Generic AI Support Resolves 25% of Queries – Vertical AI 75%

01 /The Concept

Gradient Labs claims to have built the only AI support agent in the world purpose-built for financial services – banks included.

The specialization matters. A generic AI support agent, Gradient Labs argues, can successfully resolve only 10–25% of customer inquiries – the simplest, most generic ones.

Gradient Labs' AI agent, by contrast, can autonomously and successfully handle the other 75% of inquiries in fintech specifically – through specialized training and integration approaches designed for the platforms financial companies actually use.

For example: a customer messages the bank chat saying "I lost my card, please help." The AI agent immediately freezes the card, then asks whether the customer would like a replacement – and if yes, initiates the reissue process and confirms the delivery address. No human involved.

The result is a 98% customer satisfaction score, according to surveys conducted at Gradient Labs' bank clients.

Because the agent was built for fintech from day one, it can handle 40–60% of inquiries successfully on the first day of deployment, before any additional custom training – a figure backed by real-world rollouts.

Pricing is outcome-based: not a subscription, not per-query – only per successfully resolved customer request.

Gradient Labs launched in November last year, having raised an initial £2.8 million before launch in August. Since then its client base has grown to fintech companies across Europe and the UK. The startup has now raised a new round of $13 million.

02 /Why It Matters

Customer support looks like an operational detail – but it's actually a meaningful business lever.

Support is expensive – according to Gartner, a single live-agent interaction costs an average of $13.50. And a single bad support experience drives churn roughly 30% of the time on average. In the US that number is closer to 20%; in Latin America, around 50%. Multiple bad experiences push the churn probability to around 45% – though interestingly, in the US the number rises to nearly 60%, while in Latin America it drops to around 30% (the theory being that the first bad experience already filtered out the less tolerant customers)

Speed matters too. 66% of customers expect a chat response within 5 minutes; 22% allow up to 30 minutes.

The upshot: fast, high-quality support isn't optional. And improving it requires AI agents that actually perform at scale – which generic ones don't, at least not past the simple cases. That's the structural opening for domain-specialized agents like Gradient Labs.

There's also an underappreciated upside: better customer experience allows companies to charge more. Research suggests customers are willing to pay a premium of around 16% at businesses with excellent service quality in coffee, 14% in hotels and medical clinics, 12% in restaurants, and 10% for airline tickets. Financial services aren't listed in that data, but the logic holds.

03 /Opportunities

Gradient Labs was founded by former employees of UK digital bank Monzo – which then became one of its clients.

A parallel dynamic played out at GrowthX ([covered here](/review/novaja-biznes-model-dlja-bystrogo-i-pribylnogo-rosta)), which raised $12 million in its first round this past May. Its founder built an SEO platform while working as CMO at Deepgram, then left to start his own AI SEO agency – and Deepgram signed on as the first client.

This suggests a simple and underused startup sourcing method: if you work at a company right now, ask yourself what you could build in your area of expertise that your own employer would pay for – even if you built it outside the company.

The second, more specific conclusion: vertical AI agents for specific industry segments represent a real opportunity, because generic AI agents simply can't match the resolution rates that a purpose-built one can achieve.

As a starting point for identifying which verticals to target, consider the gap between customer service expectations and actual experience across industries. Airlines have the largest gap at 33 percentage points. The 18–25% gap range covers healthcare, pharma, retail, banking, investment brokerage, telecom, and insurance.

These are sectors where customer expectations are already not especially high – and yet reality still falls short. That's a clear signal where a specialized AI support agent could have the most impact.