Agent-Native Engineering

By: Andrew Pignanelli (Cofounder/CEO, The General Intelligence Company) Source: General Intelligence Company Type: article

Summary

Fill after reading. 2-3 sentences: what is this about, what’s the core argument or insight?

Key Ideas Extracted

Fill during processing. Each idea links to a knowledge entry.

Notes

Your annotations, reactions, questions, disagreements. Written during or after reading.

Raw Content

Agent native engineering means restructuring your organization around agents as individual-contributors instead of engineers.

Background: Background Agents Changed Everything

The shift started with Devin from Cognition AI (summer 2024) — first asynchronous coding agent. Assign a task, walk away, come back to new code in a PR. What mattered wasn’t Devin’s coding performance — it was the pattern of work it enabled: engineers operating two parallel work streams.

Background Agents Engineering Workflow

Someone has a new feature idea scoped narrow enough
A Linear ticket is created
Ticket assigned to a person AND coding agent (someone still responsible)
Coding agent works in background, makes a PR
Agent iterates on CI/tests until they pass, checks for merge conflicts
Engineer reviews code, sometimes running branch locally
[Optional] Engineer leaves PR/ticket comments
Agent addresses feedback, makes new commits
Engineer approves and merges

Three Task Levels

Simple (one-shottable): Well-written prompt gets coded and merged easily. “Change the corner radius on this button.”
Easily Manageable: Background agent with some feedback cycles can solve.
Complex: Managed directly by engineer using synchronous coding agents.

Every task except complex should be delegated. Agents become primary ICs, managed by engineers. Engineers become managers, managers become staff engineers.

Agency Goes Up

Give engineers larger features with less specs. Give managers larger objectives. Stop requiring a second engineer to review every PR.

Remove Two-Engineer Code Review

“A two engineer code review totally ruins this dynamic.” If your engineer manages 5 simple, 2 manageable, 1 complex task — two-engineer review requires reviewing 10 simple, 4 manageable, 2 complex tasks. Alternatives: agentic pentesting companies, code review agents (Cursor Bugbot, Greptile), AI SRE on staging.

“In this new world you will live and die by the number of iterations you can complete in a given amount of time.”

Code Slop Prevention

Robust rulesets (claude.md, .cursorrules) — “Agents are good at reading rules”
Tests as reward signal — agent iterates until tests pass
Linters (eslint, python black) for consistent styling
Code review bots (Cursor Bugbot)

How Agent-Native Engineering Affects an Org

Engineers manage more, code less
Task decomposition time also decreasing
Ticketing system management becomes larger % of workflow
“Everyone’s a PM and team lead for whatever project they’re working on”
Products change rapidly; roadmaps >few weeks become ineffective
Tight collaboration cycles, “actions-per-minute” mentality
New problem: idea generation, not execution

You Need More Designers and Product People

Engineer:designer ratio needs to change. “If your ratio was 1:20 like is fairly common at these orgs, it’s now too low.” Features materialize at speed of thought — need more people who can reason about experience, product, and design.

Token Spend

$4,000/engineer/month on Opus tokens (January 2026). Engineers shipped average 20 PRs/day, sometimes hundreds of commits daily. “We are spending roughly 20% more on engineering intelligence to increase output by a factor of 3-4x.”

The Future

IDEs as we know them will go away
Code review by humans will go away
All engineers become product managers, all PMs become engineers
Highly capable background agent teams with everything delegated first
Advanced plan mode UIs (akin to PRDs)
Beautiful frontend code largely automated
All infrastructure becomes code
Teams become smaller and more capable
Designers matter more, non-product-thinking engineers matter less