1. Headline & intro
Insurance was supposed to be one of the last bastions of human-heavy, relationship-driven work. Yet an AI-native brokerage, Harper, has just raised enough money to seriously test the opposite thesis: that a brokerage can behave more like a software product than a traditional agency. If Harper’s bet works, it won’t just be another insurtech story – it will be a template for how AI can hollow out entire layers of white‑collar intermediation. In this piece, we’ll unpack what Harper is building, why investors are so excited, and what this means for brokers, startups and regulators worldwide.
2. The news in brief
According to TechCrunch, Harper – an AI-first commercial insurance brokerage founded in 2024 – has raised about $47 million in a combined seed and Series A round, led by Emergence Capital, with participation from Y Combinator and Peak XV Partners. The company is a licensed commercial agency that connects small and mid-sized businesses to more than 160 insurance carriers for products like workers’ compensation and general and professional liability coverage.
Harper’s CEO, Dakotah Rice, previously founded investment platform Poolit, which shut down in 2023. Harper went through Y Combinator’s Winter 2025 batch and has reportedly acquired over 5,000 customers so far. Rice told TechCrunch the company’s AI systems automate most of the brokerage workflow – from submission routing and document collection to dealing with underwriters – allowing Harper to process over 1,000 customers per month, compared to a traditional human-only sales team that might handle a few dozen deals in the same period.
3. Why this matters
Insurance distribution is one of the least digitised, most fragmented parts of financial services. In many markets, the actual risk is underwritten by a small number of large carriers, but the interface to the customer is a patchwork of small agencies still glued together by email, PDFs and Excel. Harper is attacking that layer, not by selling brokers software, but by being the broker – and letting AI do most of the work.
If the company’s claims about throughput (1,000+ customers per month) are even directionally accurate, the operational leverage is dramatic. A traditional brokerage is constrained by how many humans you can hire and train in a given geography. An AI-heavy brokerage is constrained by compute and data quality. Those are very different cost curves.
The immediate winners are:
- SMBs that struggle with complex, opaque commercial insurance. Faster quotes and simpler workflows reduce administrative overhead.
- Carriers looking for digital distribution without building their own SMB sales forces.
- Investors who see a shot at software-like margins in a business that historically behaved like a services firm.
The potential losers are small and mid-sized brokers whose value-add is largely process navigation rather than deep risk expertise. If an AI-native player can cover the standard cases – daycares, car dealerships, small manufacturers – at scale, human brokers will increasingly be pushed to niche, high-complexity or relationship-critical segments.
But Harper is also taking on significant risk. Insurance is a regulated, liability-heavy business. If AI misclassifies a risk or misses a coverage gap, the legal and reputational fallout lands on the broker. The company is implicitly betting that it can keep error rates low while scaling automation high – a difficult balancing act.
4. The bigger picture
Harper is part of a second wave of insurtech. The first wave – think Lemonade in the US, Wefox in Europe, Hippo and others – tried to reinvent the carrier or become full-stack digital insurers. Many discovered the hard way that underwriting risk is capital‑intensive, cyclically exposed, and tightly regulated. The market mood has since shifted from “disrupt the insurance company” to “make distribution and operations radically more efficient.”
AI-native brokerages like Harper and Gyde, plus tools such as FurtherAI and Vantel (also cited by TechCrunch), sit squarely in this second camp. They are closer to workflow automation and vertical SaaS than to traditional fintech. Their moat, if they build one, will come from proprietary data loops: every application, quote and renewal becomes training data that improves future automation and risk classification.
This approach echoes what we’ve seen in other industries. In real estate, for example, software-enabled brokerages like Compass attempted a similar transformation of agent productivity. In customer service, AI agents increasingly handle level‑1 support and triage, with humans stepping in only for edge cases. Insurance brokerage is ripe for the same pattern: AI handles 80–90% of the standard process; humans focus on exceptions, large accounts and delicate negotiations.
Competitively, Harper’s timing is interesting. Incumbent brokers and MGAs are already layering AI onto their workflows, but most are doing so incrementally, on top of legacy systems and incentives. Harper, by contrast, can design workflows and data models from scratch around AI, which could yield cleaner automation and better unit economics – if it can acquire and retain customers cheaply enough.
The broader industry signal is clear: the future of many “agency” businesses – brokers, recruiters, consultancies – will look suspiciously like software companies with thin human layers rather than the other way around.
5. The European angle
Harper itself is US-based and focused on “middle America”, but the model travels well – and European brokers should pay close attention. Europe’s commercial insurance market is at least as fragmented as the US, with thousands of small brokers operating under complex regulatory regimes (IDD, Solvency II, national licensing rules) and a heavy paper trail culture.
For EU players, the twist is regulatory: the upcoming EU AI Act will classify many insurance-related AI systems as “high risk”, triggering strict requirements around transparency, human oversight and data governance. An AI-native brokerage operating in the EU will not only need to be a licensed intermediary under the Insurance Distribution Directive, but also prove that its algorithms are auditable and non-discriminatory. That’s a higher compliance bar than in most US states, at least for now.
At the same time, European insurtechs such as Wefox (Germany), Getsafe, Clark, Zego (UK) and Alan (France) have already demonstrated that digital distribution and automation can scale under EU rules. The next logical step is to push much deeper AI into the brokerage workflow.
For European SMEs, especially in markets like Germany, Italy or CEE where broker relationships are strong, the winning model may be hybrid: human advisers supported by aggressive AI automation behind the scenes. The Harper story is a challenge to European incumbents to move faster – because if they don’t, a YC‑style AI brokerage could eventually cross the Atlantic, partner with local carriers and compress margins from the outside.
6. Looking ahead
What happens next will depend on a few key execution questions.
1. Can Harper maintain quality at scale? Handling 1,000 customers a month is impressive; handling tens of thousands while preserving coverage accuracy is another matter. Expect regulators and E&O insurers to scrutinise any large AI-driven brokerage closely, especially after the first publicised dispute or uncovered exclusion.
2. Where does the human sit in the loop? The most robust model is likely “AI does the grunt work, licensed humans sign off and handle exceptions.” That limits margin expansion versus a fully autonomous system, but it’s probably necessary for trust and compliance. How Harper balances automation with human oversight will be a leading indicator for the rest of the market.
3. Will carriers embrace or fear this model? Some insurers will love high-volume, digitally clean submissions; others will worry about reduced pricing power and deeper commoditisation of their products. Over the next 2–3 years, watch for strategic distribution partnerships or even acquisitions: a large carrier or global broker could decide it’s faster to buy an AI-native shop than to build one.
4. How defensible is the tech? In a world where foundation models are increasingly commoditised, differentiation will rely on proprietary data, integrations with carrier systems, and specialised underwriting logic. If Harper becomes just another front-end over public LLMs, copycats will proliferate.
For readers, the main signal to watch is behavioural: do SMBs start to treat insurance much more like buying cloud software – online, self‑service, with minimal human contact? If that shift happens at scale, it will open the door not only for Harper, but for a wave of AI-native agencies across finance, HR, logistics and beyond.
7. The bottom line
Harper’s funding round is less about one YC-backed startup and more about a structural bet: that AI can turn labour‑intensive brokerages into high‑margin software-like businesses. If the company executes, traditional insurance intermediaries will be forced to choose between embracing deep automation or retreating to ever‑narrower niches. The open question is not whether AI will reshape insurance distribution, but who will own the data, trust and regulatory licences when it does – and whether incumbents or upstarts will move fastest to claim that ground.
