MMorteza Naraghi
AI systems / software delivery / accountability

Build the system around the model.

I’m Morteza Naraghi, a co-founder and CTO building production AI systems—and writing about the architecture, economics, and operating discipline they require.

Read the book Explore field notes
A technical publishing home

Three connected bodies of work.

The book is the argument. Meros is the operating environment. The field notes are where the lessons get tested in public.

01 / THE BOOK

Agentic Software Delivery

How AI engineering can transform the software development lifecycle by preserving context, judgment, and accountability.

Read the thesis →
02 / THE WORK

Meros AI

Building and operating agentic systems inside specialty medical practices, where reliability has to be observable and accountable.

Visit Meros AI →
03 / THE NOTES

Production AI, in public

Architecture, economics, evaluation, security, and the decisions that separate a useful system from a convincing demo.

Read the field notes →
The questions I keep returning to

AI becomes real when the surrounding system can carry the responsibility.

Models matter. So do context, tools, policy, evaluation, observability, and the boundaries around human judgment.

01

What should an agent be allowed to do?

Make permissions explicit, stage actions, and design approval around consequence—not novelty.

02

How do we preserve intent across delivery?

Reduce the translation loss between discovery, requirements, architecture, code, tests, and production learning.

03

How do we know a system is getting better?

Measure outcomes and failure modes, not only latency, token count, or the number of generated artifacts.

04

What changes when people depend on the result?

Move from demo behavior to evidence trails, recovery paths, secure integrations, and accountable operations.

Cover image for Agentic Software Delivery by Morteza Naraghi
AGENTIC SOFTWARE DELIVERYHow AI
Engineering
Can Transform
the SDLCBy Morteza Naraghi
The book

AI is making code cheaper. That does not necessarily make software delivery faster.

Agentic Software Delivery is a practical guide to using AI agents across discovery, requirements, architecture, implementation, review, testing, release engineering, operations, incidents, and documentation—without turning the delivery process into an uncontrolled experiment.

Its central question is simple: what should the software-development system look like when AI becomes an active participant in it?

Explore the book →
Latest field notes

From the boundary between model and system.

Long-form essays grounded in operating questions, architecture diagrams, and the failures that teach more than the demos.