HARIO SETO S

NEVER A BETTER TIME TO START

AI is changing what a small team can build.

Work that once took months can now be compressed into minutes. The opportunity is not to build smaller companies. It is to attempt bigger problems with fewer people, faster cycles, and far more leverage.

THE NEW STARTUP WINDOW

Great startups often appear when technology shifts quickly, costs fall, and incumbents lose their advantage.

The internet, app stores, and mobile created earlier waves. AI is creating another major opening where expertise, production, and experimentation are dramatically cheaper.

WHY NOW

STARTUPS ARE MORE POWERFUL

Twenty years ago, a YC startup could spend three months building one product.

Today, a coding agent may reproduce that work in minutes. Founders can now coordinate software, research, design, operations, and analysis at a scale previously reserved for large companies.

AI RAISES THE AMBITION BAR

A weak response to AI is asking which old startup ideas are now obsolete.

A stronger response is asking what company was impossible to build before. AI gives founders capabilities that once required large teams. The advantage belongs to people willing to attempt harder and more valuable problems.

GOLDEN AGE OF STARTUPS

Hard technology remains difficult, but AI reduces barriers around research, prototyping, hiring, and coordination.

A project that once required a large institution may now begin with a small team using agents and compute. This creates room for a new generation of deeply ambitious companies.

THE STARTUP IS NOT OVER

Some people fear powerful models will absorb every valuable company.

The opportunity may be larger. New startups can become more valuable and impactful because they can build faster, reach further, and solve problems that were previously uneconomic. Intelligence on demand expands the design space.

AI LEVERAGE

BUILD WITH AGENTS

A small team can now automate major parts of a startup using AI agents.

This does not remove the need for founders. It changes the founder’s job. The scarce skills become judgment, taste, agency, customer understanding, product direction, and knowing where durable value can be created.

TOOL FLUENCY COMPOUNDS

Years of experience still matter, but AI fluency is becoming a major advantage.

People who grew up using these tools can test ideas, automate operations, and learn unfamiliar fields quickly. In a fast-moving market, working effectively with new tools may outperform traditional credentials.

CYCLE TIME IS STRATEGY

When costs fall and cycle times shrink, founders can test more assumptions with less capital.

The advantage is not merely shipping faster. Faster cycles create more customer feedback, more data, and more opportunities to correct direction. Speed becomes a learning system.

DO THREE MONTHS OF WORK

If AI compresses three months of work into minutes, the goal is not to stop after one prompt.

Use the remaining time to explore more directions, improve quality, talk to customers, and build stronger systems. The productivity gain should raise the standard of what a founder attempts.

CONVICTION

FIND THE UNPOPULAR TRUTH

Many important companies begin with a belief that experts dismiss.

Develop reasonable conviction in something the market has not understood. When new evidence continues to support the idea, being misunderstood can become an advantage because competition remains limited.

OPENAI’S EARLY SECRET

For years, only a small number of people believed AGI was possible.

That disbelief gave OpenAI time to research, recruit, and build without overwhelming competition. A startup does not need everyone to agree. It needs enough strong people who understand the same emerging truth.

STUDY NEW EXPONENTIALS

People struggle to understand exponential change.

Founders should search for technologies, costs, behaviors, or capabilities improving faster than the market expects. Opportunity appears when reality has moved, but conventional wisdom still uses an older model of what is possible.

VISION CAN BE CLEAR FIRST

A very ambitious startup may begin with a clear destination and unclear first steps.

OpenAI wanted to build AGI long before it knew it would become a product company. Uncertainty is acceptable, but progress is required. Take imperfect steps that generate evidence and reveal the next move.

CONVICTION NEEDS EVIDENCE

Strong conviction should grow through data, experiments, technical progress, or customer response.

When nobody credible shares the belief and no evidence improves over time, pay attention. The goal is not blind confidence. It is a belief that survives contact with reality and becomes more precise.

