Six Companies, One Bet: AI Can Widen Who Gets Help, Not Just Who Profits From It

Nearly half the world’s population, 4.6 billion people, still lacks access to essential health services, and progress toward universal coverage has stalled since 2015. That figure shaped how we at Horasis approached a task we rarely take on, too. 

Most of our year goes to convening, which means putting business leaders and policymakers in the same room and asking them to disagree productively about where the global economy is heading. So when we decided to name six AI companies to watch in 2026, we held the list to a standard that reflects the problems those leaders bring to us, rather than the funding tables that shape most year-ahead rankings.

Health is only one of those problems. In 2022, 2.1 billion people faced financial hardship paying for care, and 1.6 billion of them were living in poverty or pushed deeper into it by health expenses. 

Financial exclusion follows a similar pattern. The World Bank’s latest Global Findex found that 1.3 billion adults around the world still remain unbanked, and more than half of them live in just eight countries. 

We kept coming back to one detail in that report: of those 1.3 billion adults, nearly 900 million own a phone, including 530 million with a smartphone. The device that could deliver an AI-enabled service is already in the hands of hundreds of millions of people whom the service has never reached.

Readers will notice who is absent from our list. We left off the frontier labs and the chip makers, along with the companies whose main promise is to shrink a workforce faster than their competitors can. Those firms matter, and they already receive plenty of attention, but our question was narrower and, we think, more useful to the people we convene. Who has historically been locked out of a service? And does AI genuinely change that? 

Upfirst AI, Get Covered, Sonata Software, SkyPSI, Effie AI and 360 Health Data each gave us a specific answer. 

For Get Covered specifically, its platform now runs 16 AI agents, chasing missing documents, classifying insurance paperwork, flagging vendor compliance gaps and predicting lapses across more than 3 million rental units. Insurance, as anyone who’s signed for it knows, is tedious work, but precisely the one that decides whether a tenant learns about a lapsed policy before a fire or after one. 

Sonata Software, meanwhile, came to our attention partly through an argument it made in a piece we published in June: that the way evidence is generated and published leaves gaps between what science knows and what the record reflects, and that technology can help close them. We found that point difficult to set aside.

What the six share is that each can point to particular people who now have access to something that was previously out of reach.

That is the measure we believe the leaders in our rooms most need to see applied to AI, and it is the one we intend to keep using.

Horasis AI healthtech company to watch in 2026: The evidence gap between languages

360 Health Data closes the list by working in Spanish, in Colombia, on a problem that rarely makes it into English-language coverage of AI in medicine: most clinical evidence is published in English, and most of Latin America’s clinicians are not. 

Its platform, Coralia Health, and an AI layer called SALUS translate medical research and real-world evidence into a form a Spanish-speaking physician can use at the point of care, including, notably, over WhatsApp, the most popular social media platform in the region, used by 420 million throughout Latin America. 

As operations leader Manuela Gutiérrez put it when the platform launched, the aim is to “strengthen daily medical practice” and reduce the regional inequities that leave Latin American clinicians working with data built for someone else’s health system.

Horasis AI insurtech company to watch in 2026: The insurance form nobody can read

Insurance carries a similar asymmetry, just measured in a different currency: comprehension. Most leases now require renters to carry a minimum policy, yet they’re buying something mandatory that they were never really asked to understand. On the other side of the lease, property managers are stuck proving compliance across a patchwork of carriers, certificates, and inboxes.

The industry’s own AI investment has mostly gone toward underwriting speed and claims automation, tools built for the insurer’s side of the desk.

GetCovered sits in the gap between the lease and the policy. It gives residents a simple way to buy the coverage their community requires, and gives property teams one place to track policies, verify certificates, and work claims, with the adjuster, the manager, and the resident sharing a single case thread instead of three separate inboxes.

Horasis AI company in the legal sector to watch in 2026: Reshaping access to legal services with AI

Every U.S. state bar has now weighed in on AI in legal practice, and the pattern in the data is consistent and a little uncomfortable: adoption tracks firm size almost exactly.

