The Agent that doesn't need a rewrite
The agent that survives its first year in production is usually a thin layer over the system you already have. Your ten-year-old codebase is not the obstacle. It is the asset.
Notes on Systems
Writing on systems, AI, data, and software for teams trying to reduce operational drag without losing clarity.
Latest posts
Evals Are the Tests You Should Have Written First
Renaming tests to evals did not create a new discipline. It created permission to skip an old one - the obligation to define what correct means before shipping.
Why Agents Get Canceled
An agent is a production system that happens to be probabilistic, and it has to be operated like one - the surviving minority are not the teams with a better model, they are the ones operating better.
What I Optimize for Now
What I optimize for has shifted from capability, elegance, and speed toward concreteness, reversibility, and earlier clarity — the system that will exist after me, not just the one I am building.
Systems That Surprised You
The most useful audits of long-lived software come from the moments your assumptions about durability turned out to be wrong.
Decisions That Aged Badly
Some decisions were wrong. Others were defensible at the time and aged badly. The discipline is keeping the categories separate, and making it cheaper to be wrong.
Stewardship Begins Where Ownership Ends
Ownership makes responsibility visible in the present. Stewardship makes systems survivable across time.
What You Refuse to Do Is Architecture
A system without explicit constraints is not flexible. It is undecided. The refusals you write down are the load-bearing structure.
AI in Long-Lived Systems
Every AI integration is a bet that you can monitor something you do not fully control.
Tooling and Accidental Complexity
No one decides to make a system heavy. It happens one reasonable choice at a time.
Abstractions That Age Poorly
Good abstractions compress complexity. Bad abstractions relocate it.
Refactoring as a Signal
The code that needs refactoring most often is rarely the worst code. It is the code sitting on top of a structural assumption that no longer holds.
Defaults Nobody Revisits
Best practices are useful until they become borrowed conclusions that no longer match the system’s actual constraints.
The Carrying Cost of Flexibility
Speculative flexibility often looks responsible at first, then quietly raises the cost of understanding, testing, and changing the system.
Time As A Design Constraint
Systems age whether we plan for it or not, and the best design decisions are often the ones that remain legible long after their original context is gone.
When Code Is No Longer Scarce
As code generation gets cheaper, the real bottleneck shifts from production to coordination, comprehension, and judgment.