PEOPLE AND NETWORKS

FIND YOUR PEOPLE

Startups are difficult and lonely to build alone.

You do not need universal support, but you need a few people who understand the mission. Strong co-founders, early teammates, and trusted peers provide complementary skills, resilience, and shared conviction when progress is uncertain.

NETWORKS COMPOUND SLOWLY

Important startup relationships often begin years before their purpose becomes clear.

Sam Altman met Greg Brockman while helping Stripe recruit him. Eight years later, they started OpenAI together. Useful networks are built through repeated trust, not last-minute transactions.

BE MILDLY HELPFUL

One of the best ways to build a valuable network is to help capable people in small, genuine ways.

You gain exposure to interesting problems, earn trust, and create future possibilities. Help should be useful now, even when there is no visible return. The long-term effects are unpredictable.

CHOOSE HIGH-DENSITY ROOMS

Founders benefit from environments where ambitious builders repeatedly meet each other.

Accelerators, technical communities, startup cities, and strong networks increase the probability of finding co-founders, early hires, investors, and collaborators. Proximity creates more chances for luck to compound.

BUILDER MINDSET

OPTIMISM CREATES MOMENTUM

Paul Graham helped early founders leave each weekly dinner believing their companies could matter.

That optimism was not empty motivation. It redirected energy toward action. In the early stage, founders often need someone who can convert confusion and rejection into a clear next move.

THE FLIGHT INSTRUCTOR MODEL

Startup advice works better like flight instruction than a classroom lecture.

A useful mentor sits beside the founder and points out the immediate mistake, missed signal, or next action. Specific feedback changes behavior faster than abstract theory because it connects directly to the current situation.

EARNESTNESS BEATS SARCASM

It is easy to mock people attempting difficult things and receive attention for a clever comment.

The cost is subtle. Sarcasm trains the mind to observe instead of participate. Put energy into building, helping, and learning. Over time, real output compounds while internet points disappear.

EXPECT RESISTANCE

The more a company threatens the existing order, the more criticism it may attract.

Dismissal does not prove an idea is correct, but it is often part of important work. Founders must hear useful criticism without allowing every attack to redirect the mission.

POWER AND RESPONSIBILITY

STARTUPS DISTRIBUTE POWER

AI can distribute capability across millions of people, or concentrate power inside a few companies and models.

Successful startups help widen access by creating alternatives, new applications, and independent economic value. Building a strong company can contribute to a healthier ecosystem.

SAFETY IS REAL

As AI systems become more capable, alignment, security, and loss-of-control risks become practical concerns.

Founders should not treat safety as someone else’s problem. Products need secure infrastructure, clear safeguards, responsible deployment, and human values guiding the systems.

PROTECT HUMAN AGENCY

A future with material abundance but no privacy, freedom, or meaningful choice would still be a failure.

Progress should increase human agency. People should gain more control over their time, work, creativity, and decisions. Technology should expand participation, not create a comfortable surveillance system.

EXECUTION

TAKE THE FIRST IMPERFECT STEP

A bold vision is useless if it never produces forward motion.

The first product, experiment, or structure may be wrong. It still matters because action generates evidence. Build something small enough to test, but meaningful enough to reveal whether the larger direction has real potential.

AMBITION NEEDS AGENCY

AI gives founders more intelligence, tools, and speed, but the founder must still choose the problem and drive the system.

Agency means noticing what matters, deciding without perfect information, coordinating resources, and continuing through uncertainty. Tools amplify the person who acts.

FAIL WITHOUT ENDING THE STORY

Early startups can fail without destroying a founder’s future.

Technology ecosystems often reward learning, persistence, and the ability to try again. Keep the ambition, but do not make every setback a final judgment. Experience, relationships, and judgment can compound even when one company fails.

ENJOY THE BUILD

Startups are stressful because the outcome is uncertain and the work feels personal.

Keep the drive, but allow room for happiness during the process. Careers unfold through many attempts, mistakes, and unexpected connections. The current company matters, but it is part of a much longer journey.