The American Bar Association’s Legal Technology Survey Report found 30% of lawyers using AI-based tools, nearly triple the 11% recorded in 2023, but the gap by size is stark: 46% at firms with 100 or more attorneys, against 18% at solo practices. 

Meanwhile, Thomson Reuters’ own professional services research found something similar: nearly 40% of firms said they’d received contradictory instructions from different clients about whether AI use was even permitted on a matter. 

Solo practitioners, who represent the overwhelming majority of America’s lawyers and the ones most likely to serve people who’ve never had a lawyer before, are the least equipped to sort any of it out.

Upfirst AI, an answering and intake tool built by an attorney rather than a legal-tech vendor, starts at $24.95 a month, against the $35,000-plus annual cost of hiring a receptionist. Its bet is that the thing keeping a solo immigration or family-law practice from taking on clients was never legal judgment, but the missed call at 7 p.m. from someone who needed to talk to a person and got a busy signal instead.

Horasis AI engineering company to watch in 2026:  Speed as its own kind of fairness

Sonata Software is the outlier on the list in that it isn’t chasing an access gap so much as a speed one, though the two turn out to be related. The company, a 25-year Microsoft alliance partner and one of the few outside the Redmond campus to hold both Inner Circle and Frontier Partner status, works with enterprises looking to modernize.

Its CEO, Rajsekhar Datta Roy, put it plainly when the Frontier badge was announced in March: the goal is helping clients “evolve into AI-first organizations” through early investment in Fabric and Azure AI Foundry, instead of waiting for tooling to trickle down once the big players have already captured the advantage.

Velocity, in other words, is itself a kind of equity question.

A company that has to wait two years for enterprise AI infrastructure to become affordable is a company that spends two years losing ground to one that didn’t.

Horasis AI company to watch in 2026: Creating more blue-collar jobs

SkyPSI is the plainest illustration of the group’s whole thesis, partly because its founder has lived both sides of the argument it’s making. Vic Pellicano built and sold two companies before putting $5 million of his own money into a business that trains independent operators to run commercial cleaning drones, then hands them the contracts and the equipment to work for themselves rather than for a franchise. 

“Tech turned a homeless kid from a trailer park into a multimillionaire,” Pellicano has said of his own path. “I’m putting $5 million of my own money into making sure working people get the same shot.”

The framing matters more than the drones do, however. Most of the AI-and-labor conversation this year has run through economists like MIT’s David Autor, who argues that AI’s real opportunity is extending expert-level decision-making to workers who were never given the credential to make those calls themselves. 

SkyPSI is a fairly literal test of that theory, aimed at exactly the tradespeople who’ve had the most reason to distrust every previous wave of automation.

Horasis AI retail company to watch in 2026: Moving decisions to the shelf

Retail runs on a quieter kind of gap than most. The person standing in the aisle is usually the last to learn what headquarters wants and the first to be blamed when the shelf doesn’t match the plan. For decades, merchandisers and field sales reps have photographed shelves one by one and filed reports that head office reads days later, long after anything could be fixed.

effie.ai is trying to move that decision to the aisle itself. Its AI agents gather a store’s requirements from several sources, read the shelf through image recognition, draw up a plan for the visit, guide the rep through it step by step, and then verify the work and report it straight back to headquarters. The company’s newest agent lets a rep scan a shelf in one continuous pass instead of stopping for photo after photo, flagging gaps and confirming fixes as they happen. Nestlé was the first to take the approach public, presenting a joint case study with effie.ai at the Promotion Optimization Institute’s spring summit in Chicago.

The pitch to brands is compliance and sales. What caught our attention is who does the work. The people on the receiving end of these tools are frontline field staff, and in effie.ai’s own client accounts, the benefit they describe is less time spent on reporting and more on the job itself. It is an early, practical test of an argument that runs through this list: AI’s most valuable use may be handing better judgment to the people doing the work, rather than replacing them.

These six companies may not show up on a list of the most valuable AI firms for 2026. And that’s arguably the case we make by naming them all: the more interesting question about AI this year might just not be how big it gets, but who it finally starts letting in.