The Direction of PM in the AI Agent Era — and Why It Overlaps With the Builder Mindset

A Lower Learning Curve Than SaaS — and Faster Adoption

One of the most notable differences is that AI agents don’t require deep onboarding or UI mastery the way SaaS tools do.

Traditional SaaS required:

• learning feature sets,

• memorizing menu structures,

• practicing workflows.

AI agents rely on natural language as the interface.
The barrier to adoption is dramatically lower, and as a result, experimentation and deployment cycles are becoming faster than ever.

For PMs, this means a fundamental reshaping of how we test, ship, and validate.

A World Where “You Only Know After You Try”

While working at Vertical Bar on Text2SQL research, I saw this firsthand.
The model hallucinated — even creating columns that never existed in the schema.

It was a clear reminder:
AI agents operate in an inherently uncertain, nondeterministic environment.

You might see:

• success 9 out of 10 times but one unexplained failure,

• unexpected reasoning paths,

• context misinterpretations,

• or odd creative leaps.

This uncertainty changes the PM skill set entirely.

The New PM Skill Set: Becoming a Builder

These are the three capabilities I’ve found most crucial.

1) Embracing Uncertainty

AI doesn’t behave with strict determinism.
The ability to accept unexpected results, investigate them, and move on quickly is essential.

2) Rapid Prototyping

A simple UI prototype often communicates ideas better than long explanations.
Fast iterations — build, show, scrap, repeat — become a core competency.

3) Technical Curiosity

You don’t need to master every low-level detail.
But you do need the instinct to explore:
“How does this pattern behave? Why does the model respond this way?”
This is exactly where the PM role overlaps with the Builder mindset.

My Current Focus Areas

Recently, I’ve been diving deeper into two domains:

• AI-driven content generation

• Data-driven growth using AI

I’m especially interested in how AI will evolve from a rewriting tool to a strategy-aware growth operator.
Soon, content planning, execution, and performance learning may become fully autonomous loops.

Learning From Global Standards — The Power of Speed

Watching the U.S. and European ecosystem closely, one element stands out:

Speed determines survival.

Product iteration cycles are shorter.
Ambiguity tolerance is higher.
Teams are more willing to ship imperfect versions and learn from real data.

Experiencing this shaped my own conviction:

I need to build faster, take more risks, and move with less hesitation.

This is the Builder mindset in action.

Conclusion — The AI Agent Era Has Just Begun

Whether you're a PM, a founder, or a builder, one truth remains:

The AI agent era is still in its early days.
And the people who jump in now will gain the deepest learning.

The winners in this new cycle will be:

• curious,

• fast-moving,

• comfortable with uncertainty,

• and eager to uncover new behavior patterns in AI systems.

I’m choosing to push harder into this shift — to build faster, learn deeply, and move forward without fear.