When agents write the code, judgment should outrank titles
Career ladders reward scope because writing good code at volume used to be scarce. Agents moved the scarcity to judgment, and hiring, leveling, the ladder and pay all need to follow.
Written by
-
Benedikt Stemmildt Founder & Co-CEO
Published: 11 September 2026 Updated: 2 min read
A team lead told me something that I have not stopped thinking about. His strongest contributor that quarter was, on paper, his most junior engineer. Two years of experience. She was not writing the most code. She was the best in the team at one thing: deciding what the agents should build and catching where they were wrong.
His problem was that the career ladder had no row for that.
The ladder rewards what used to be scarce
The ladder most engineering orgs run was built for a specific scarcity. Writing good code, at volume, was hard and slow. So we leveled people on it. More scope, more output, more people managed, higher title. The ladder is a machine for rewarding whatever was scarce, and for two decades the scarce thing was the ability to produce correct code reliably.
Agents moved the scarcity to judgment
Agents changed what is scarce. Producing code is no longer the hard part. Deciding what is worth building, judging whether the output is right, knowing what must not break: that is the hard part now. And the ladder does not measure it well, because the ladder was built before it mattered.
So you get the mismatch the team lead hit. The person creating the most value is not the person the ladder ranks highest. Title and contribution came apart.
Four places that need rework
This shows up in four places at once, and each one needs rework:
- Hiring. Testing how fast someone writes a function tests the thing agents do now. The signal worth hiring on is judgment under ambiguity: can they tell a good plan from a plausible one, and say why.
- Leveling. Scope stops working as a proxy for seniority once a two-year engineer can direct a factory’s worth of output. Level on decision quality, not on how much work someone supervises.
- The ladder itself. A rung defined as writing more code and managing more people rewards a skill that is getting cheaper. Redefine the rungs around what does not compress: taste, judgment, holding a system in your head.
- Pay. Compensation bands are bolted to ladder level, so they inherit its mistake. When level and value come apart, your best performer is not your best-paid one, and that gap is where a competitor recruits.
Experience still matters, for the judgment it produced
I want to be careful here. This is not “experience stopped mattering.” A lot of judgment is earned by twenty years of seeing systems fail. The point is narrower: experience matters because of the judgment it produced, not because of the title it earned along the way. When those two come apart, follow the judgment.
The candid version of the talent shift is uncomfortable for everyone. It means some senior people are leveled for a skill that is now partly automated, and some junior people are doing the work that actually moves the business. The org that says that out loud and fixes its ladder gets the talent. The org that keeps promoting on scope keeps losing its best people to the one that does not.
If you rebuilt your career ladder this quarter, what would the top rung actually reward?
Further reading
Written by
-
Benedikt Stemmildt · Founder & Co-CEO
More on Organization and roles
- Agents finished the portal, then it waited in a queue An agent-built portal was ready to ship and waited on a gateway, a firewall rule, a certificate and QA. A factory shortens the build, and only a self-service platform shortens the wait around it.
- Double throughput needs a mandate, not another pilot A pilot asks a team to prove agentic engineering works here. A mandate states the outcome and makes the boring blockers somebody's problem, which puts the uncertainty on the organization.
- Coding agents are an organizational-design problem Cheaper local generation pushes the coding-agent constraint into team practice, the shared platform, governance, and learning. One practitioner's four-level model of that shift, held up against what DORA can and cannot support.
Test the constraint in your system
The Factory Readiness Assessment connects this research to your value stream, your data, and the next decision.
Training on this topic: Agentic Engineering for Engineering and Product LeadersWant to check one team first? The self-check: twelve yes or no questions, three minutes, instant result, no account. Take the self-check