THE FOUNDER’S PLAYBOOK

1. FIND THE SHIFT

Identify the capability, cost curve, or behavior that changed.

Do not begin with a fashionable feature. Begin with the structural shift. Ask what can now be built, delivered, or distributed that was impossible or uneconomic only a few years ago.

2. FORM A STRONG BELIEF

Choose a large problem and develop a clear point of view.

Look for evidence that the market has not fully absorbed. Speak with experts, build prototypes, and track progress. Keep refining the belief until it becomes specific enough to guide product decisions.

3. GATHER THE MISFITS

Find a small group with complementary skills and shared conviction.

The team does not need to look conventional. It needs trust, speed, technical or market strength, and the willingness to continue when outside validation is limited.

4. USE AI AS LEVERAGE

Use agents to compress research, coding, design, operations, and analysis.

Then reinvest the saved time into higher standards, more experiments, and deeper customer understanding. AI should increase the ambition and quality of the company, not merely reduce effort.

5. BUILD, LEARN, REPEAT

Ship an imperfect first version, measure the response, and improve quickly.

Each cycle should answer a real question: Does the technology work? Does the customer care? Can the system scale? Evidence turns a bold idea into an executable company.

THE OPPORTUNITY

There may never have been a better time to start an ambitious company.

AI has lowered the cost of capability, shortened the path from idea to product, and expanded what small teams can attempt. Choose a meaningful problem, take the first step, and begin generating evidence.

PICK ONE IDEA AND GO DEEP

Certainty does not come from thinking longer.

It comes from choosing a direction, meeting customers, building something real, and learning from what happens.

STOP SEARCHING FOR PERFECT

THE PERFECT IDEA DOES NOT EXIST IN YOUR HEAD

A brilliant idea may solve nothing urgent.

A modest idea may reveal a huge market once real users expose the deeper pain.

REALITY IS THE DECISION-MAKING TOOL

Use conversations, usage, payments, objections, and retention as evidence.

The goal is not to prove perfection. The goal is to learn what reality says next.

YOU DO NOT NEED PERFECT FOUNDER-MARKET FIT

Relevant experience helps, but it is not a gate.

A founder who studies deeply and works closely with customers can become highly credible in a new domain.

CURIOSITY CAN BECOME EXPERTISE

Founder-market fit can be built.

Learn the language, observe workflows, understand incentives, do the job, and solve real problems repeatedly.

BOOM SUPERSONIC PROVES THE POINT

His background did not make him the obvious aerospace founder.

He chose an ambitious problem, learned the domain, and built the expertise required to pursue it.

COMMIT TO ONE DIRECTION

MULTIPLE IDEAS CREATE BAD DATA

Shallow experiments produce ambiguous results.

You may abandon a strong idea because execution was weak or continue a weak idea because the signal was noisy.

DEPTH BEFORE COMPARISON

The commitment does not need to be permanent.

It needs to be serious enough to generate clear evidence from customers, usage, and willingness to pay.

BURN THE OTHER BOATS

Commitment requires subtraction.

Stop developing alternatives. Tell customers when you have pivoted. Remove ambiguity and give one idea a fair test.

WEAR A NEW SKIN

Change the name, website, email, positioning, and internal narrative when needed.

The new market cannot remain a side project attached to the old company.

COMMITMENT CREATES SIGNAL

You ask better questions, ship more relevant features, and notice recurring patterns faster.

Focus improves both execution and learning quality.

GOVDASH COMMITTED THROUGH REPEATED PIVOTS

Each pivot became a real company, not a side experiment.

Deep commitment to government contracting eventually produced more demand than the team could easily handle.

BECOME THE CUSTOMER EXPERT

COULD YOU RUN THE CUSTOMER’S BUSINESS?

Learn the workflows, crises, margins, lost revenue, buying authority, and current workarounds.

Know what customers will pay to improve and why.

