Belief Has to Survive the Work

Cogensus

Aging is creating one of the largest healthcare challenges of our time, and Cogensus is betting that better conversations can become part of the solution. This week, the Sheridan, Wyoming-based AI health technology company announced a follow-on investment from Caduceus Capital Partners, building on its participation in the Caduceus LAUNCH Spring 2025 cohort. The news arrives as Founder and CEO Matthew Edgerton, Co-Founder and CCO Raju Joshi, and Co-Founder and COO Hall Wang continue building a platform focused on cognitive health, social connection, and legacy preservation for older adults.

Healthcare systems spend enormous resources responding to cognitive decline after warning signs appear. Cogensus is focused on the period before intervention becomes more difficult. Loneliness and social isolation are increasingly connected to broader health outcomes.

The platform combines generative AI, machine learning, and analytics to support older adults through an AI companion named Joe while helping providers monitor cognitive and mood-related signals over time. Families gain tools to preserve stories, memories, and connections that might otherwise disappear.

The company's mission is rooted in a personal experience. Matthew Edgerton's discovery of the largely untold story of his grandfather, Milton Sawitz, a decorated war veteran, highlighted how easily important family histories can be lost. That realization became part of the foundation for Cogensus.

The leadership team brings expertise from multiple disciplines. Raju Joshi contributes more than 30 years of experience across biotechnology, medical devices, digital health, and healthcare operations. Hall Wang brings experience spanning product strategy, operations, and commercialization. Ralf Brockhaus, CTO, leads technology development, while Tom Robershaw, Head of Finance, supports the company's financial strategy and growth.

The follow-on investment comes as Cogensus expands partnerships with Insight Senior Living and Integrated Senior while progressing toward commercial availability in Q3 2026. As populations age and healthcare systems search for scalable ways to support cognitive wellness, the company is positioning itself where healthcare, technology, and human connection meet.

Strategic Advisor Dr. Gary Small is widely recognized for his work in brain aging and dementia research, while Clinical Advisor Dr. Robert Bilder brings decades of experience in neuropsychology and cognitive assessment. Their involvement reinforces the company's focus on combining clinical insight with technology-driven solutions.

The company name itself points to the mission. Cogensus brings together ideas of cognition, connection, and understanding, reflecting a platform designed to help older adults remain engaged, give providers better visibility, and preserve stories that might otherwise be lost.

Leal Therapeutics

Leal Therapeutics put $30M behind a thesis that is easy to say and brutally hard to prove: stubborn brain disorders may be attacked by correcting metabolic imbalances inside the central nervous system.

The Worcester biotech announced its Series A on August 27, 2025. SV Health Investors’ Dementia Discovery Fund led the round, with OrbiMed, Newpath Partners, Chugai Venture Fund, Euclidean Capital, Alexandria Venture Investments, and PhiFund participating. The announcement supports a $30M Series A, but not the card’s second close label.

The capital was assigned to specific clinical jobs. LTX-001 is a brain-penetrant oral small molecule that inhibits GLS1 and is being developed for schizophrenia. LTX-002 is an intrathecal antisense oligonucleotide targeting SPTLC1 for genetic and sporadic ALS. One goes after excessive glutamate. The other targets dysregulated sphingolipid metabolism. Different tools, same wager: brain biology is also chemistry with consequences.

CNS drug development has punished confidence for decades because promising mechanisms often dissolve between elegant preclinical data and human outcomes. Leal is not selling immunity from that reality. It is trying to make the bet more measurable through human genetics, biomarker data, target engagement, and defined metabolic pathways.

Asa Abeliovich, MD, PhD, founded Leal after building Prevail Therapeutics, which Eli Lilly acquired in 2021, and co-founding Alector. Eduardo Paredes, PhD, MBA, serves as CTO, bringing oligonucleotide and CMC experience. The executive bench also includes CMO Johannes Tauscher, MD, and CBO Raymond Jordt, MBA, MS. This is not a team learning CNS development from a slide deck.

The financing also marked a board-level commitment. Christian Jung, PhD, a partner at SV Health Investors and co-leader of DDF, joined Leal’s board. DDF is not a tourist in neurodegeneration. Its mandate is specialized investment in companies developing or enabling novel dementia therapeutics.

The real scorecard is already clinical. Leal reported supportive Phase 1 safety, tolerability, and target-engagement findings for LTX-001 in April 2026. In June 2026, the company dosed the first participant in the Phase 1/2 NeurALS study of LTX-002. Funding headlines are snapshots. Those milestones show what the money was supposed to become.

