Mesa

Musings, notes, scribblings, and rants. Thinking about the future of software development.

Stop Using Claude to Reviewing Cold Code

March 11, 2026 · Peter Rallojay

The most expensive moment to find a bug is after the PR. We built continuous review that runs inside your coding session, on your existing subscription.

Beyond the PR: Shifting Review into the Coding Session

February 26, 2026 · Peter Rallojay & Noah Mitchem

Code review is not a gate you pass through after writing code. It is a constraint that shapes the code as it’s written.

Code Review Was Built for a Slower World

February 20, 2026 · Noah Mitchem & Peter Rallojay

PR-based code review assumed a human author who sticks around. When your author is a 90-second agent session, every assumption breaks.

Agent Blame: A Deep Dive into Line-Level AI Code Attribution

January 26, 2026 · Murali Varadarajan

A deep dive into how Agent Blame attributes code changes to AI agents.

Coding Agents Are a New Infrastructure Primitive

January 26, 2026 · Ben Warren & The Mesa Team

Coding agents are transforming from app -> infra. That changes how you build agentic products.

The Secret Mesa Master Plan (just between you and me)

January 24, 2026 · Ben Warren, Co-Founder & CTO, Mesa

A peek behind the curtain of Mesa's long-term vision.

Why we built our own LLM observability system

January 22, 2026 · Ben Warren

We needed something that Sentry spans couldn't provide.

Using an expensive model made our agent 75% cheaper

January 19, 2026 · Ben Warren

How we reduced code review costs by treating cost like a product metric, building the right measurement, and running a tight loop of experiments.

Raise the Ceiling

January 18, 2026 · Oliver Gilan & The Mesa Team

A manifesto for the next generation of software development.