Memos
Our thinking on AI, each piece tied to a decision a leader has to make.
As of Q4 2026
The big trends we’re tracking and researching right now.
Falling frontier prices and open-weight models closing the gap, and what that does to the order of commitments.
Emerging standards for how personal and enterprise agents find, trust and transact with each other.
Not individual launches, but how fast models and personal agents are improving, and in which directions.
How governments think about the pace of development, what transparency labs owe and to whom, whether antitrust reaches companies poised to do everything, and how accountability works when AI decides on people’s behalf.
Today’s AI infrastructure is built for one person working with an agent against one system. The next layer, where my agents work with yours across companies, has a common language but no settled way to establish who an agent is, what it may agree to, or what happens when two disagree, and no neutral party owns it yet. Leaders should prepare for their agents to deal with other people’s, and investors should watch who earns the trust to broker it.
Every week the frontier makes a new case for waiting: why build or connect anything of your own if your customer’s agent will soon do the work? But people, markets and regulators move far slower than the models do. Commit first to what pays for itself now, plan in parallel for the world where it matters less, and get the data ready for both.
Stories from working with agents, on the R5 Substack.
An AI assistant emails a prospect on its own, and what that taught us about explicit limits.
An agent slows to a crawl, throws no errors and can’t diagnose itself.