Leal Therapeutics, Asa Abeliovich, Eduardo Paredes, DDF, and the investor syndicate now have a more consequential scorecard ahead. The $30M matters, but the sharper signal is disciplined capital moving into medicine where easy stories were never the problem. Evidence was.

Xpander

Xpander just raised $7.5M in Seed funding, led by PICO Venture Partners, and the real story is not another pile of capital chasing another AI demo. It is a bet on the layer enterprises discover they need right after the demo works.

Building an agent is getting easier. Letting hundreds of them touch real systems, carry real permissions, make real decisions, and survive real audits is where the invoice for all that enthusiasm finally arrives.

Xpander is positioning itself as the control plane for that moment. The platform lets companies build, run, and govern agents across different models, frameworks, and environments while keeping identity, permissions, approvals, observability, and action logs in one operating layer.

Enterprise AI is not heading toward one model, one framework, or one cloud. It is heading toward a mixed fleet assembled by engineering, security, operations, finance, and every ambitious department that found an API key.

Someone still has to know which agent can reach which system, whose authority it is borrowing, what it changed, what it cost, and whether a human was supposed to approve the action first. The model can be brilliant while the operating environment remains chaos wearing a dashboard.

Xpander's answer is vendor neutrality with grown-up controls. Its Universal Harness is designed to support different models and frameworks, while deployment can stay in a public cloud, a VPC, on-premises, or even an air-gapped environment. The company also offers Omni, an agentic forward-deployed engineer that helps teams build and operate agents without handing the entire pattern to outside services.

The business lesson inside the round is that AI infrastructure is shifting from "Can we build an agent?" to "Can we trust an organization full of them?" The winners will not just make agents more capable. They will make that capability governable, portable, observable, and reusable across the company.

CEO David Twizer and Ran Sheinberg have been working on this problem since before enterprise agent sprawl became the fashionable headache. PICO Venture Partners backing the next phase says the market is starting to value the plumbing once everybody notices the building is full of water.

Xpander, Twizer, Sheinberg, and PICO Venture Partners now have $7.5M of new capital behind that thesis. The larger signal is sharper: the model race gets the attention, while the control layer may collect the durable enterprise value.

Medicare Platform

Medicare Platform just announced a $50M revolving credit line, and the structure matters as much as the number. This is not an equity round dressed up for applause. It is debt capacity for scaling Medicare accountable care.

Nestpoint Group facilitated the facility for the Orange County company. The lender, pricing, maturity, collateral, covenants, and amount drawn were not disclosed. The announcement did not arrive with the credit agreement stapled to it.

The disclosed uses are national ACO partner onboarding, AI and data investment, more support for physicians and provider groups, and payments to physician organizations participating in high-quality programs.

That model lives in the unglamorous middle of value-based care, where slogans stop and claims files begin. Medicare Platform says it combines CMS data, attribution analytics, physician dashboards, RAF and HCC workflows, care management, compliance, and ACO formation support so doctors can see how performance becomes shared savings.

Dr. Brian James, COO of Medicare Platform, framed the capital around valuing primary care physicians for keeping seniors healthier and out of the hospital. Henry Huang, Managing Director of Nestpoint Group, described it as private capital backing population-health infrastructure.

The market underneath that pitch is real. CMS estimates that 14.3M Medicare beneficiaries receive ACO-coordinated care in 2026. Shared Savings Program ACOs earned $4.1B for performance year 2024 while generating $2.5B in net savings for Medicare. Those are national figures, not Medicare Platform results, but they explain the attention.

Credit is especially revealing here. Equity usually funds a bet on what a company might become. A revolving line is built for working capital and staged expansion. It can be tapped as needs arise, making borrowing discipline and deployment pace part of the story. The test is whether Medicare Platform can turn access to capital into durable ACO operating capacity.

The details still matter. The borrowing terms remain private, and the company has not disclosed how much of the $50M was drawn. That leaves the next chapter in partner growth, technology execution, physician economics, and measurable outcomes. The money now has to move with evidence, not merely ambition.

Medicare Platform now has capital with a specific job to do. The real question is whether better data, operations, and incentives can make accountable care feel less like policy architecture and more like something physicians and seniors experience.

Dimension Energy

Dimension Energy just secured $857M in additional capital, and the structure matters as much as the headline.

