Hype vs. Value: Assessing Early-Stage AI Startups for Real Potential

Two years ago, venture capitalists were broadly investing in everything AI. Now, we’ve all become more selective and educated, seeking solutions to real-life problems where AI adds real value. 

So let’s make the final conclusion from the start: having “AI” in the company name is no more a decisive factor. It took the market two years to get over the AI Big Bang and start separating real value from illusory prospects. 

The real challenge has fallen on pre-seed investors. We had to sharpen our eyes to recognize innovations among hundreds of equally bright pre-revenue AI-based B2B SaaS solutions. Each could turn either a unicorn, or an AI wrap for nothing.

At Pre-Seed to Succeed (P2S), we specialize in evaluating AI startups at the earliest stage, where traction is minimal, teams are lean, and vision matters as much as code. We’re going to share the core principles and practices that guide our investment decisions in this dynamic sector and detect ventures with high potential.

The Misconceptions Around Pre-Seed AI

Investors considering entering the AI space should understand that both the scale and uniqueness of the solution matter. There’s a common misconception about early-stage AI startups, that simply adding a thin AI layer, like a chatbot interface, is enough to build a successful business. The real value lies in owning the data, solving specific problems, and effectively reaching users.

We’ve seen founders focus on surface-level integrations that were easily replicated. The market is abundant with copilots and assistants, but only those grounded in deep domain expertise, with differentiated distribution channels or unique user insights, have a chance to become thriving businesses. Examples include legal copilots for contract intelligence, talent strategy solutions for CHROs, and construction site copilots – complex projects that require hands-on industry experience and a genuine understanding of the processes AI can optimize. These types of solutions hold the highest value.

Another trend is the shift from copilots to agents – autonomous systems that can not only assist but act independently across workflows. These models handle multi-step tasks, reason across systems, and coordinate actions without human prompts. While still early, the startups building real agent frameworks for specific verticals (e.g., financial reporting, legal operations) are showing signs of high-leverage scale.

Also, together with infinite opportunities for businesses, the implementation of AI came along with higher expenses than initially expected. B2B customers are willing to invest in expensive custom solutions only if they seamlessly integrate into their existing workflows and provide either financial benefits or operational relief.

Assessing Innovation In the Absence Of Revenue

In the absence of revenue and considerable traction, the team’s expertise and domain experience are by far the most important things we’re investing into at the pre-seed stage.

If a startup team understands the nuances of AI adoption in high-stakes industries and can accurately discover specific business challenges, that’s already half the battle won. The other half lies in execution – the ability to build thoughtful infrastructure, hybrid interfaces, and leverage edge AI where it truly makes a difference.

Among other “speaking” factors that demonstrate a startup’s value in the absence of revenue, are the speed of execution, and early signs of user love. 

Remember the story of the Tortoise and the Hare? Today, speed of execution matters more than ever. Vanity metrics don’t move us. Startups that can iterate quickly, incorporate user feedback, and show positive progress within a set timeframe, for example, achieving a 10% conversion rate in a pilot project or hitting key milestones, are likely to receive much more investor interest.

Users’ retention matters too: are users coming back? Are they engaging deeply with the product? In LLM-powered platforms, token consumption can be a meaningful proxy for usage depth too.

While some AI companies manage to scale fast, the ecosystem still remains volatile. Margins may compress, moats are hard to sustain, and competition is fierce. In these conditions, user engagement, initial traction and theme strength are the strongest cards in the startup’s deck.

Market Defensibility Is Must

We’ve passed on startups that initially looked compelling but lacked defensibility – proprietary workflows, domain-specific labeling advantages, or user-generated data loops that deepen with time.

One team had a sleek interface but no proprietary data or technical depth. Others rode the hype of early LLM releases but were quickly eclipsed by new capabilities in popular platforms.

For example, we’ve seen startups that attempt to bolt on “emotional intelligence” layers to ChatGPT during user interaction, only to be disrupted a few months later by OpenAI itself, releasing similar features natively. The entire premise of the startup just vanished. These cases taught us the importance of investing in companies with independent technical cores and domain focus.

Red Flags We Watch For

In our assessments, certain warning signs consistently surface:

• Jargon-heavy pitches, light on specifics
• Founders unable to identify paying customers or articulate a pain point
• Replicating standard solutions without a unique technological layer or niche focus

We steer clear of build-and-flip mentalities. We’re in the business of backing founders committed to solving problems and becoming leaders in their space.

Another red flag we watch for: startups that over-index on OpenAI plugins or build entirely inside Notion, Slack, or Discord. These platforms can cut off access or absorb the value themselves. We ask: what survives if the platform changes its API tomorrow?

Advice to Investors Entering AI Without Technical Expertise

For non-technical investors eager to fund AI startups, our advice is simple:

First, partner wisely. Bring in advisors, co-investors, or managers with technical backgrounds who can vet the technology and the team.

Second, back founders who can clearly communicate the problem they’re solving and the steps to solve it. If a pitch lacks clarity, it likely lacks direction.

Third, diversify. Spread investments across sectors, problem types, and business models. It increases the chance of finding breakout winners and limits downside risk.

If a pre-seed startup under consideration shows clear evidence of user need (growing audience, strong engagement and retention levels) and frequent product iterations based on feedback – these are the indicators of a high-potential venture with scalability prospects.

Conclusion

In an AI market the temptation to chase the next big thing is understandable. But early-stage investing requires discipline, skepticism, and a deep understanding of what separates hype from substance.

Investors who apply rigor, partner with experts, and focus on fundamentals will be best positioned to navigate this high-velocity space and emerge with a portfolio of durable, high-impact AI companies.