On June 3rd, one of the most anticipated sessions at our Diia.City United “FutureTech: Artificial Intelligence — The Future in a New Reality” meetup was the “Investments in AI” panel. Moderator Hanna Shuvalova, Principal at Horizon Capital and member of our Strategic Board, joined investors and business leaders to explore how AI startups are actually evaluated and what drives investment decisions.

The panel featured Leonid Podobedov, Partner at hi5ventures; Andriy Dovzhenko, Partner at SMRK; Zoya Dronshkevych, Business Development Director at Kyivstar; and Anton Verkhovodov, Partner at DefenseTech D3.

Here are the most compelling insights from the discussion.

What Defines an AI Startup?

With nearly every company now using artificial intelligence to some degree, the moderator’s opening question was crucial: How do you actually determine which products qualify as “AI” startups and which don’t?

Leonid Podobedov outlined hi5ventures’ criteria:

I actually analyzed all our portfolio companies — they all use AI in some way now. But here’s an example of doing it right: our portfolio company SOC Prime took the best existing models, trained them on their proprietary dataset, and created an AI product they sell as their core offering.

Andriy Dovzhenko shared his perspective:

Only the customer can determine what is AI and what isn’t. You approach them and say: ‘I have this thing, it’s AI-powered, so it costs $20 more.’ If the customer is willing to pay even $30 — then your company is truly AI. But if they say ‘I don’t care if it’s AI or human’ — then maybe it’s not really an AI product.

According to the investor, the key distinction is whether AI forms the foundation of the product or is just one of many features.

What Investors Look For

The experts revealed key criteria for evaluating AI startups:

Team and Technology Understanding

Anton Verkhovodov emphasized the importance of expertise:

When a team comes to us wanting to use AI tools, we try to determine whether they actually understand what this is and what they’re getting into. Together with our pool of technical experts, we assess whether they can handle the complexity of the task.

Customer Willingness to Pay

Leonid Podobedov shared a telling example:

We don’t evaluate the cost of using AI as much as whether customers are willing to pay for it. We looked at a startup that could save huge costs for fashion brands. The question was: spend $200,000 on a photoshoot or pay $5,000 to have AI create everything? Everyone understood they’d be willing to pay.

However, this project didn’t succeed for other reasons:

The problem was that it turned out to be very easy to replicate. General models started learning very quickly. Plus, competitors emerged who raised $100 million in the US and with those budgets could train their models faster and better, Leonid explained.

Specificity as Competitive Advantage

Andriy Dovzhenko outlined a strategy for Ukrainian startups:

If a team builds their own model and trains their neural network to solve a very specific task, they’ll lose to large LLMs on general questions but win in their narrow niche. It’s like a narrow specialist beats a generalist.

Financial Results Trump Technology

Zoya Dronshkevych revealed Kyivstar’s approach:

Our priority is using AI across all our businesses. AI is an added benefit that makes business more efficient, but we evaluate based on results in numbers.

For partnerships with Kyivstar, startups need:

  • Strong product team
  • Clear product and use case
  • Potential for significant business impact
  • Scalability to other markets

Defense Tech: The Most Promising Direction

The experts paid special attention to defense technologies.

Leonid Podobedov is confident:

In the near future, stories at the intersection of defense and AI will develop incredibly well. There are still many unsolved problems. Ukrainian teams now have the opportunity to test, but they need to move fast — attract capital and resources.

Anton Verkhovodov shared his view of the real situation in defense:

In defense, the potential for AI is enormous, and it’s almost unrealized. We might think we have guided drones, but this is still in very early stages. It works in 10% of situations, but we need to reach 90-100%.

According to him, over 60 teams in Ukraine are working on AI for drones, but there’s still no scalable solution.

Many teams have built themselves as those who quickly make something. Though the world is more complex than they imagined. We now need more fundamental, systematic work on military products, Anton added. 

Cybersecurity as a New Opportunity

Zoya Dronshkevych highlighted another promising direction:

We at Kyivstar get attacked 15 times a week — that’s normal. There’s huge potential in businesses that can automatically stop Russian cyberattacks using AI-based solutions here and now.

Advice for Startups

Andriy Dovzhenko summarized the philosophy of successful AI startups:

Don’t develop AI — develop products that use AI to meet the real pains.

The experts agreed that competing with giants in the general market is a bad idea. Instead, focus on specific, narrow tasks where AI provides real advantage. Create products that solve specific problems better than anyone else.

Use existing AI to make a product that doesn’t currently exist in the market. The better you use AI, the better your product will be.

_____

* FutureTech is a series of open Diia.City United events dedicated to the emerging technology trends. 

Join Diia.City United business association and gain access to unique business events and development opportunities. Together, we influence the establishment of clear business rules and protect them for the global growth of technology businesses from Ukraine.

Want to be part of the community shaping the future of the Ukrainian tech industry? Feel free to contact us at members@diiacityunited.ua.