Teamday
← People

CJ Hess

Special Projects Lead at Tenex

agentsenterpriseai-engineeringfuture-of-worktenex

About CJ Hess

CJ Hess is the special projects lead at Tenex, the New York applied AI partner co-founded by Alex Lieberman. He was Tenex's first hire, joining before the company had settled on its name. A software engineer by trade, he previously ran his own engineering consultancy and moved into AI consulting there.

He says hourly billing pushed him toward Tenex: as AI made his output better and faster, charging by the hour meant earning less. Tenex's model of paying engineers for output rather than time matched his bet that leaning into AI would raise his earning capacity.

His special projects work covers everything outside client engagements. That includes Tenex's internal "meta harness" and the platform Tenex uses to deploy background agents into enterprises.

Career Highlights

  • First hire at Tenex, now special projects lead
  • Founded an engineering consulting company before Tenex
  • Builds Tenex's internal meta harness for agent-driven engineering work

Notable Positions

On the Meta Harness

Hess describes Tenex's meta harness as a standardized set of skills, a structured documentation layout inside each repo, and a CLI that models use to drive work: plans, visualizations and client readouts. Every project has a canonical plan document, and agents log their decisions and progress in a set way. Everything is checked into Git, so the reasons behind the code are on record and not just in someone's head. He says Tenex has loosened the required structure as models have improved. See Agent Harness.

On Measuring Engineer Output

Tenex tracks average story points per engineer per month. Hess estimates output has roughly doubled this year, starting with the Opus 4.5 release in December, and says the team recalibrates its story-point criteria at least once a quarter.

On Automating Enterprise Work

His rule for agent projects: you can't automate a process that does not exist. When many people do the same job in slightly different ways, an agent can't know whether it's doing the job right and there's nothing to build evals against. He recommends standardizing the process first, then handing it to a background agent whose work people review.

On Jobs

Hess believes AI will change and displace jobs but sees it mostly as reallocation: when an agent takes over a function, the same subject-matter experts can move on to the next problem, which in large companies keeps coming.

Key Quotes

  • "You can't automate a process that does not exist." (on deploying agents in enterprises)
  • "It's never like in someone's head. It's always in the repo." (on keeping engineering context in Git)

Related People

alex lieberman