This is not a venture round wearing an infrastructure costume. It is a $200M upsize of Dimension Energy's corporate credit facility plus a $657M construction-to-term debt and tax-equity package. One piece advances the development pipeline. The other finances 29 distributed-solar projects totaling 149 MW across five states.

That separation shows how serious solar platforms get built. Development capital moves projects toward construction. Project debt and tax equity take a defined portfolio toward operation. When both arrive together, the company is building a repeatable machine for turning pipeline into power.

Nuveen Energy Infrastructure Credit and funds and accounts managed by HPS Investment Partners are leading the upsized corporate facility, which now totals $650M. MUFG Bank, First Citizens Bank, ING Capital, and National Bank of Canada are Coordinating Lead Arrangers on the $657M package. Fifth Third Bank is Joint Lead Arranger, while Advantage Capital is providing tax equity.

That is a deep bench making a specific bet: distributed generation can move faster than giant centralized projects because it is built closer to the customers using the electricity. This syndicate is backing Dimension Energy's ability to repeat the work across portfolios.

The company says it has developed more than 1 GW of distributed-energy assets and invested more than $2B across 600 MW operating or under construction. The new financing is intended to help Dimension Energy reach 1 GW of operating assets by 2028. Those numbers make this less about a single close and more about the industrialization of a local-power model.

Community solar reaches customers who cannot put panels on their roofs. NREL estimates that 42% of U.S. households cannot use behind-the-meter solar, while community solar had passed 11 GW-AC by June 2025. The question is who can finance, build, and operate at the speed the grid demands.

Co-founders Rafael Dobrzynski, CEO, and Sam Younes, Chief Development Officer, have built Dimension Energy for that test. Credit to the team translating complex capital stacks into projects people can use, and to partners willing to fund a platform instead of admiring a pipeline from a safe distance.

$857M buys capacity, but the more valuable asset may be proof of repetition. When a distributed-energy company can keep returning to institutional markets with larger portfolios and a broader financing toolkit, scale stops being a slide-deck promise and starts becoming operating discipline.

Ranchbot

Ranchbot just sold $15.8M of equity, and the interesting part is not the size of the check. It is the kind of problem investors are backing: water, distance, and the expensive habit of discovering an operational failure after the cattle already know about it.

A new SEC Form D from Ranchbot Technology Holdings Inc. reports $15,783,741 sold toward a $20M offering. The first sale occurred June 30, 2026, and 20 investors had participated by the July 15 filing.

That language matters. Ranchbot had said it was preparing for a Series B, but this filing identifies equity and does not name a round stage, lead investor, valuation, or use of proceeds. Good reporting does not dress an SEC filing in a venture costume just because the closet is nearby.

What the filing does show is meaningful. Twenty investors put nearly $15.8M behind a Fort Worth AgTech company that treats water monitoring as operating infrastructure, not a dashboard waiting for somebody to admire it.

Ranchbot's monitors track tanks, troughs, pumps, rainfall, flow, and pressure across properties where a routine check can burn hours, fuel, and patience. Satellite and cellular connectivity feed alerts into MyRanchbot so a producer can see trouble before another long drive becomes the diagnosis.

The company says more than 12,000 ranchers use its technology. In 2025, Ranchbot also signed an MOU with Texas A&M AgriLife Research and Extension to explore satellite-connected monitoring, water data, remote sensing, field demonstrations, and applied research.

That is the business case hiding inside the hardware. Ranching has no shortage of tradition, but tradition never required wasting a morning checking a tank that was fine or finding out too late that a pump was not.

CEO Andrew Coppin built the underlying Farmbot business with Craig Hendricks after seeing how much time producers lost to manual checks. Chief Product and Technology Officer Davide Defendi now leads product and technology across the Australian and U.S. operations as the platform expands beyond a single monitor into connected ranch infrastructure.

Farmbot's A$4.6M bridge in 2024 supported U.S. expansion and included Telstra, Macdoch Ventures, and new investors. Those names are prior-round context, not participants that can be assigned to this 2026 financing without evidence.

Andrew Coppin, Craig Hendricks, Davide Defendi, and the Ranchbot team are turning a plain operational truth into a serious AgTech platform: if water is critical, visibility cannot be optional. The next phase now has $15.8M of disclosed equity behind it.

Worldscape

Worldscape raised a $10M Seed Round extension, and the story is not more money chasing another AI model. It is capital moving toward the layer where models meet missions, messy data, security rules, and decisions that cannot be fixed with a prettier demo.

