MMXXVI // Post-Agentic Systems
Siddhartha
Chaturvedi.
I build AI systems for organizations. I'm interested in the work around the model: how a team checks its output and decides what it can safely do.
Thinking in Public
Notes on building AI into everyday work, from choosing an interface to deciding when an agent needs human review.
The Enterprise Refactor
For leaders deciding how much work to hand to an agent: where human review belongs, and how to test generated work before customers depend on it.
Read the essayThe App That Builds Itself
Edit a brief on screen, or ask an agent to run an analysis? Compare the options through worked examples, including a generated interface for planning across teams.
Explore the interface demosAI Won't Simplify Work
Doctors need time to think about patients. Researchers need time to pursue a question. This essay asks where AI could take on the paperwork that competes for that attention.
Read the argumentThe Value Function
A worksheet for reviewing AI with your team. Look for convincing language with weak evidence, and answers that sound right but leave the user to finish the work.
Open frameworkExperience behind the writing
The writing comes from shipped systems: edge AI in the field, federal-scale AI programs, and enterprise workflows where a demo is not enough.