2026: Enterprise AI Evolves to Scalable Agentic Collaboration

Key Quotes

"the more you can align with the sort of company's general long term perspective about where powerful ai will come from like the smoother things will go because anthropic is nothing but focused right and i think that that's shown through in sort of the you know the things that we choose to make versus not and definitely there's this belief that you know for very powerful ai you need uh the ability of the model to sort of reason about things to plan agentically and and work for a long term horizon but and also to be able to write and run code"

Mike Krieger explains that Anthropic's product development is guided by a long-term vision for powerful AI, which includes reasoning, planning, and the ability to write and run code. This focus allows for smoother product development by aligning with the company's core beliefs about AI capabilities.

"going into the sort of the last couple weeks of the year last year we had built something internally we called claude cli which we later released as claude code and that was the emergence of that came from our labs team which is a team that really focuses on trying to do sort of disruptive zero to one ideas and that was everything from like early compute user explorations and some wacky things and also this claude cli thing and in between i think september when the first version got sort of rolled out internally to december it rapidly overtook every other sort of coding tool that we had internally"

Krieger describes the internal development of Claude CLI, later released as Claude Code, which originated from Anthropic's labs team focused on disruptive ideas. This tool quickly became the most used coding tool internally, indicating its significant potential and user adoption even before its public release.

"we have product principles inside anthropic and one of them is right the exponential which is like we're trying to build products that both you know meet the moment so they're they're useful today or at least they poke at something useful today maybe the ones that are a little earlier we won't release yet but that they can naturally improve and it's been interesting even on the claude code side we've deleted parts of the harness over time rather than added to it because the model can do more"

Krieger highlights Anthropic's product principle of "writing the exponential," which aims to create products useful today that can naturally improve over time. This approach is demonstrated by the evolution of Claude Code, where parts of its supporting structure have been removed as the underlying AI model's capabilities have increased.

"yeah so this is one of those questions that i'm most interested to see in coming year but even the coming years is what what it takes to kind of rewire people with these new tool set it's like this this whole language this whole infrastructure that they have access to especially if they that they haven't before especially if they're they're not developers do you think that some of this you know if on the spectrum from this early kind of usage of claude code for non coding use tinkers who are kind of you know more technical than they let on on one end of the spectrum versus actually kind of heralding a different set of interaction patterns that people are going to have how how do you see that evolving"

Krieger expresses interest in how people will adapt to new AI toolsets and infrastructure, particularly non-developers. He questions whether this will lead to a new set of interaction patterns, moving beyond early tinkerers to a broader adoption of AI for various tasks.

"i think for for a lot of enterprises that we talked to there's still this gap between sort of the idealized like great if you ran this perfectly on this like one cloud with all you know your you know permissions all perfectly set and you were okay with inference happening in this way then we could unlock use cases tomorrow and the reality which is there's legacy systems there's often sort of regulatory reasons why they you know for example will only run in this particular way on aws in this particular kind of setup and so a lot of the work that we are doing for next year is the word we've been using is distributability which i think the spell checker tells me is not really a word but what we really mean is uh if we want to bring our intelligence and our even our agentic primitives you know whether it's skills whether it's the agent sdk whether it's storage whether it's memory all of these pieces into actual enterprise workloads we need to really actually embed and meet them where they are"

Krieger identifies a significant gap for enterprises between idealized AI deployment scenarios and the reality of legacy systems and regulatory constraints. He explains that Anthropic's focus for the upcoming year is on "distributability," meaning embedding their AI intelligence and agentic primitives directly into existing enterprise workloads and infrastructure, meeting them where they are.

"i think that that probably more than anything is what will define the year is you start seeing this already with coding so -- we did this get up partnership with their agent hq piece where now you tag claude in a pull request and then you go have your coffee and you come back and it's done whatever you needed to do and we did the same integration with cloud code that's sort of pointing at the kind of interaction that you might expect now is it already going to be at the place where it can onboard onto the organization understand the understand the problem space understand the sort of dynamics of all the relationships and just pick up work no i don't think we're going to be there maybe near the end of the year we'll have some kind of early glimmers of there but i do think the the sort of more like piece of the job function that has like a a clean sort of you know all right great we need -- we need to prepare this kind of report here's the work i've done already here's where you can go get more information here's what good looks like report back to me you know in a way that you might delegate to somebody else that's very much around the corner"

Krieger suggests that the ability of AI to take on larger workloads, moving beyond tools to colleagues, will define the coming year, citing coding as an early example. He anticipates that while AI may not fully onboard and understand complex organizational dynamics by year-end, it will become capable of handling more defined tasks, similar to delegating work to a human assistant.