The People Who Stay With You

Trace Biosciences

Trace Biosciences just put a $7M signal flare over a problem surgeons have long been forced to navigate with memory, anatomy, and a very expensive kind of caution. Axogen invested the capital on July 28 as part of Trace Biosciences' Series A financing. The public filing also discloses a minority equity stake and a limited right of first refusal. Additional investors are expected, so $7M is not necessarily the final size of the round.

Trace Biosciences is developing Nerve Trace, a near-infrared fluorescence platform designed to make peripheral nerves visible during surgery, including nerves hidden beneath tissue. Surgeons cannot protect what they cannot see, and anatomical knowledge is not the same as real-time visualization.

The company's lead agent, LGW16-03, also known as NerveTrace Dx, has an FDA-cleared IND and is currently being evaluated in a recruiting Phase I study. The single-arm trial plans to enroll up to 38 adults undergoing open-field orthopedic surgery and is assessing safety, pharmacokinetics, dosing, and fluorescence contrast. That is not clinical proof yet. It is the first step in determining whether a promising image can become a practical operating-room tool.

The founding team has been advancing that vision since the technology emerged from Oregon Health & Science University. Co-founders Connor Barth, PhD, CEO, Summer Gibbs, PhD, CSO, and Lei Wang, PhD, CTO, have built a clinical-stage company around a straightforward thesis: invisible anatomy should not remain invisible simply because medicine has learned to work around it.

Axogen is not a random logo on the financing slide. The company operates across peripheral nerve repair and regeneration, with products used after nerves have been damaged or severed. Trace Biosciences is working further upstream, where better visualization could help surgeons avoid injury, identify critical anatomy, or evaluate nerves before repair becomes necessary.

That adjacency is the more interesting transaction. Axogen President and CEO Michael Dale is placing capital alongside a technology that could expand how peripheral nerve care approaches intervention. Repair matters. Seeing the nerve before injury occurs could matter just as much.

The financing is intended to support planned Phase II and Phase III trials leading toward a New Drug Application. The limited right of first refusal gives Axogen future optionality, but it does not announce an acquisition, commercial integration, or guaranteed outcome. Clinical-stage biotech still has to earn each milestone.

What comes next is measurable: Phase I safety and imaging performance, completion of the Series A, later-stage trial design, and evidence that fluorescence contrast can translate into clinical utility. Trace Biosciences made the invisible its business. Axogen made it much harder for the market to overlook.

Biota

Biota, Inc. is not announcing a $3M seed round. It is showing investors what the company plans to finance before pretending the wire has reached the bank. The Longmont environmental diagnostics startup is seeking a $3M seed round and will present at the 2026 Rocky Mountain Life Sciences Investor and Partnering Conference. "Funding sought" is a live opportunity. "Funding raised" is a completed transaction.

CEO and Founder Rose Nash, PhD, is building Biota around a simple problem with expensive consequences. PFAS cleanup cannot move faster than the testing behind it, and conventional workflows can leave treatment teams waiting days or weeks for results while equipment, people, and decisions sit in neutral.

Biota addresses that delay from both directions. PFAS RapidTest is a field-compatible screening system that the company says can deliver quantitative results in approximately 15 minutes. MagnaPrep uses magnetic nanoparticles to streamline laboratory sample preparation. One brings faster intelligence into the field. The other moves more samples through the laboratory.

That is a sharper value proposition than simply saying, "we built another sensor." Biota is selling time back to laboratories, remediation firms, utilities, and operators that cannot make treatment decisions until the chemistry says yes. The company is building the measurement layer for the organizations cleaning up PFAS.

The foundation already extends beyond a pitch deck. Biota opened a 7,500-square-foot facility in Longmont in June. Colorado OEDIT awarded the company a $250,000 Advanced Industries grant for its PFAS extraction system, and the Colorado BioScience Association lists a $305,000 NIEHS SBIR Phase I award among Biota's recent milestones.

Antler and Mazarine Ventures publicly list Biota in their portfolios. Rose Nash has also identified Zymo Research, Unpopular Ventures, and Aerion Capital among the company's investors. That mix reflects the nature of the opportunity, where regulation, laboratory operations, public health, infrastructure, and liability all converge around the same sample bottle.

