Posted inBlog
Why Your AI Agent Keeps Starting From Scratch
There is a particular kind of failure that plays out in production agents again and again: the agent does something useful in one run, and by the next run it…
Posted inBlog
Build the Room Before You Write the Memo
For a while, the gap between a useful AI draft and a mediocre one appeared to come down to prompting skill. Add more context, be more specific, break the task…
Posted inBlog
The 70/30 Rule: Why AI Agents Fail in Production
Shaping the flow of ideas at the frontier of cognitive systems. --- A car buyer saved $4,200 by pointing an AI agent at a vehicle purchase while he sat in…
Posted inBlog
The Deck Was Forwarded. The Number Was Wrong. Here Is Why That Keeps Happening.
There's a particular confidence that comes with a polished file. The fonts align, the numbers populate every cell, the slides flow cleanly from problem to solution. We've been trained, over…
Posted inBlog
The next AI platform winner won’t have the best model
For the past couple of years, most of the conversation in AI has been about access — which model can reach more things, use more tools, click more buttons. That…
Posted inBlog
The Build Room: Why Enterprise AI Value Is No Longer in the Model
For most of the past three years, the dominant question in enterprise AI was: which model do you use? Procurement cycles stall on benchmark comparisons, vendor scorecards, and capability matrices.…
Posted inBlog
Six Things Have to Be True Before AI Changes a Workflow
Something shifted in enterprise AI conversations around late 2025. A year earlier, the question was whether a frontier model could handle a given task at all. Now the question is…
Posted inBlog
When AIs Discover Management
Shaping the flow of ideas at the frontier of cognitive systems. --- Sixteen AI agents built a working C compiler in two weeks. The compiler processes 100,000 lines of Rust…
Posted inBlog
The Bottleneck Has Moved: What Your Organization Has to Become Now That the Model Can Do the Work
About a year ago, most AI conversations were still capability debates — does the model understand enough context, can it handle our codebase, will it do more harm than good?…
Posted inBlog
Not All Agent Protocols Are Equal: How to Tell the Foundation From the Frontier
A year ago, the question being asked was still whether agents were ready for production. Now the question has moved on — agents are being deployed, and the new confusion…