Competition in the AI ​​field intensifies. What conditions will startups need to break through in 2025?

Source: Heart of the Metaverse

In 2024, thousands of startups emerged based on the power of cutting-edge large language models (LLMs). According to statistics, only one in five companies will survive until the end of 2025. To survive this year, these companies need a little advantage.

AI-driven startups will reshape our world in ways we can't yet imagine, as long as they have the elements of success.

Twilio selected 12 winning companies from more than 500 companies that registered for the "Searchlight" competition. All winners embody several basic principles that startups need to keep in mind when building artificial intelligence solutions.

Keep these principles in mind, and you will have a good start for an AI company that can develop for a long time.

01. Focus on business basics

Don't just count on AI to bring competitive advantages, it's everywhere. The challenges faced by starting and running a startup before the emergence of LLMs still exist today.

What businesses need to do is attract, convert and retain customers while controlling costs. Although the cost of AI is constantly decreasing, it may still lead to rising costs if complex workflows are created.

It is important to note that 72% of IT and finance leaders said that the cost of AI is becoming "uncontrollable".

The power of AI can also become an advantage if the currently complex operations can be encapsulated into an easy-to-use API framework. This is what Twilio did for telephones ten years ago, and it is what large AI models are doing today.

If you want to build a sustainable technology business today, think about how to do it through application programming interfaces (APIs).

02. Don't stay in the "shell"

If you just make a "shell" for an existing LLM, it will be difficult to maintain differentiation in the long run.

For example, if you want to create a tool to help create code, do voice transcription, or scan PDF files and extract information, it doesn't matter how beautiful the interface is, because LLM is already excellent at these tasks.

Entrepreneurs should focus on areas where they can provide differentiated services and bring compound advantages to individuals through data flywheels or network effects.

For example, Goodcall, one of the "Searchlight" award-winning companies, provides automatic voice calling services to enterprises. The company has been collecting anonymous data from more than 4 million customer calls to build a more powerful database and improve analysis.

Another area that startups can focus on is extracting data from unstructured customer conversations.

SpokeAI does just that, extracting data from customers’ voice calls so that business users can see who’s calling them, what they might want, how they’re feeling, and what they’ve previously talked about with colleagues.

03. Understand the trajectory of AI

AI is changing incredibly fast. The number of annual AI patents has grown 31-fold since 2010, and will exceed 62,000 in 2022.

When deciding where to focus your efforts, first understand the trajectory of large language models and where they’re likely to go in the next 12 months. Otherwise, solutions may be outdated before they even make it to market.

For example, large AI labs are currently working to enhance the reasoning capabilities of these models, improving their capabilities in a variety of complex domains. Don’t focus on advanced reasoning unless you have billions of dollars in funding.

In contrast, one of the “Searchlight” award winners, CuraJOY, is a nonprofit grassroots technology company that uses AI and entertainment to improve the accessibility, effectiveness, and fairness of social and mental health support.

This is definitely not an area that large AI models focus on, but it meets a major social need.

04. Catch the hype

New AI solutions attract a lot of interest, but this hype is fleeting. Without a plan to capture these "jumpers" and convert them into long-term customers, the business will quickly fade with the hype.

Products need to maintain high attention, and one way to do this is to continuously improve the product based on customer input.

For example, AI can be used to capture and classify customer feedback and send the highest-value feature requests directly to the product team. This will keep customers buying and promote sustainable growth.

Another way is to continuously improve your capabilities, gain new recognition, and launch customer-friendly services.

For example, Alpharun, an AI phone interview platform, was part of the OpenAI accelerator this year and won the audience award at the 2024 Staffing Industry Analysts conference.

Even though it has achieved good results, the company is not satisfied with the status quo. Now, it has received key technical certifications and offers customers uptime guarantees, international support, and best-in-class reliability, all of which are essential for its target enterprise customers.

05. Looking Forward to the AI-Driven Economy

2024 was the year of AI experimentation, but 2025 will be defined by AI-driven startups that deliver measurable business impact. Through my work at Lightspeed and my experience judging Twilio’s “Searchlight” competition, one thing is clear: the most promising companies are not just creating clever AI implementations, they are building robust businesses that can withstand inevitable technological change.

The winners of the “Searchlight” competition exemplify this approach, building real competitive moats with compounding advantages.

These companies show us that lasting success comes from combining AI capabilities with deep domain expertise and a strong business foundation.

We are at the dawn of a new wave of technology boom, and there is no doubt that some of today’s builders will become tomorrow’s tech giants.