From Hype to Substance: VCs Demand Real AI Innovation

Venture capital investment soared to a 10-quarter high, reaching €108.3 billion in the first quarter of 2025. This significant surge was largely driven by artificial intelligence, which alone accounted for over €44.6 billion of the capital raised.

In recent years, the AI sector often felt like an automatic money-maker. Investors, eager to capitalize on the next major trend, were quick to fund almost any startup that simply mentioned AI in its presentation. The actual implementation or practical value of the idea wasn't always a prerequisite. In some instances, merely the perception of innovation was enough to garner a unicorn valuation. However, investors are now growing increasingly discerning about the practice of ‘AI-washing’.

As the CEO ofGradient Labs— an AI customer service platform designed for highly-regulated industries — I have observed investors becoming more cautious regarding AI-washing: the tendency for companies to overstate their AI usage or capabilities.

And rightly so. Despite its immense promise, AI technology comes with substantial risks. Gartner predicts that 40% of agentic AI projects will be canceled by 2027, while MIT research indicates that 95% of pilot projects ultimately fail. Even Sam Altman, arguably the sector’s most prominent beneficiary, has underscored that we are currently in an AI bubble.

History demonstrates that such rapid investment spikes are unsustainable. While AI remains a vibrant sector, overall VC investment declined by 21% between Q1 and Q2, signaling an end to the era of readily available capital. Startups can no longer rely on mere buzzwords to stay afloat.

Despite this broader slowdown, I recently guidedGradient Labsthrough a successful €11.1 million Series A funding round, completed in just one week. My key takeaway? Investors are no longer primarily concerned with missing out on the next big wave. Instead, their focus has shifted to a company’s proven ability to deliver. They seek tangible evidence: functional demonstrations, marketable products, and customer endorsements that validate ambitious claims, rather than just promises.

In the past, incorporating industry jargon throughout a pitch deck might have been sufficient to secure a term sheet. But today, simply labeling oneself an “AI-native startup” no longer serves as a differentiator.

This doesn't imply that the opportunity for AI innovation has passed. Many AI companies are pitching in similar niches, often lacking a truly distinctive product or innovative vision — with founders simply trying to ride the hype. However, investors are becoming increasingly adept at identifying AI-washing.

The positive aspect is that genuine innovation — products crafted for a clear, specific use case — now truly stands out. This is particularly true when the founders and their team possess a deep and authentic understanding of the market they aim to serve.

For my co-founders and me, establishing an AI startup was never solely about financial gain. Our motivation stemmed from a problem we encountered while working at Monzo, a leading UK fintech: highly regulated industries were unable to leverage automation due to strict compliance mandates. AtGradient Labs, we developed a solution to address precisely this challenge.

Our approach wasn't AI for AI’s sake; it was AI with a clear purpose — and that distinction proved crucial in our boardroom discussions.

AI technology is advancing rapidly, meaning what seems novel today could become standard tomorrow. You must critically assess what truly differentiates your product and whether that uniqueness will endure when you make your pitch. Consider the likelihood of major players like OpenAI solving the same issue with their next GPT model release. If that probability is high, you may be pursuing the wrong strategy.

Our approach centered on hiring individuals with profound expertise, designing something truly distinct, and rigorously proving its effectiveness. We weren’t satisfied with an agent that provided accurate information 95% of the time. In highly regulated sectors, even a single error can result in irreparable reputational damage.

We dedicated 14 months to obsessing over the product, prioritizing its development over our pitch. Every detail had to be perfect before we went live, and this meticulousness paid off: our platform consistently outperformed human customer service agents — and our customers were genuinely impressed.

As a result, we didn’t need to rely on flashy marketing speak or inflated promises to attract the attention of VCs. They perceived the inherent quality, observed the compelling metrics, and recognized the potential for a category-defining product.

While product excellence is paramount, who you know also makes a difference — especially when distrust is prevalent. We meticulously laid the groundwork for months before our funding round, actively meeting investors and providing regular updates.

By the time we were prepared to formally pitch, we weren’t just another email landing in an inbox; we were continuing ongoing conversations with individuals already familiar with us and our story. For investors, this meant they had ample opportunity to assess our credentials, verify our claims, and speak directly with our customers. They were confident in our legitimacy, and when the time came to invest, they were ready to proceed swiftly.

Not every investor will say yes, but even rejections hold significant value. VCs are deeply embedded in networks, and information travels. The relationships we cultivated and the trust we garnered meant many were willing to open doors for us, even if they ultimately passed on the direct investment opportunity. This network effect created its own momentum — lending credibility, fostering urgency, and signaling that our endeavor warranted serious consideration.

The intense AI boom may be moderating, but dedicated founders have no cause for alarm. Abundant capital remains accessible — provided your intention is to genuinely solve problems, rather than to deceive.

Dimitri Masinis the co-founder and CEO ofGradient Labs, which offers the only customer operations AI agent specifically built for financial services.Masinpreviously held roles at Google and Monzo, where he served as the Vice President of Data Science, Financial Crime, and Fraud.