The EPA is preserving enforceable PFOA and PFOS limits at 4 parts per trillion. It has also announced nearly $1B to help states address PFAS in drinking water. More investment in cleanup creates greater demand for measurement technologies that can keep pace.

Biota still has work to prove. Both products are listed as coming soon, independent performance data has not been made public, and the current seed round has not been verified as closed. That is not a reason to dismiss the company. It is the real investment question. Can Rose Nash and the Biota team turn faster PFAS testing from a compelling laboratory thesis into the operating infrastructure of a growing cleanup economy?

LearnVector

Coursera just put $100M behind LearnVector, the new AI-native learning company founded and led by Andrew Ng. In return, Coursera receives a one-third ownership interest on a fully diluted basis. This is a strategic equity bet on what happens when AI stops acting like an answer machine and starts behaving like a tutor.

LearnVector is barely out of the gate. The Mountain View company was founded in 2026, has no public product yet, and is targeting its first learning experiences for early 2027. Its goal is to build agentic systems that plan a personalized learning path, adapt to each learner, practice alongside them, and stay engaged until they can demonstrate mastery.

That distinction matters. A chatbot can help someone finish an assignment more quickly without making them better at the underlying skill. Research cited by LearnVector found that unguarded AI assistance improved practice performance but reduced later unassisted performance, while tutoring guardrails mitigated that decline. Helping the learner own the answer is the more difficult business.

Andrew Ng has been working toward this problem for years. He co-founded Coursera, led Google Brain, built DeepLearning.AI, and continues to serve as Coursera's Board Chair. LearnVector extends that trajectory toward a model where software adapts to the learner instead of forcing every learner through the same sequence.

Coursera contributes trusted content and distribution. The combined Coursera-@Udemy ecosystem reports more than 300M learners and 12,000 enterprise customers. The companies are exploring commercial collaboration, but the products and partnerships still have to be built. Distribution opens the door. It does not prove the learning model works.

There is also a governance story behind the investment. Andrew Ng founded LearnVector while remaining Chair of Coursera, so an independent and disinterested special committee evaluated, negotiated, and unanimously approved the related-party transaction. That does not eliminate execution risk, but it demonstrates that the governance process was addressed.

The startup is hiring across AI engineering, learning science, full-stack development, and operations. No official valuation, additional investors, customers, revenue metrics, model providers, or broader executive team have been announced. This is a well-funded thesis before it is a proven product.

Coursera is betting that the next generation of education platforms will win by helping people develop skills they can still apply after the machine steps away. If LearnVector can prove that difference, the traditional boundaries between course platforms, tutors, and workforce infrastructure may become much harder to define.

PEX

PEX just secured $160M in debt and equity financing, and the structure says more than the headline. Bluff Point Associates led the financing, while Clear Haven Capital Management provided the credit facility supporting PEX's charge-card expansion. That division of capital gives software growth and credit capacity distinct roles, which matters when a card program has to support increasing transaction volume.

The timing reflects operational momentum. PEX says its charge-card business has sustained triple-digit growth and that its platform has processed more than $11.7B in spend since inception. The company also reports serving more than 10,000 organizations and issuing more than 10M cards. Those are company-reported figures, but they point to something meaningful: PEX is not arriving with a deck and a dream. It has been building corporate payment infrastructure since 2006.

The product thesis is difficult to execute. Growing businesses do not need a card in one tab, receipts in another, policy enforcement in somebody's inbox, and reconciliation waiting at month-end like an unpaid bar tab. PEX combines charge, prepaid, virtual, and disbursement cards with spend controls, approvals, receipt capture, reporting, and accounting integrations. The objective is not more finance software. It is less financial friction.

Founder and CEO Toffer Grant has spent two decades in commercial payments, and CTO Alexander Najem leads the technology behind a platform built for control, compliance, and secure spending. The new capital gives the team room to expand the charge-card program, increase transaction capacity, and build more AI-powered automation for routine finance operations.

That last point is where the market becomes especially interesting. AI in finance is easy to announce and even easier to get wrong. Finance teams do not need a chatbot wearing a CFO costume. They need accurate data, enforceable policies, auditable workflows, and fewer hours lost to receipts, coding, and exceptions. If PEX can make that automation reliable enough to disappear into everyday workflows, the feature becomes infrastructure.

