14+
years building & selling software
AI-Native
engineering:
the machine does
the work. I judge it.
I'm Alpesh Nakrani, Chief Revenue Officer at Devlyn. I help companies go AI-Native: advising on which AI bets are real, building delivery pods that ship, and teaching teams the operating model behind it. Fourteen years building and selling software, ten as CTO and COO. I also write about all of it, 33 books and counting.
33
books on AI-Native work
10
years as CTO & COO
CRO
now, at Devlyn
The thesis
AI-Native is not AI-assisted. The machine does the whole job; the human's role narrows to one thing: judgment. Specifying intent, evaluating output, owning the call. Everything I write follows from that single shift.
01 / How I help
Work with me
01 Advise Whether your AI bets are real, and which to kill. A blunt, instrumented answer, fast.
02 Build AI-Native delivery with Devlyn: pods that ship, with evaluation built in from day one. 03 Teach Courses and workshops that turn the thesis into your team's operating model.
04 Speak Talks and working sessions on AI-Native engineering and the judgment economy.
The book · 2026
AI-Native
How the machine took the work and left us the judgment
For three years we called it AI-assisted and meant autocomplete with ambition. AI-Native is the harder claim: the model does the job, and the only thing left for us is judgment. This book draws the line precisely: where the machine is now doing real work, where it is not, and what an organization has to rebuild once the bottleneck moves from production to evaluation.
02 / From the library
Selected books
03 / Coverage
What I write about
5 titles AI-Native When the machine does the work and the human evaluates. 5 titles Evals Measuring what will actually break in production. 3 titles RAG Retrieval that survives a growing corpus. 3 titles Agents The narrow band where autonomy earns its keep. 4 titles Inference The economics of shipping models you can run. 4 titles Leadership Teams and org design after automation.
04 / Blog
Latest writing
Blog · 20 min
LLM Evaluation: Measuring What Will Break
LLM evaluation is the harness that gates a real deploy. Learn what to measure, which metrics lie, when to trust an LLM judge, and who should own it.
EvalsEngineering
Blog · 10 min
How to Build an LLM Evaluation Framework
A good LLM evaluation framework tests what will break in production: a golden set from real traffic, task metrics, blinded rubrics, and a drift cadence.
EvalsEngineering
Blog · 10 min
LLM Evaluation Metrics That Matter (and the Ones That Lie)
The LLM evaluation metrics that matter measure what breaks in production. The ones that lie measure what looks good in a deck. Here is how to tell them apart.
Evals
05 / Praise
