AI Agent Funding: Suki Lands $90M as Leela and Loops Raise Seed

AI agents that do the work, not just show it

Three AI startups announced funding on October 12, 2026. Read together, they show investors paying for software that finishes tasks instead of software that only presents information. Healthcare AI company Suki raised a $90 million Series E. Financial-compliance startup Leela came out of stealth with a $4.6 million seed round. Commerce-infrastructure company Loops closed a $3.5 million seed.21 The three work in very different markets, but each one pitches its product as an agent that acts on a user's behalf.21

The dollar amounts are small next to the huge rounds that dominated 2026. AI companies took in $102 billion in the third quarter, which was 64% of all venture capital. A record 27 startups raised rounds of $1 billion or more, and those companies absorbed about a third of the quarter's funding.6 Against that, a $90 million Series E and two seed rounds look minor. They matter because they show what investors expect from applied AI below the frontier-model labs: a narrow vertical, measurable workflow results and agents trusted with real tasks.

One detail on the source roundup: it was titled as four funding rounds but described three companies, Suki, Leela and Loops.21 This article covers only those three.

Suki: a late-stage bet on agents in healthcare

Suki's Series E is the biggest of the three. Inc. reported it first as an exclusive.21 The investors include Zoom Ventures, Philips Ventures, Healthier Capital and Hercules Capital. Crosslink Capital is new, and existing backers Venrock, Flare Capital Partners and Hedosophia also put in money.21

Suki did not say what the company is now worth, and that silence is telling. Its $70 million Series D in October 2024 reportedly valued it at close to $500 million post-money, according to Reuters sources.21 Many AI companies have been announcing valuations that double between rounds. Valon Technologies, for example, said its Series D doubled its valuation to $2.3 billion.23 A company that keeps its number private may simply not want to draw attention to it, but that is speculation. What can be said is that investors put $90 million into Suki two years after its last round, at a stage where the price usually reflects revenue more than hope.

The product is a voice assistant for clinicians. It drafts clinical notes, automates billing and other back-office work, and recently added a feature called "Suki Ask" that gives best-practice medical guidance.21 Founder and CEO Punit Soni, who worked at Google and Flipkart before starting Suki in 2017, calls it a Siri for healthcare.21 More than 450 medical institutions use the software, and over 30 health-tech partners build their own products on Suki's platform.21

The new money is meant to move Suki from documentation into action. The company plans to add agent features on top of its "ambient clinical intelligence" so it can carry out follow-up tasks from a patient visit, such as arranging referrals and booking appointments.21 This is a sensible next step. Many companies now sell AI note-taking, so it is getting harder to stand out on transcription alone. Handling what happens after the visit gives Suki more ways to earn revenue and makes it harder for customers to switch.

Leela: compliance as an agent workflow

Leela's seed round was led by Array Ventures and Female Founders Fund. Northwestern Mutual Future Ventures, The Council and several angel investors also took part.21 Founder and CEO Preeti Krishnan used to be a lawyer at Davis Polk and has also run fintech businesses.21 The company sells to financial firms regulated in the United States.21

Published reports disagree slightly on the size of the round. The PYMNTS article gives both $4.8 million and $4.6 million in different places. Other reporting, including Law.com, says $4.6 million, and that appears to be the correct figure.21

The platform turns regulations, internal policies and contracts with sponsor banks into step-by-step workflows. Agents then handle reviews, collect documents and prepare evidence for auditors and regulators.21 Leela stresses that people stay responsible for the important decisions, and that every action an agent takes is recorded so compliance officers can check its work.21 One early customer reportedly turned 40 policies into 262 specific obligations it could act on.21

Leela belongs to a busy category. Valon, which sells AI agents to mortgage servicers, reported more than $200 million in contracted annual recurring revenue within six months of launching.23 Legal AI company Legora raised a $550 million Series D at a $5.55 billion valuation this month.3 Leela's seed round is tiny by comparison. The lesson is that regulated work with heavy paperwork has become one of the most reliable places for agent software to win customers. In these settings, an agent that keeps a clear record of what it did is a feature buyers want.

Loops: commerce as the first test case

San Francisco-based Loops raised its $3.5 million from a16z speedrun, NTT DOCOMO, Connecticut Innovations, Alumni Ventures and Monday Capital.21 The product connects three tools that retailers usually buy separately: search that understands plain-language requests, a shopping assistant that works across channels, and software that rewrites catalog data so customers and AI systems can read it more easily.21 All three share the same product knowledge and customer context. Co-founder Ilker Selim Zorluoglu argues that treating AI as a set of separate features is the biggest mistake companies make.21

The company says it has more than 70 customers and pilots combined. It also says 64% of pilots turned into paid contracts.21 At Turkish supermarket chain A101, sessions that used Loops converted to purchases about 83% more often than a comparison group. Loops notes that this was an observed result, not a randomized test.21 It plans to expand into financial services and energy, and it is building small language models designed for specific business tasks.21

How this fits the October funding wave

Funding news from the rest of the week shows the same direction at a much larger scale. Nous Research raised a $90 million Series B at a $1.5 billion valuation to sell its open-weight Hermes agent to businesses. Nvidia, Microsoft's M12 and Samsung were among the investors.612 AI evaluation startup Arena raised $200 million at a $3.1 billion valuation. Its value has nearly doubled since January, as more companies need ways to test how autonomous agents behave.9 Hone raised a $60 million seed round at a $285 million post-money valuation, only five months after it was founded, to build agents that pursue a business goal over weeks or months.5

The main acquisition story in AI this month is a deal that fell apart. Manus, a Chinese maker of AI agents, raised more than $500 million in a round led by Boyu Capital and IDG Capital after Beijing forced Meta to undo its roughly $2 billion purchase of the company.17 Manus did not disclose its new valuation. Bloomberg had reported that the round would double it to $4 billion.7 Analysts quoted by CNBC said the round showed investors were willing to back Manus as an independent company.7 For founders building agent companies, the case is a reminder that government reviews can now block an acquisition even after it has been announced.

The list of new unicorns is still growing. Germany's RobCo passed a $1 billion valuation through an employee share sale.11 Lithuania's Source, the parent of Oxylabs, said it was valued at $3.6 billion and had a $350 million annual revenue run rate, driven by AI labs that need web data.15 Some market watchers are starting to push back, though, and argue that expectations around AI investment have become unrealistic.19

The takeaway

The October 12 deals point to a clear conclusion. In applied AI, "agent" no longer means a chatbot in front of a database. It means software trusted to finish work in fields where mistakes are costly, including healthcare, financial compliance and retail sales. Suki, Leela and Loops each back their pitch with operating evidence, not just model performance: hundreds of institutions using Suki, policies turned into specific obligations at Leela, and pilots converted to paid contracts at Loops.21

The frontier labs and infrastructure companies will keep the billion-dollar headlines. The steadier part of the market looks like this group: vertical agent companies, funded at prices tied to customer results, raising rounds whose size depends on whether those results hold. Suki's decision to keep its valuation private suggests that even late-stage AI companies now have to show the metrics before they show the price.