There is also a signal in the financing mix. Corporate cards sit at the intersection of software, underwriting, and capital. Scaling them responsibly requires credit capacity, disciplined risk management, and a product customers continue using long after the initial demo. Bluff Point's long-standing relationship with PEX and Clear Haven's credit facility give this financing practical shape. The next test is not whether $160M attracts attention. It is whether PEX can turn increasing financial complexity into fewer operational decisions for the businesses doing the spending.

Spur Intelligence

Spur Intelligence secured a $200M investment from Insight Partners, and the size of the check is only half the story. The other half is what Insight is betting the internet can no longer ignore: an IP address is not an identity.

Residential proxies, VPNs, bot networks, and AI-driven infrastructure allow attackers to walk through digital front doors looking like ordinary users. Traditional IP reputation can leave security and fraud teams staring at a familiar address with little context about who is actually behind it.

That is Spur Intelligence's opportunity. The company transforms obscured network activity into actionable intelligence through APIs, on-premises data feeds, session enrichment, and integrations for authentication, fraud scoring, investigations, and real-time enforcement. It provides context between a connection and a decision.

Spur Intelligence's 2026 study of more than 200 security practitioners found that 94% of surveyed organizations encountered anonymizing VPN or residential proxy infrastructure during security incidents. Nearly half planned to implement, replace, or upgrade IP intelligence within the following 12 months.

The company also reported 210% year-over-year new-business growth, 42% ARR growth in Q2 2026, 109% net revenue retention, and 100% renewal retention across scheduled renewals. Those company-reported metrics help explain why a software investor with Insight Partners' cybersecurity track record paid attention.

The capital will support product development, intelligence coverage, integrations, and enterprise operations. No financing series, valuation, or additional investors were disclosed. A $200M headline does not need imaginary furniture added to the room.

CEO Kevin Hickey now has additional capital to move Spur Intelligence from enrichment toward infrastructure for live security and fraud decisions. CTO Alastair Parr leads the technical execution of that strategy. Co-founders Ethan Smith and Riley Kilmer built the company around a blunt reality: attackers borrow legitimacy from the networks they hide behind.

For operators, the lesson is not to buy more data. Security stacks already generate enough signals to ruin a perfectly good morning. The opportunity is to make those signals explainable and actionable early enough to stop fraud without penalizing legitimate users. That is where IP intelligence becomes an operating layer.

Insight Partners is backing the shift from post-incident enrichment to real-time decision-making across authentication, fraud, and edge security. The next chapter is not about seeing more traffic. It is about seeing through it.

Jump Capital

Jump Capital just closed its 8th fund at $350M, and the more interesting number is not $350M. It is eight. A repeatable venture franchise does not get eight chances by buying whatever happens to be loud on the internet. It survives market cycles by knowing which changes are structural, which are costume jewelry, and which founders can turn a technical shift into a business customers will continue paying for.

This fund is aimed at expanding Jump Capital's early-stage AI investing across applications, infrastructure, and cybersecurity. That framing matters. AI is no longer a single category. It is a pressure wave moving through every layer of software: the products users interact with, the infrastructure beneath them, and the security controls trying to keep autonomous systems from wandering through the building with the master key.

Jump Capital is also focused on legacy industries, where decades of manual work and operational complexity kept better software from gaining traction. That is where AI becomes interesting. Not another feature designed to win a demo, but a workflow that compresses a week into an hour, makes scarce expertise repeatable, or gives an established market a reason to buy software it rejected five years ago.

The firm has already been building around that thesis. Its recent writing and portfolio span enterprise applications, AI infrastructure, data tooling, and cybersecurity. Standard Kernel is improving GPU efficiency. Prompt Security focuses on enterprise GenAI risk. Compa applies AI to compensation intelligence. Different products, same question: does the technology survive contact with a real customer's operating environment?

The market backdrop is both generous and unforgiving. AI captured more than 65% of 2025 venture investment, according to NVCA, while fundraising became increasingly concentrated among established managers. Capital is abundant at the headline level and highly selective everywhere else. Founders do not need another tourist with a checkbook. They need investors who can distinguish technical novelty from durable distribution, enterprise curiosity from committed budgets, and a clever model from a sustainable company.

