You better move quickly

If you want the agentic runtime your team is developing to stay at parity with those offered by AI companies, you'll need to move quickly.

Always-on, user-dedicated autonomous agents (Spark, Muse, Dots, Grok Bot, etc.), dynamic work analysis with model autorouting (monomodal, multimodal, omnimodal; decisioning, etc.), cost management (frontier, flagship, mid-tier, lightweight, open weight/source, etc.), data (ontologies, semantics, taxonomies, etc.), DBs, (relational, vector, graph, KV, multimodal, streaming, etc.), governance (RBAC, ABAC, logging, auditing, etc.), federated learning, output consistency, and more, more, more.

Note that I’m not including any of the DevOps, SRE, or IT infrastructure (compute, storage, network, etc.) and teams required to run these systems, or the legal, compliance, and privacy liability you take on by building, running, and owning these runtimes.

And we're seeing the emergence of new user experiences, shifting from static text generation to dynamic interactive environments: OpenAI's GPT-6 Intelligent UI, Google Labs' Playground, and Microsoft, with Satya Nadella describing Windows PCs as an "infinite software factory." That's not a new interface for an app; it's the app itself, generated on demand.

I've gotten feedback on my posts about this topic, and it's worth addressing. I'm not saying: "Do not build an agentic runtime." I'm saying: "Understand the business problem you're trying to solve, evaluate TCO, including the ongoing run costs (TOKENS!), full-boat people costs, liability, and the opportunity cost of building it given the pace of innovation across the AI landscape, and then make a decision."