LLenny's Podcast
← All frameworks
LeadershipTomer Cohen (LinkedIn CPO)

The Five Traits to Keep, Automate the Rest

Protect where humans shine — vision, empathy, communication, creativity, judgment — and automate everything else

Difficulty
Moderate
Time to result
~weeks to results
Steps
2
Confidence
92%

A decision rule for what to keep human versus what to hand to AI across the product development lifecycle. Builders spend their time on five high-leverage traits; every other sub-step is a candidate for automation. Judgment ('taste-making') is the single most important trait.

Origin

Articulated by Tomer Cohen (LinkedIn CPO) as the organizing principle behind the full-stack builder model's automation strategy.

Core principles

  • 01Great builders should spend time only where humans clearly outperform AI
  • 02Judgment — high-quality decisions in ambiguous situations — is the most important trait
  • 03AI is not yet great at next-level creativity; it surfaces the known, not the beyond-obvious
  • 04Everything outside the five traits should be automated 'really, really hard'

How to run it

  1. 1

    Name the five keep-traits

    Define the human-owned zone: vision (a compelling stance about the future), empathy (profound understanding of an unmet need), communication (rallying others around an idea), creativity (possibilities beyond the obvious), and judgment (high-quality decisions in ambiguous situations).

    Pro tip Weight judgment above all — if forced to pick one trait that defines a builder, it's taste-making ability.

  2. 2

    Audit every other sub-step as an automation candidate

    Take each remaining sub-step of the lifecycle — data pulls, spec review, build fixes, QA, analytics queries — and target it for an agent. Assume it can and should be automated unless it requires one of the five traits.

    Pro tip Coding is already the most-accelerated zone; look for untapped gains earlier in the 'idea to design' phase.

    Watch out Don't automate creativity yet — humans still lead on the beyond-obvious; AI mostly recombines what you already know.

In the wild

Maintenance and QA agents doing failed builds

LinkedIn built a maintenance agent and QA agent for the 'code to launch' phase; when a build fails the agent fixes it instead of an engineer hopping on the issue.

Close to 50% of failed builds are handled by the maintenance agent, freeing engineers to 'finish your coffee' instead of redoing builds.

Common mistakes

Trying to automate judgment and creativity too early

AI is not yet great at next-level creativity or high-quality judgment in ambiguous situations; handing these to agents produces weak, derivative decisions rather than genuine breakthroughs.

Over-investing in the coding phase while ignoring idea-to-design

Coding was already accelerated, so the highest untapped quality gains sat in the neglected idea-to-design phase — the source of much of the eventual output quality.

Is it for you?

Best for

Product leaders deciding which parts of their workflow to hand to agents versus protect as human craft

Not ideal for

Teams treating AI as a full replacement for judgment or creative direction

From the transcript

the key trait that I'm emphasizing for for builders is where I want them to spend their time

12:30

ultimately what I think is the most important trait for a builder is judgment. That's, you know, some people call it test making but it's…

13:30

I don't think AI yet is great at creativity

13:00

I think we're close to 50% of all those builds being done by the maintenance agent and a QA agent

27:00

From the episode

Why LinkedIn is turning PMs into AI-powered "full stack builders”

Tomer Cohen (LinkedIn CPO)