That is the bet Jump Capital is making with Fund VIII. Co-founders Sach Chitnis and Mike McMahon built a Chicago firm around operating experience and early-stage conviction. The next chapter is less about declaring that AI will reshape every industry and more about identifying the founders who know exactly which workflow to change first. $350M can finance a lot of experiments. The real returns will come from knowing which ones become infrastructure.

Vivid Dx

Vivid Dx did not raise a single $15M seed round. The cleaner story is more interesting: founder and CTO Jeff Liu's Y Combinator profile says the Oxford diagnostics spinout raised $15M across a $5M seed round and a $10M pre-Series A. That distinction matters. Capital history should not be flattened to fit a tidier headline.

Oxford Science Enterprises is the one investor Vivid Dx has publicly identified. The company says OSE committed £4M in April 2024 to support its Raman diagnostics platform. Vivid Dx has not disclosed the date, full investor syndicate, valuation, or financing terms for the reported $10M pre-Series A, so those details remain appropriately undisclosed.

What investors are backing is not another dashboard wearing a lab coat. Vivid Dx is working to eliminate blood culture as the starting point for infection diagnosis. Its platform is designed to isolate pathogens directly from blood, analyze their Raman fingerprints, and use a proprietary spectral database with AI to classify infections.

The company's target is broad pathogen identification in 30 minutes and phenotypic antimicrobial susceptibility testing in three hours. Blood culture workflows can take days, forcing clinicians to begin treatment with broad-spectrum antibiotics while they wait. During that interval, the patient is not waiting, and neither is antimicrobial resistance.

The scale of the challenge is severe. WHO and UN estimate that bacterial antimicrobial resistance directly caused 1.27M deaths and was associated with 4.95M deaths in 2019. Faster diagnostics alone will not solve AMR, but earlier therapy-guiding information could help clinicians target treatment sooner while reducing unnecessary broad-spectrum antibiotic exposure.

Vivid Dx reports 99% pathogen-identification accuracy in laboratory studies and 92% concordance with gold-standard AST across 64,000 single-cell spectra. Promising is the operative word. The harder work now is clinical validation, regulatory evidence, workflow integration, and demonstrating the platform's performance outside a controlled environment.

That is why the reported funding matters. Vivid Dx says its next phase includes expanding its scientific, engineering, and microfluidics capabilities, completing an integrated proof of concept, and advancing clinical validation. Those are company plans rather than a disclosed use-of-proceeds schedule, but they identify where execution will be measured.

Jeff Liu and Mike Risley, Chief Operating and Product Officer, are advancing an Oxford-born technical thesis toward clinical practice. The more important question is whether this capital can turn a faster laboratory signal into a dependable clinical decision, because that is where the real diagnostic value begins.

Ellis

Ellis just emerged from stealth with a $10M seed round and a clear read on one of finance's least glamorous, most expensive problems: private credit grew into a massive asset class while the machinery underneath it remained scattered across spreadsheets, fund administrators, ledgers, loan systems, portfolio tools, and documents.

When a fund team spends days rebuilding the same operating picture, every reconciliation becomes a scavenger hunt, every report introduces manual risk, and every decision starts with a debate over which number deserves to be trusted.

Ryan A. Williams knows that view from the inside. After founding Cadre and spending a decade modernizing access to private markets, Williams began building Ellis in 2025 to tackle the operating layer the industry left behind.

Ellis connects existing systems and reconciles their data into a single, source-verifiable book of record. Its AI agents assist with close, reporting, cash forecasting, scenario analysis, and portfolio monitoring, while material decisions remain with accountable human professionals.

That distinction matters. Credit is not a sandbox where a confident answer earns points for presentation. Every number has to trace back to the loan tape, the ledger, the fund administrator, or the document that created it.

Ellis says it developed the platform alongside private credit managers representing more than $50B in assets under management. That is company-reported, but it is still a meaningful design signal: the product is being shaped around live operating pain rather than a demo looking for a budget.

The seed round includes First Round Capital, 645 Ventures, Harlem Capital, Khosla Ventures, Thrive Capital, Slow Capital, Kearny Jackson, and Mellody Hobson. No lead investor or valuation was disclosed, which keeps the useful part of the announcement where it belongs: the operating thesis.

