform Team? A question I get a lot, let's dissect this a bit! It is the central team that owns an enterprise's agent infrastructure: the engine room of any agentic platform. It encompasses the models (accessed, routed, and, where the enterprise hosts its own, served), alongside the agent harness and agent governance. The team builds and runs that infrastructure as a product, so every vertical can build and operate its own agents on one governed substrate. How these teams are being formed, what their jobs to be done are and where you get started: I summarize all this in my latest article on Weave Intelligence, links in the comments.

Your platform team just got renamed. Again. Welcome to life as an AI Platform Team! But this one isn't just a new title on the org chart. It comes with real work: → Managing agents at scale + all the adjacent security concerns → A whole new set of users: beyond developers into any vertical of the org I sat down with Sam to dig into what actually changes. The full breakdown is in my latest article on AI Platform Teams, link in the comments. Honest question: is your team being called an AI Platform Team yet? Real shift, or just a new label?

I feel like the best way to approach "agents" is to think about them like another short lived service. Almost like a lineage from VMs, over dockerized workloads, functions as a service and here we go: agents! That demystifies the whole thing a lot. There's something running in a sandbox. Has env vars, can call out to a bunch of things, has governance, gets tested etc. Sure, some things differ. More egress focus, less ingress and a different way of testing but all in all? Very similar. Standard software architecture practices apply!

Interesting question at the moment: how much of the Agentic Platform will be PaaS, and how much will we build in house? The answer will likely make you want to become an AI Platform Engineer. I'm deeply convinced that every single enterprise will run an agentic platform within the next 24 months. This platform is the curated environment that agents run on. From sandbox management to context serving, from evals to observability. But how much of this will come "out of the box" from a unified provider, and how much will enterprise AI Platform Teams need to build themselves? I feel like the history of platform engineering can teach us some valuable lessons here: - Since the early days of Heroku, enterprises could have "just used a PaaS". They didn't. In fact, so few did that Salesforce really struggled with the product. Enterprises kept building Internal Developer Platforms instead; over 80% now have one. - You could argue that lock-in was already a problem for developer platforms. Today it's a nightmare if your platform is single-model or dependent on specific vendors. The models are specializing, and it's entirely unclear which ones will win the race. Staying model-agnostic is simply a must. - Platforms may be one of the last remaining differentiators for the business. Enterprises will go to great lengths to differentiate and build their own. - Enterprises are very different animals from SMBs. They require an insane depth of integration. We've never really seen a PaaS provider hit a nerve with the enterprise. So, long story short: while many components of an Agentic Platform will be available as tools, building the final platform will be an incredibly interesting challenge. What a time to be an AI Platform Engineer!