What can we learn from Y Combinator?
Y Combinator is often associated with celebrated startups and large funding rounds. Its most durable lessons, however, concern the work that comes before fundraising: finding a real problem, staying close to users, building something useful early and learning quickly from evidence.
Y Combinator, or YC, is best known as the startup accelerator behind companies including Airbnb, Stripe and Dropbox. That record can make YC appear to be a gateway to Silicon Valley. Its deeper value lies elsewhere: in a small set of demanding principles distilled from watching thousands of young companies succeed, struggle and fail.
A strong startup does not begin with an impressive pitch. It begins by solving a problem that matters to someone.
Make something people want
YC's best-known instruction sounds almost too simple. In practice, founders often fall in love with a technology or solution and then reinterpret the world to justify it. A stronger starting point is not “Which technology should we use?” but “Who repeatedly loses time, money or opportunity because this problem remains unsolved?”
This distinction matters especially in the age of artificial intelligence. An “AI platform” is not itself a value proposition. The founder must identify which decision will improve, which task will become shorter or which loss will be reduced.
Understand users through their real behaviour
YC repeatedly asks founders to talk to users themselves. This does not mean asking whether people like an idea. Polite enthusiasm and predictions about future behaviour are weak evidence. Better conversations examine specific past events: When did the problem last occur? What did the person do? How much did it cost? Who made the decision? Why was the existing method inadequate?
In mining and industry, “the user” may not be one person. An operator uses the product, an engineer validates it, an IT team integrates it, procurement contracts it and management funds it. A serious product must work within the realities faced by all of them.
Launch early without treating quality carelessly
Waiting for perfection postpones contact with reality. YC therefore encourages founders to launch, observe use and improve. A minimum viable product is not permission to release something careless. It is the smallest complete solution capable of testing the most important assumption and delivering recognisable value.
In safety-critical industries, moving quickly cannot mean ignoring risk. A pilot can be bounded, retain human oversight and define success and stopping conditions in advance. Responsible work and rapid learning are not opposites.
Do things that do not scale
Paul Graham's advice to “do things that don't scale” has become central to YC thinking. Founders may need to recruit the first users personally, install the product themselves, stay close when something fails and perform parts of the service manually.
These actions cannot continue unchanged at large scale. Early on, however, they reveal what the product must do, where trust breaks down and where value is actually created. Manual work can be treated as research that discovers the requirements of the future product.
A few users who care deeply are better than broad indifference
Many people saying that a product is interesting is weaker evidence than a small group using it repeatedly, recommending it and paying for it. The first objective is not to serve an entire market. It is to become exceptionally useful to a clearly defined group.
Retention, repeat use and willingness to pay reveal more than press attention or download counts. Growth matters, but only when it rests on users receiving real value.
Measure progress without worshipping empty numbers
A startup is designed to grow, so YC pushes teams to select a meaningful measure and review it frequently. Yet not every rising number is progress. Traffic, followers and registrations can increase while users quietly leave. A useful metric connects the customer's success to the company's success.
Focus also matters. Events, awards, partnerships, new markets and extra features can all appear valuable. If they displace building, speaking with users and measuring results, a team can remain busy without advancing. A small team's advantage is its ability to abandon weak work quickly and change its decisions when evidence changes.
Fundraising is a tool, not the result
YC's relationship with venture capital can make fundraising seem like the objective. Capital is useful when it accelerates a demonstrated opportunity. It cannot make an unwanted product valuable.
Founders must understand their runway, burn and path to revenue. The durable company is not the one that announces a round. It is the one that repeatedly delivers enough value for customers to support its continued existence.
Applying the lessons to Mongolian mining and industry
Much YC advice emerged from software and fast-moving markets. Mines and industrial operations have longer procurement cycles, difficult integrations and strict safety requirements. Silicon Valley's pace should not be copied blindly.
The learning discipline does transfer. A Mongolian team can begin not with a broad “mining AI platform,” but with one expensive, recurring problem: unplanned equipment downtime, ore misclassification, slow reporting, inefficient water or energy use, or repetitive work that exposes people to hazards.
The team can observe the work on site, build with the people who perform it and test within a controlled scope. It can prove value through downtime avoided, recovery improved, decisions shortened, cost reduced or exposure removed. Some steps may initially remain manual; repeated knowledge can then be encoded in the product.
Done well, Mongolia's mines can become more than buyers of imported technology. They can become demanding proving grounds where local solutions are tested, strengthened and prepared for export. Difficult operating conditions can be an advantage when they produce products robust enough for the wider world.
Take the discipline, not the costume
The YC path is not appropriate for every company. Not every good business requires venture capital, and not every organisation should pursue extreme growth. Industrial systems cannot be tested like consumer apps.
The transferable principles are more fundamental: begin with a real problem, understand the user's work, test the most important assumption early, measure genuine results and change course when reality contradicts the plan.
YC's deepest lesson is not how to raise capital. It is how to face reality early, learn quickly from users and keep building what proves useful.
Sources and interpretation
This note draws on Y Combinator's “YC's Essential Startup Advice” and “How to Talk to Users,” together with Paul Graham's essays “Do Things That Don't Scale” and “How Not to Die.” The application to Mongolian mining and industrial ventures is my own interpretation.