Private credit is measured in trillions, yet regulators continue to point to data gaps and limited loan-level visibility. The asset class scaled faster than its information layer, and reconciliation can no longer depend on whichever associate knows the sacred workbook.

The bet behind Ellis is that institutional quality should describe how a manager operates, not how much infrastructure it can afford. If the platform can turn fragmented inputs into evidence-backed answers without replacing the systems firms already trust, the value extends beyond faster reporting. It is operating leverage with a paper trail. The sharper signal is what investors are funding: an attempt to rebuild the private-credit control room before complexity outruns the people responsible for keeping it honest.

Perceptron Network

Perceptron Network just closed a $6.5M strategic round to expand a decentralized AI data network built around a stubborn idea: the people producing useful data should not be treated like invisible plumbing.

The company is building a mesh of distributed nodes, community agents, and human data tasks that AI companies can use to commission specialized datasets. Perceptron says the network now spans more than 800,000 nodes and supports more than 300,000 daily active users. Those are company-reported figures, but they describe the scale of the bet clearly enough.

Most conversations about AI infrastructure begin with chips and models. Perceptron is working one layer earlier, where real-world inputs are collected, validated, and transformed into something a model can actually use. Its new data-questing platform is designed to let companies request high-value datasets directly from contributors while allowing those contributors to retain ownership of their data and participate in the economics.

That changes the operating question. The challenge is no longer whether a decentralized network can gather information cheaply. It is whether that network can repeatedly deliver useful, verified datasets quickly enough for serious AI buyers without turning quality control into a trust exercise wearing a token costume.

Co-founder and CEO Peter Anthony is moving Perceptron from passive node participation toward commissioned human intelligence. The financing will support the platform launch, contributor tooling, rewards infrastructure, and the company's stated path toward 5M nodes.

The investor group reflects the crossover Perceptron is trying to own: Sigma Capital, Selini Capital, QCP Capital, P2 Ventures, CoinDCX Ventures, Momentum6, DeFi Capital, Walrus Foundation, Aethir, Colosseum, GuruDev Capital, Tempo Finance, NewTribe Capital, Digital Consensus Fund, and CodeCraft Capital. No lead investor was disclosed, which matters because a long wall of logos is not the same as a single institution setting the terms.

The business lesson is sharper than the fundraising headline. Distribution attracts attention, but repeatable quality wins contracts. Perceptron already has the contributor footprint. Now it has to prove that human-authored data can become reliable infrastructure rather than a large community waiting for instructions. The next milestone is not another node count. It is demonstrating that a distributed network can deliver data with the speed, specificity, and trust that AI companies will pay for repeatedly.

Discern Security

Discern Security just raised $13M in Series A funding, and the sharper story is not another cybersecurity company promising to find more problems. Security teams already have enough alerts to qualify as a weather system.

The round was led by Forgepoint Capital, with participation from First Rays Venture Partners, Growth Enjin Partners, BoldCap, WestWave Capital, Vela Partners, and other angel investors. The capital will help Discern grow its engineering and product teams, expand its library of AI Skills and autonomous capabilities, and accelerate adoption and go-to-market execution.

Discern is building around an uncomfortable truth: companies have spent heavily on security products, yet much of that licensed capability still sits unused. Discern reports that its customers use less than 25% of the features available in the security products they already own. Buying another dashboard does not make the existing dashboards more useful.

That is where the Discern Security Loop gets interesting. The platform unifies data across an organization's environment, identifies control gaps, prioritizes remediation using business context, and coordinates action through intelligent agents and human-approved workflows. The goal is not AI theater. It is a closed loop that moves from identifying an issue to proving the security posture actually improved.

Discern is betting the next control plane will connect those systems while keeping humans at the decision points that matter. Its AI Skills handle repeatable analysis. Its agents move work forward. Its approval model keeps automation accountable. That combination is more compelling than pretending autonomy means removing judgment. It is also a useful standard for any security AI platform.

Founder and CEO Sai Venkataraman says Discern's agentic loops can at least double tool utilization for customers. That is a company claim, but it highlights the commercial logic behind the round: the fastest path to better security may be extracting more value from the stack organizations already own.