KNOW THE PROBLEM BEHIND THE REQUEST

A request for faster support may hide missed revenue, weak follow-up, staff overload, low trust, or poor scheduling.

Find the structural problem.

COULD YOU TEACH THE PROBLEM?

Aim to become one of the most informed people on the problem.

Use interviews, observation, research, product data, and direct experience.

DO THE JOB YOURSELF

Doing the work reveals edge cases, shortcuts, emotional pressure, trust requirements, and hidden constraints.

Observation turns assumptions into evidence.

LEARN AND BUILD IN ONE LOOP

Understand the need.

Build the smallest useful solution.

Watch customers use it.

Learn from the result and improve the next version.

VALIDATE THROUGH REAL BEHAVIOR

CUSTOMER PULL IS THE STRONGEST SIGNAL

Compliments are weak evidence.

Stronger signals are payment, repeated usage, referrals, expansion requests, and disappointment when the product is unavailable.

USAGE TURNS THEORY INTO DATA

Interviews create hypotheses.

Usage shows whether the product saves time, increases revenue, reduces risk, improves quality, or earns a place in daily work.

MEASURE THE PROBLEM, NOT THE EXCITEMENT

Ask what happens when the problem remains unsolved.

Measure lost money, wasted time, affected users, frequency, urgency, and existing budget.

LOOK FOR AI-ERA ADVANTAGE

The strongest AI companies combine frontier capability with deep ownership of customer outcomes.

BUILD AT THE EDGE OF MODEL CAPABILITY

Know why the product fails today.

Track accuracy, latency, cost, context, reliability, and review needs. A stubborn bottleneck may become the company.

LIVE IN THE FUTURE AND BUILD WHAT IS MISSING

Operate close to the frontier.

Look for missing tools, workflows, safeguards, and infrastructure that prevent the future from working reliably today.

UNDERSTAND EVERY BOTTLENECK

Do not rely on vague faith that AI will improve.

Map each failure point and decide which bottlenecks you can solve directly.

SELL THE OUTCOME, NOT JUST THE SOFTWARE

As software becomes cheaper to produce, code alone becomes less defensible.

Value shifts toward trust, licenses, operations, data, distribution, and outcomes.

VERTICALIZE THE BUSINESS

Instead of selling software to an industry, deliver the service itself where feasible.

Owning the outcome creates deeper value and stronger economics.

CORGI INSURANCE OWNS THE OUTCOME

The company aimed to control underwriting, service, and insurance operations.

Owning more of the stack improved speed, pricing, and economic control.

OWN THE FULL STACK

Owning more of the outcome is harder but strategically powerful.

It allows the company to redesign the whole process around AI instead of legacy constraints.

CHOOSE THE MOST AMBITIOUS VERSION

Pursue the version that can transform a sector.

Greater ambition can attract stronger talent, justify deeper investment, and create a moat worth building.

FAILURE CAN REVEAL THE REAL COMPANY

A failed idea can still produce the customer truth needed to discover a stronger opportunity.

A FAILED IDEA CAN PRODUCE VALUABLE TRUTH

Failure after commitment is informative.

You gain customer data, stronger execution skill, clearer conviction, and a better basis for the next decision.

THE FIRST PROBLEM IS OFTEN TOO SHALLOW

Going deep does more than validate the original idea.

It exposes broken systems, missing infrastructure, ignored workflows, and problems outsiders cannot see.

THE BOTTLENECK MAY BECOME THE COMPANY

The best pivot often appears after sustained market contact.

You discover what every customer struggles with and what must exist before the original vision can work.

MOVE FAST THROUGH THE FOG

Early-stage founders can only see a short distance ahead.

Choose one direction and move quickly. Serious action reveals opportunities invisible from the starting point.

THE WORST FAILURE IS NOT DECIDING

Pick one idea.

Burn the other boats.

Learn the customer’s business.

Build in tight loops.

Measure real pull.

Go deep enough to find the better idea underneath.