Solidroad raised $25 million to do something most companies still consider impractical: evaluate every customer conversation instead of the usual random handful.
The Series A round, led by Hedosophia and announced April 16, 2026, brings total funding to $33 million for the Dublin-based startup, which also keeps a footprint in San Francisco. First Round Capital, Y Combinator, and Sony Innovation Fund joined the latest financing—a notable vote of confidence for a company that graduated from Y Combinator's Winter 2025 cohort just over a year earlier.
Its pitch is deceptively simple. While traditional quality assurance programs in customer support typically sample 2% to 3% of interactions—cherry-picking a few calls here, a handful of chats there—Solidroad automates scoring across every conversation. All of them. Phone, live chat, video, email. Whether a human agent or AI bot handled the exchange.
Then it connects those evaluations to personalized training simulations, closing what co-founder and CEO Mark Hughes describes as "the feedback loop that's been broken in customer support for decades."
One client now runs more than 800,000 conversations monthly through the platform. Each one gets scored. According to Hughes, the company has helped customers slash manual review hours by up to 90% while expanding coverage twentyfold. At Podium, a customer engagement platform, the CEO says resolution times dropped 33% after deploying Solidroad's training tools, with new hires hitting performance benchmarks in half the usual onboarding period.
Those are the kinds of numbers that get attention in a market suddenly flush with interest in AI-powered support infrastructure.
An Old Problem, New Tools
Quality assurance in customer service has long been something of a necessary fiction. Teams review a tiny sliver of interactions, extrapolate from that sample, and hope the rest looks similar. It's cost-effective, sure. Comprehensive? Not exactly.
"You're flying blind on 97% of what's happening," Hughes said in a recent interview, framing the problem with the clarity of someone who spent years inside the machinery. He and co-founder Patrick Finlay both worked at Intercom, the customer messaging giant, before launching Solidroad in 2023. Hughes had already built and sold Gradguide, a career platform; Finlay co-founded Monaru, another YC-backed venture.
Their timing intersects with broader momentum in the category. Zendesk acquired Klaus, a QA-focused rival, in early 2024—a signal that incumbents see value in the space. Competitors like Observe.ai and MaestroQA have also drawn investor interest, though the market remains fragmented. Perhaps more fragmented than it should be, given how universal the underlying problem is.
Solidroad counts Ryanair, ŌURA, Crypto.com, and ActiveCampaign among its customers. The platform integrates with major helpdesk software—Zendesk, Intercom, Freshdesk, Gorgias, Talkdesk—making adoption relatively frictionless for teams already embedded in those ecosystems.
What Comes Next
The new capital will fund team expansion across both Dublin and San Francisco, along with continued product development. The company hasn't disclosed its current headcount or post-money valuation, keeping those details close as many early-stage startups do.
Its fundraising trajectory has been brisk. A $6.5 million seed round landed in June 2025 from First Round Capital and Y Combinator, following a $1.2 million pre-seed the previous July. Fourteen months from pre-seed to Series A isn't unheard of in today's market, but it's also not the norm—suggesting either strong traction, compelling unit economics, or both.
Still, scaling from a few high-profile customers to a genuinely large user base presents familiar challenges. Building AI that reliably evaluates conversational quality across industries, languages, and contexts is technically demanding. And while reducing manual review hours sounds appealing, it also requires convincing quality assurance teams to trust algorithmic judgment on something as subjective as customer satisfaction.
Hughes seems unfazed. "The question isn't whether this should be automated," he said. "It's whether you want to keep guessing about 98% of your customer interactions."
Fair point. Though perhaps the tougher question is whether companies are ready to confront what they might find when every conversation gets scrutinized.