The founding team knows the territory. Sai Venkataraman, Santhosh Purathepparambil, and Rohan Puri built Discern around the gap between security posture on paper and security controls in practice. Santhosh Purathepparambil also serves as Chief Product Officer, giving the company a technical leadership bench capable of translating that thesis into product. The $13M matters, but the more interesting test starts now: can the Discern Security Loop turn security software shelfware into operating leverage at enterprise scale?

Nexus Power

Nexus Power just landed a strategic investment from Madison River Capital, and the most important word in that sentence is not "investment." It is power. The electrical infrastructure market has plenty of equipment. What it lacks is coordination. Data centers, utilities, semiconductor plants, mines, and industrial operators need design, engineering, procurement, and delivery to move as one system. When those pieces drift apart, timelines stretch, and expensive equipment waits on expensive decisions.

That is the lane Nexus Power has built. Formed in 2022 through the merger of two established manufacturer representatives, the Salt Lake City company serves customers across 11 Western and Midwestern states. Its model combines exclusive manufacturer relationships with embedded electrical design engineering and project coordination.

The depth matters. Nexus Power is ABB's exclusive manufacturer representative across its territory, extending a relationship that spans more than 20 years. It also works with Hitachi and approximately 20 other electrical equipment manufacturers across five divisions.

Madison River Capital is investing in that operating position, not a slide-deck promise. The firm's support is expected to help Nexus Power expand geographically, pursue targeted design-engineering acquisitions, broaden manufacturer partnerships, and grow its team. Terms were not disclosed, which is the right place to stop counting and look at the strategy.

Grid modernization, AI infrastructure, data center construction, and reshored manufacturing are turning power availability into a board-level issue. The opportunity is not simply selling more switchgear. It is helping customers specify systems, source equipment, manage delivery, and keep complex projects from becoming a relay race where everyone drops the baton.

Pete Jones, President and CEO of Nexus Power, points to manufacturer relationships, technical depth, and end-to-end execution as the company's advantage. Rich Haag, Co-founder and Board Member, remains involved. David Wittels and Matthew Clancy of Madison River Capital are backing a team that built market coverage before asking capital to accelerate it.

The operator lesson is straightforward. Distribution is not old economy when expertise is scarce. Representation is not a commodity when the representative can engineer the answer. Growth capital becomes more compelling when it is attached to a business solving the coordination problem inside a market moving this quickly. That is the signal worth watching.

Antora Energy

Antora Energy just closed a $550M Series C, and this is what climate infrastructure looks like when it stops asking permission to be taken seriously. G2 Venture Partners and Eclipse co-led the oversubscribed round. Ribbit Capital, Salesforce Ventures, Activate Capital, John Doerr, Westly Group, StepStone Group, and Liberty Mutual Strategic Ventures joined as new investors. Decarbonization Partners, Impact Science Ventures, Trust Ventures, Breakthrough Energy Ventures, and Lowercarbon Capital came back for more.

At POET's bioprocessing facility in South Dakota, Antora deployed a 5 GWh thermal battery system that moved from initial construction to delivering energy in less than 12 months. More than 200 factory-built modules now turn low-cost electricity into stored heat for an industrial operation that cannot pause because the sun went down or the grid got moody.

The technology is almost offensively practical. Antora heats blocks of solid carbon when electricity is cheap and abundant, stores that energy at extreme temperatures, then delivers it around the clock as heat or power. There is no scavenger hunt for supply-constrained critical minerals. The same modular architecture can serve a chemical plant, food producer, steelmaker, data center, or the grid.

That matters because the next industrial race is not only about who can build the smartest factory or the biggest data center. It is about who can bring reliable energy online quickly enough to operate either one. Power has become a deployment constraint, a capital constraint, and increasingly, a competitive constraint.

Antora's answer is to manufacture the infrastructure. Its San Jose operation expanded into a three-building campus this year. The Series C will fund additional large-scale projects, expanded production, a second U.S. manufacturing hub, and a stronger domestic supply chain. Antora says its total corporate and project financing now stands at approximately $1B.

Co-founder and CEO Andrew Ponec is framing the moment around speed and scale. Co-founders David Bierman, CCO, and Justin Briggs, COO, have spent years turning thermal storage from a clever materials story into equipment that can be built, shipped, and commissioned the way industry actually needs it.

The question after a round this large is no longer whether thermal batteries can earn attention. It is how quickly Antora can make repeatable deployment feel ordinary, because that is when infrastructure starts changing markets.

