It starts with a pitch deck and a promise.
Somewhere in Bengaluru or Gurugram, a founder clicks to the second slide of their presentation and utters the magic words: “We are an AI-first company.” Cue nods, raised eyebrows, maybe even an impressed investor or two. But scratch beneath the surface, and “AI” often turns out to be nothing more than a few API calls to OpenAI or a glorified use of ChatGPT to automate customer service scripts.
An application programming interface (API) is a bridge for one software to request data or services from another application.
“Every company I come across says that they are an AI start-up. At times, even while using AI in a minimal way, say using ChatGPT to write the summary of their product, start-ups claim that they are AI start-ups. This in turn, makes us investors cautious before investing in a company,” says a venture capital (VC) investor scouting for AI companies.
As companies overstate the usage of AI technology in their products, the tech industry is grappling with how to distinguish genuine AI start-ups from those trying to simply play the buzz.
Take the numbers. As of February 2025, Tracxn reported that India's AI sector comprises approximately 5,100 startups. So, how to cut the noise and identify which is an AI start-up?
Industry insiders highlight that a true AI start-up falls into two primary categories. One is in terms of the new use cases enabled by AI.
“Companies that solve problems that weren’t possible before AI or LLMs became widely available. These companies leverage AI for reasoning, understanding, and problem-solving in ways that were not feasible before,” says Prayank Swaroop, Partner, Accel.
Another aspect is the substantial efficiency gains driven by AI. For businesses in established sectors such as recruitment or social media analytics, the primary consideration is the extent to which AI enhances efficiency. If AI can accelerate a process by 10 to 100 times, it qualifies as a genuinely AI-powered solution, believes Swaroop.
But many experts caution that only improving efficiency isn't enough. If the core product was feasible without AI, it’s not quite a true AI company.
"The fundamental test for an AI company is whether its work was possible before AI existed. If it merely improves efficiency, it’s not an AI company per se, but if it enables something entirely new, it truly is,” says Sajith Pai, Partner, Blume Ventures.
Lifting the AI Mirage
As the hunt for true AI start-ups intensifies, VCs point out that a true AI startup is more than just a company that throws around AI-related buzzwords—it is built on a foundation where artificial intelligence is not just an enhancement but the core driver of its innovation.
Such a start-up should possess proprietary technology and leverage advanced machine learning or deep learning models in a way that fundamentally defines its product, rather than merely improving an existing process.
For example, if a company builds an AI model that can generate human-like code automatically, that’s a fundamentally AI-driven product. Without AI, such a solution wouldn’t be possible. On the other hand, if a company simply uses AI to speed up an existing process—like automating resume screening in recruitment—it’s enhancing an old method rather than creating something entirely new.
“Key criteria include a skilled technical team, access to high-quality data, scalable AI models, and distinct approach in how AI solves a real-world problem. Ultimately, real value lies in how seamlessly the AI technology translates into practical, scalable solutions for end-users effectively and efficiently,” says Anisha Patnaik, founder, LexStart Partners.
Zooming out, experts argue that real AI start-ups operate across different levels of the so-called AI stack—everything from software applications to infrastructure purpose-built for AI.
That could mean developing AI-specific compute and data center solutions, or building apps that rely on sophisticated architectures and foundational models. ChatGPT, for instance, is a textbook case: a product born from GPT architecture, fed by massive datasets, and driven by cutting-edge deep learning. Real AI start-ups create or enhance such core technologies—not just plug them in as a feature.
“In the broadest sense, AI companies are those playing in one or more layers of the AI stack. In addition, one can think of companies building hardware infrastructure built from the ground up for AI use cases such as compute, memory, data centres, etc., as part of this category too,” says Sheetal Bahl, Partner, Merak Ventures.
Some VCs draw parallels to earlier tech transitions—like the cloud or mobile wave. Just as those technologies became foundational, AI is now a universal tool in the arsenal.
So maybe the question isn’t whether a start-up is an “AI company,” but whether it’s leveraging AI in a way that truly unlocks value.
“We believe that there is a new set of tools for problem solving, and that value will be unlocked in new and interesting ways as a result. If companies are able to maximally leverage these tools, then it’s a compelling investment,” adds Sonal Saldanha,Vice President, Investments, 3One4.
The Hype Around AI
AI whitewashing or companies exaggerating the usage of AI is not a completely new thing. It became more widespread in the mid-2010s as AI gained mainstream attention. The hype around machine learning, deep learning, and automation led many companies to overstate their AI capabilities to attract funding and customers.
Just to give an example, Theranos and Juicero are two prominent US based companies that misused the usage of the word “AI” and eventually led to loss suffered by investors.
Theranos, a biotech start-up from Silicon Valley that raised approximately $700 million from investors over its lifetime, shutdown in 2018. Theranos misled investors and the public by claiming that its blood-testing devices, known as Edison machines, used AI-powered diagnostics to analyze a single drop of blood and provide fast, highly accurate medical results.
However, in reality, it couldn’t perform most of the tests that it claimed to, as per the Wall Street Journal. Similarly, Silicon Valley startup Juicero, which raised a total of $120 million from prominent investors including Google Ventures (GV), shut its shop in September 2017. The start-up claimed that its $400 juicing machine, used AI and smart technology to press juice more efficiently by scanning QR codes on special juice packets. However, a Bloomberg report in 2017 highlighted that the juice could be squeezed by people by hand without the usage of an expensive machine.
As AI becomes a more fancy word, it is obvious that more such cases will emerge, claim experts. What becomes important on the part of investors is to do a thorough background check as well of the company.
You don’t want another Theranos in the start-up ecosystem. What you need is to check thoroughly, claim VCs as the race for AI start-ups increases.
