A small team can now build in days what previously required months.
Prototypes are cheaper.
Software is faster to ship.
Design is easier to iterate.
Technical leverage is increasingly available to founders who do not have large engineering teams.
That is an extraordinary change.
But it creates a problem for founders and investors.
If everyone can build faster, building fast becomes less differentiated.
The advantage moves somewhere else.
We think that advantage is learning velocity.
Not how quickly you can write code.
Not how many features you can ship.
Not how polished your first version looks.
How quickly you can move from:
assumption → experiment → signal → decision → iteration.
That may become one of the defining characteristics of the next generation of exceptional seed founders.
the MVP is becoming a weaker signal
For years, having a working product was one of the strongest early startup signals.
That made sense.
Building software required time, technical expertise and resources.
A founder who had managed to get something into the hands of customers had already demonstrated a meaningful amount of execution.
AI changes the equation.
A capable founder can now go from idea to prototype dramatically faster.
That is good for entrepreneurship.
It also means investors need to interpret the MVP differently.
A working product tells us less than it used to.
The more interesting question becomes:
what did the founder learn while building it?
Did they discover a painful customer problem?
Did customers use the product differently than expected?
Did the founder change the product because of those observations?
Did they find a narrow group of users who care deeply?
Did distribution become easier or harder?
Did the founder kill an assumption quickly?
These are harder signals to manufacture.
And they tell us much more about the future company.
learning velocity
We think of learning velocity as:
the rate at which a founder converts uncertainty into useful information.
Every early-stage startup is built on assumptions.
Who is the customer?
What problem matters?
How painful is it?
Who pays?
Why now?
Why this product?
Why this founder?
Why will customers discover it?
Why will they stay?
At pre-seed, most of these questions are unanswered.
That is normal.
The best founders aren’t necessarily the ones who already know every answer.
They are the ones who can discover the answers faster than everyone else.
That is learning velocity.
A founder says:
“We believe this customer has this problem.”
Then they test it.
The market disagrees.
They don’t spend three months defending the original thesis.
They update it.
Then they test again.
That loop is the company.
speed without learning is just motion
Startup culture has developed an obsession with speed.
Ship faster.
Launch faster.
Raise faster.
Hire faster.
Grow faster.
Speed matters.
But speed by itself is not the goal.
You can move very quickly in the wrong direction.
A founder can ship ten features nobody wants.
They can acquire thousands of users who never become customers.
They can raise capital before understanding the market.
They can build an impressive product around a weak problem.
They can spend six months optimizing something that should have been killed in week two.
That is not high velocity.
It is high activity.
The distinction matters.
Execution velocity asks: how quickly can we do things?
Learning velocity asks: how quickly can we become less wrong?
The second question is more important at the beginning.
what high-learning-velocity founders do differently
There are a few patterns we increasingly care about.
1. they get close to reality quickly
They don’t spend months debating hypothetical customers.
They talk to customers.
They watch workflows.
They sell before everything is perfect.
They observe what people actually do rather than what people say they might do.
The closer the founder is to reality, the faster the learning loop becomes.
2. they run small experiments
The best early-stage founders don’t always make huge bets.
They make cheap bets.
Can we get five customers?
Can we get one person to pay?
Can we solve this workflow manually?
Can we get someone to use the product twice?
Can we find ten people with the same problem?
Each experiment reduces uncertainty.
The objective isn’t to prove the startup works.
The objective is to discover what needs to be true for it to work.
3. they change their mind
This is underrated.
Some founders interpret changing direction as weakness.
We often see the opposite.
A founder who changes their mind because reality changed is demonstrating something valuable.
Conviction is important.
But conviction without updating becomes stubbornness.
The strongest founders tend to have a strange combination:
strong ambition + weak attachment to their initial assumptions.
They know where they want to go.
They are flexible about how they get there.
4. they treat distribution as a learning system
Distribution isn’t just a growth problem.
At the earliest stage, it is also an information system.
Every customer conversation teaches you something.
Every failed outbound campaign tells you something.
Every demo that converts tells you something.
Every channel that doesn’t work narrows the search space.
This is why we increasingly think product and distribution should not be treated as completely separate functions at pre-seed.
Distribution helps you learn what the market wants.
The market helps you build the product.
The loop should be continuous.
5. they know what not to build
One of the biggest advantages of an early founder is the ability to remain small.
That means saying no.
No to unnecessary features.
No to premature hiring.
No to vanity metrics.
No to customers who pull the company away from its core thesis.
No to fundraising milestones that don’t matter.
No to building things because competitors have them.
Capital efficiency isn’t simply spending less money.
It is spending resources on information that changes the trajectory of the company.
founder-market fit becomes more important
There is another consequence of this shift.
If building becomes easier, the founder’s understanding of the market becomes more important.
AI can help generate software.
It cannot automatically give a founder deep customer empathy.
It cannot replace years of domain exposure.
It cannot manufacture trust with a particular buyer.
It cannot guarantee that a founder understands why a workflow exists the way it does.
This is where founder-market fit becomes interesting.
A founder with unusual proximity to a problem can often learn faster because they start with better questions.
They know what to investigate.
They know which signals matter.
They know when a customer is politely interested versus genuinely desperate.
That advantage compounds.
the next generation of founders may look different
We think the next generation of exceptional founders will increasingly come from people who can combine:
domain expertise + AI leverage + distribution ability + extreme learning velocity.
They don’t necessarily need large teams.
They don’t necessarily need to be the best engineers.
They don’t need to know exactly what the company will look like in five years.
But they need to be unusually good at navigating uncertainty.
That is the job of a founder at the beginning.
Not predicting the future.
Finding it.
what this changes for investors
This also changes how we think about evaluating pre-seed companies.
At this stage, historical metrics can be limited.
Revenue may be tiny.
Users may be early.
The product may still change.
The market may not yet have a clear category.
So the question cannot simply be:
“How big is this today?”
It has to include:
“How quickly is this team becoming more right?”
We want to understand the learning loop.
How long does it take to identify a problem?
How quickly do customers respond?
How quickly does the founder incorporate feedback?
How often do assumptions change?
Does each iteration produce a better product, better customer understanding or better distribution?
And perhaps most importantly:
does the founder know what they still don’t know?
That is often more informative than confidence.
the new seed advantage
The early-stage environment is becoming more interesting because the cost of experimentation is falling.
More people can build.
More products can launch.
More ideas can be tested.
That will create more companies.
It will also create more noise.
The challenge for founders is therefore not simply to move faster.
It is to learn faster than the noise around them.
The challenge for investors is not simply to identify the founder with the most impressive product.
It is to identify the founder whose learning loop is already accelerating.
Because a startup at pre-seed is rarely a finished company.
It is a set of hypotheses.
The best founders turn those hypotheses into evidence.
Then they turn evidence into decisions.
Then decisions into a better company.
Again and again.
That is learning velocity.
And in a world where AI makes building increasingly cheap, it may become one of the most valuable founder advantages of all.
Absolutely! In an AI-first era, shipping fast is just motion; the real moat is how quickly you evaluate user telemetry and failure modes to update your product thesis. Speed without high-signal feedback is just noise.
If everyone can build faster, building fast stops being the moat. That tracks with what I see on the content and distribution side too, production capacity got cheap and the constraint moved to knowing what is worth producing. Learning velocity is a good name for the thing that actually compounds.