SoundHealth

SoundHealth just raised a $12.25M Series A after moving SONU from an ambitious idea into an FDA-cleared wearable with repeat-use data. In a medtech market that has become allergic to fuzzy commercialization stories, that sequence matters.

SONU uses personalized acoustic vibration to relieve moderate to severe nasal congestion associated with rhinitis. A smartphone scan maps facial anatomy, the system calculates individualized resonance frequencies, and the wearable delivers therapy without a spray or pill. The FDA granted De Novo classification in 2023, followed by a 510(k) clearance in 2025 supporting the current product for people ages 12 and older.

The company-reported usage gives the round some weight. SoundHealth says it has delivered more than 500,000 treatments to 15,000 users, increased monthly new-user acquisition by more than 15x over 24 months, and saw 3,600 of 4,000 monthly active users complete treatments in April 2026. Patients average 15 sessions per month. Those figures point to the behavior every consumer medical device has to earn: people returning after the novelty wears off.

Founder and CEO Paramesh Gopi built the company from a personal struggle with lifelong allergies, then assembled a team that does not need to cosplay as clinical credibility. Co-founders Bryant Lin and Peter Hwang lead clinical affairs, product design, and research, while Vivek Mohan leads product development and engineering. That combination of operator, physician, and technical expertise is doing real work here.

SoundHealth announced a $7M Seed round in 2024 led by Moai Capital and J4 Ventures, with TeleSoft Partners, Tau Ventures, TechU Ventures, and Rhythm Venture Capital participating. Add the new Series A, and the company has disclosed $19.25M across those two rounds. The Series A investors were not identified in accessible public disclosures, so the cleaner signal is the stage of the business they are backing: regulated hardware entering commercialization, not a slide deck asking everyone to imagine product-market fit.

Now comes the expensive part. SoundHealth has to turn regulatory progress and user engagement into durable distribution, manufacturing scale, clinical trust, and consumer demand. That is where medtech stories stop being clever and start becoming companies. The company found a literal use for sound capital. The next question is whether personalized acoustic therapy can become a category patients reach for before the medicine cabinet.

Smallest.ai

Smallest.ai just raised a $13M Series A to make voice AI feel less like waiting on a machine and more like speaking with someone already following the conversation. Seligman Ventures led the round, with Sierra Ventures and 3one4 Capital participating, bringing the company's total disclosed funding to more than $21M.

The architecture matters as much as the amount. Most voice agents still run a relay race: speech becomes text, text goes to a large model, the answer comes back, and another system turns it into speech again. Every handoff adds delay, and people notice that silence immediately.

Smallest.ai is taking a different route. Its compact model is designed to listen, think, and speak simultaneously, closer to the rhythm of a real conversation. For routine, domain-specific exchanges, it responds in real time. When a question falls outside that lane, the system can hand the task to a larger foundation model and return with the answer.

That is a cleaner approach than asking one giant model to be fast, fluent, multilingual, and perfect in a noisy call center. Smallest.ai is betting on a focused model for live conversations and a larger model for more demanding reasoning.

The company says its technology has powered more than 1B minutes of voice AI across more than 10 enterprises. TechCrunch identified RingCentral and Truecaller among its customers. The usage figures are company-reported, but the operating signal is clear: this platform is built for production workloads, not a polished demo that falls apart when someone interrupts.

Co-founder and CEO Sudarshan Kamath wants people to speak with the model without knowing whether it is AI or human. Co-founder and CTO Akshat Mandloi is building the compact speech architecture behind that goal, where accents, background noise, language, and response timing are core engineering constraints.

The round also shows how conviction compounds. Sierra Ventures led Smallest.ai's $8M Seed round in 2025 and returned for the Series A alongside 3one4 Capital. New lead investor Seligman Ventures is backing a vertically integrated infrastructure thesis rather than another customer-service wrapper with a microphone attached.

Smallest.ai says the capital will accelerate model development, real-time infrastructure, and enterprise deployments. That puts the money where this market becomes difficult: model quality, inference speed, deployment control, and the uncomfortable reality that a half-second pause can expose an entire AI experience. Voice may be becoming a primary interface, but the sharper idea is treating conversation itself as infrastructure and asking whether smaller models can make the experience feel much bigger.

