The Humble Origins and Giant Leaps An Inside Look at the AI startup founding story

Have you been following social media recently? If so, chances are you have seen the headlines: “This AI unicorn just hit $10 billion in valuation” or “These teenagers developed an AI app in one weekend.” It sounds like a fairy tale, doesn’t it? Like someone typed ‘create a rich company for me’ as a command and – voila! As someone that hangs out amongst the tech community in KL and Singapore, I can honestly tell you that the truth is much more ‘sweat and kopi’ than ‘fairy tales’ or ‘commands. You see, every big tech company, that you currently see, had an AI startup founding story that started from something basic to fix. But no tech company has started with ‘tech’ as their reason for starting. They always started from a reason as to why they did.

The real spark behind the AI startup founding story

When we talk about an AI startup founding story, we envision a Silicon Valley garage. Today, that “garage” might actually be a co-working space in Mid Valley or an apartment in Singapore. Regardless of the physical location, founders get the idea for their startup when they see how long it takes for people to complete a manual task. For example: sifting through thousands of legal documents or responding to customer inquiries. The founders of small AI companies such as Perplexity or smaller niche AI companies did not just decide one day that they would create a large neural network. They considered their vision for the AI company and how it could improve access to information. They observed how difficult it was to find things on the internet because of advertisements. So they wondered, “Can we provide the answer directly?”

This is the primary aspect of every successful AI startup founding story: It does not matter how well-designed the code is. If the founder has the belief that something can be accomplished differently than it has before. Then they can proceed to test their concept. Before they begin, the founders of AI companies spend time speaking to others, determining if the problem they are trying to solve is a “pain point” vs a “nice-to-have” or, in business parlance, finding a hook. Without this kind of hook, it does not matter how much GPU processing power a startup has; it cannot succeed without one.

The messy reality of the AI startup development journey

After you’ve got the concept in place, you enter the phase of endurance. This is when the fun part of building an AI startup begins. Building Artificial Intelligence is costly; unlike building a conventional application, there are more expenses than just the hosting costs. Massive processing power is required for Artificial Intelligence. They typically make the MVP (Minimum Viable Product) by putting together an MVP with existing 3rd party models (such as OpenAI or Anthropic) and adding a layer to those. This is a significant part of the innovation strategy of AI Startups. Tey don’t re-invent the wheel; they develop a better car.

The crucial aspect of this phase isn’t only coding but rather collecting data. For artificial intelligence to function, you require good data. In the past I have seen teams spend 80% of their time cleaning excel spreadsheets and scraping websites legally. This is very unexciting work. But this is where the product-market fit of AI startups is validated, if users keep returning with a poorly designed UI and AI with little responsiveness, you have a market.

The AI startup founding story and strategy

Now, let’s delve into the business part. How can these businesses stay alive long enough to reach ‘unicorn’ status? It comes down to having a robust AI startup business model. In earlier days, software was sold with a single license. After that, it became Software as a Service (SaaS) with a monthly subscription. Today, with AI, that is moving toward a “usage-based” model where the business pays based on what work is completed by AI. A founder needs to have an obvious competitive advantage over other AI startups. Since most businesses working in AI share the same ‘starting point’ by using the same fundamental AI models, the ‘moat’ will need to come from somewhere else.

For some founders, the competitive advantage might come from their own proprietary data. As an example a large volume of palm oil logistics information or an in-depth knowledge of the Southeast Asian banking system. For others, the competitive advantage could come from building an intuitive user experience. Typically, an AI startup’s long-term growth strategy will be to drill down deeper into ‘niche’ markets. Instead of trying to compete against Google at the general search level, a savvy founder would build the ‘best AI for Malaysian tax law.’ So by the time large corporations become aware of them, they have already built a customer base and accumulated a wealth of proprietary data specific to that market. This is how ‘underdogs’ can win.

Scaling up without losing the soul

Once a startup completes its first two years, it will start the growth phase of scaling. Although it’s great to see the growth; it can be daunting as well. You will need to hire 50 new employees, find a larger space for the office, as well as make sure the servers don’t crash from the overload of 100,000 new users from Indonesia or Thailand! The founders must balance the need for speed in their product development while also ensuring the safety of their customers. There is enormous pressure on AI entrepreneurs and companies to develop new features and get them released on a weekly basis. If your AI starts making up information that doesn’t exist for a corporate customer. You will lose your customers’ trust immediately!

When looking at AI startup today’s other founders have a similar vision to where the market is headed in AI. It is no longer just about who has the best AI, but who has the most effective AI. The trend is rapidly transitioning from “chatbots that talk” to “agents that perform”. For example, think of an AI that can not only tell you about a flight but actually schedule the flight… process your refund when the flight’s late… and find a hotel that is better than what you had booked.

There are moments when you feel like you are the smartest person in the world and there are times when you feel like you are an idiot because some large telecommunications corporation has created a new service for free that provides the same service as your business. However, it creates a very creative environment; and through that creativity, both the ability to develop and deliver to customers will continue to benefit every day.