Pranav Gujarathi

Pranav Gujarathi

Generative AI & ML Engineer. Building multi-agent AI for digital health at Twin Health. MS Data Science, Indiana University · IIT Roorkee.

Posts by Tags

AI Engineering

Agentic AI in Production: A Delegation Problem, Not a Model Problem

17 minute read

Published:

This post is adapted from a talk I’ve been developing on shipping agentic AI systems at scale — drawing on experience across Walmart, Cigna, and now Twin Health. The thesis: most agent failures aren’t model failures. They’re harness failures.

Agentic AI

Agentic AI in Production: A Delegation Problem, Not a Model Problem

17 minute read

Published:

This post is adapted from a talk I’ve been developing on shipping agentic AI systems at scale — drawing on experience across Walmart, Cigna, and now Twin Health. The thesis: most agent failures aren’t model failures. They’re harness failures.

LLM

Agentic AI in Production: A Delegation Problem, Not a Model Problem

17 minute read

Published:

This post is adapted from a talk I’ve been developing on shipping agentic AI systems at scale — drawing on experience across Walmart, Cigna, and now Twin Health. The thesis: most agent failures aren’t model failures. They’re harness failures.

ML-Ops

Agentic AI in Production: A Delegation Problem, Not a Model Problem

17 minute read

Published:

This post is adapted from a talk I’ve been developing on shipping agentic AI systems at scale — drawing on experience across Walmart, Cigna, and now Twin Health. The thesis: most agent failures aren’t model failures. They’re harness failures.

Production

Agentic AI in Production: A Delegation Problem, Not a Model Problem

17 minute read

Published:

This post is adapted from a talk I’ve been developing on shipping agentic AI systems at scale — drawing on experience across Walmart, Cigna, and now Twin Health. The thesis: most agent failures aren’t model failures. They’re harness failures.