Offscript Worldwide

Offscript Worldwide just took a strategic minority investment from ESSENCE Ventures, and the most important part of the deal is not a mystery dollar figure. It is the architecture.

ESSENCE Ventures now holds a minority stake in Revolt New Media Holdings LLC. Sundial Media & Technology Group founder and Chairman Richelieu Dennis gains a board seat. Offscript Worldwide and REVOLT remain independently operated under CEO Detavio Samuels. That last point matters. This is not a merger dressed up in partnership language. It is a governance and commercial link between two Black-owned media ecosystems that already know how to turn community trust into attention, experiences, and business.

On one side sits Offscript Worldwide: REVOLT, REVOLT Sports, Rap-Up, 440 Artists, 3BLACKDOT, Offscript Creator Network, and Six Zeros. On the other sits a Sundial and ESSENCE portfolio that includes ESSENCE, Girls United, Refinery29, AFROPUNK, and Beautycon.

The portfolios serve different roles, but the commercial thesis is straightforward. Give advertisers one point of access to audiences moving through entertainment, music, women's lifestyle, beauty, creators, sports, gaming, technology, and live experiences. That is a sharper answer to modern media fragmentation than asking every brand to assemble the culture puzzle one publisher at a time. Reach still matters. Context matters more. Trust is the variable many media-buying spreadsheets cannot calculate until it is missing.

The announcement says the combined portfolios reach millions across owned media, social, creator, video, and live experiences. It also cites approximately 500,000 attendees at the 2026 ESSENCE Festival of Culture. Those company-reported figures explain the investment thesis: screens are crowded, while credible communities remain scarce.

Offscript Worldwide was built around creator ownership. Creators keep their platforms and intellectual property while gaining infrastructure, distribution, and brand partnerships. ESSENCE Ventures brings an operating portfolio with decades of audience equity and a history of building where culture, commerce, and community intersect.

Financial terms were not disclosed, so there is no honest valuation theater to perform here. The market signal is the structure itself: minority capital, board-level alignment, independent operations, and a shared push for larger advertiser relationships. The more interesting question is whether this alliance can make cultural fluency easier to buy without sanding down the independence that made these brands valuable in the first place.

Pantheon

Pantheon just closed its largest dedicated co-investment program at $3.2B. Pantheon Global Co-Investment Opportunities Fund VI, or PGCO VI, is not a pool built to spread capital across every polished logo. The strategy is narrower, and that is the point.

The fund will back mid-market companies alongside general partners Pantheon believes are best positioned to create value in specific sectors. The stated focus is non-bank financials, industrials and business services, and technology. That makes manager selection, access, and underwriting the product, not decorative language around it.

Scale matters because the prior program, PGCO V, closed at more than $2.4B in 2023. PGCO VI adds at least $800M to that benchmark, an increase of roughly one-third. In private markets, bigger does not automatically mean better. It does mean investors are giving Pantheon more capital to prove that selectivity can survive scale.

Pantheon says the close follows a record 2025, when the firm deployed approximately $1.3B across 30 co-investment deals through a platform that reports $41B in private equity assets. The dedicated co-investment strategy began in 2009, decades after founder Rhoddy Swire established Pantheon in London in 1982.

The capital base is also worth examining. Pension funds remain the largest source of commitments, while sovereign wealth funds, insurers, asset managers, family offices, and endowments broaden the investor mix. Pantheon says Asia delivered the strongest growth in its investor base.

That is a useful signal. Co-investment is increasingly becoming a portfolio-construction tool rather than a sidecar attached to a broader fund relationship. It can offer more direct exposure, but only if the manager has enough deal flow to reject most of what crosses the desk.

Jeff Miller, Pantheon's CIO and Global Head of Private Equity, pointed to the firm's relationships, proprietary sourcing, and ability to underwrite mid-market opportunities across market cycles. Florence Dard, Pantheon's Chief Client Officer, emphasized the diversification and diligence investors expect from the program.

The real test starts after the fund closes. A $3.2B vehicle creates deployment pressure, and disciplined portfolio construction gets harder as the checkbook gets heavier. Pantheon says it will diversify by manager, sector, and vintage while keeping the strategy centered on the mid-market. The job now is to make $3.2B behave like conviction, not inventory.