Scout Ventures led the extension, with Radius Capital participating. Worldscape previously announced a $6M seed in March led by Scout Ventures, with Radius and Washington Harbour Partners participating. That is $16M across two announced 2026 financings, while valuation and deal terms remain private.

Founder and CEO Mark Bolz is building Worldscape as an AI-native Decision Advantage Platform for mission engineering. The product brings agentic AI, simulation, digital engineering, distributed data, and secure cloud infrastructure into one environment where teams can model, test, plan, and optimize complex operations.

The market has plenty of models that can produce an answer. Government and industrial users need systems that connect the answer to live data, evaluate alternatives, respect access controls, and operate when the network is not feeling cooperative. Intelligence without an operating environment is just an expensive opinion.

Worldscape says its platform supports deployments from IL2 through IL7, including SAP and fully air-gapped environments, plus multiple Department of War and Department of Energy initiatives. Those are company-reported claims, not disclosed contract values, but they signal software built for real operational constraints.

The new capital will expand engineering and AI research, mission-engineering capabilities, secure and classified deployments, partner integrations, autonomous agents, large-scale simulation, and a marketplace for mission-ready applications, models, datasets, and intelligent agents.

Scout Ventures built its strategy around national security and critical technology, so the repeat backing tracks. Worldscape sits directly in that dual-use lane: infrastructure that can support defense planning while extending into commercial operations, supply chains, industrial systems, and enterprise decision support.

The broader signal is simple. AI value is migrating from chat windows toward governed operating systems that can see context, test consequences, and help humans choose among imperfect options. The winners will not merely generate intelligence. They will make intelligence deployable.

Mark Bolz, Frank Savage, the Worldscape team, and its investors now have a harder assignment than financing the vision. The business is building the infrastructure that lets organizations act on a view of the world.

Higgsfield

Higgsfield just raised a $400M Series B at a $5.4B post-money valuation, and the number is only half the story. DST Global led a syndicate including Tribe Capital, Goldman Sachs Alternatives, Smash Capital, Fifth Wall, Valor Capital, Intel Capital, Liberty Global Tech Ventures, Mirae Asset Capital, NTT Docomo Ventures, and returning backers Accel, Menlo Ventures, AI Capital Partners, GFT Ventures, Capra Ventures, BAM Corner Point, and BroadLight Capital.

That is serious institutional conviction for a company founded in 2023 and launched broadly in 2025. Seven months ago, an $80M Series A extension brought that round above $130M and valued Higgsfield above $1.3B. Now the valuation is $5.4B. Capital did not stroll into this market. It hit fast-forward.

The bet is not simply that AI video gets better. The sharper thesis is that the workflow layer around those models can become the operating system for commercial creativity.

Higgsfield builds its own image and video tools while integrating outside models into a single production environment. Creators and marketing teams can move from ideation and storyboarding through generation, editing, and delivery without rebuilding the process every time a new model wins the weekly benchmark beauty contest.

That is where the company starts looking less like another generator and more like infrastructure. Higgsfield says its platform now serves 25M users who have produced 850M generations, including 300M videos. In June, the company also reported a $500M annualized revenue run rate. Those are company-reported figures, not audited revenue, but they help explain why a $400M check found a pen.

Alex Mashrabov, Yerzat Dulat, and Mahi de Silva are building around an uncomfortable truth for creative-software incumbents: access to a powerful model is not the moat. The moat is turning unpredictable model capability into repeatable production, useful controls, brand consistency, distribution, and a product that nontechnical teams will pay to use every day.

The round is also a loud market signal. Investors are pricing the generative-media application layer as a category with real economic weight, not a novelty attached to somebody else's foundation model.

Alex Mashrabov, Yerzat Dulat, Mahi de Silva, and the Higgsfield team now have considerably more capital behind that thesis. The next test is bigger than making impressive video. It is proving that creative production can behave like software without sanding away the judgment, taste, and control that make the work worth watching. That is the real scale test.

MyGolfSpy, Pin High Golf

MyGolfSpy just became the anchor asset in a new golf media platform, and the interesting part is not the undisclosed check. It is what Graham Sports & Entertainment Partners and Bolt Ventures believe is worth buying: trust, testing data, and an audience trained to question the sales pitch.

The firms acquired and invested in MyGolfSpy through Pin High Golf. The transaction closed in May 2026 and was announced on July 23. Financial terms were not disclosed, so anybody attaching a price tag or round label is practicing a different kind of math.

Adam Beach founded MyGolfSpy in 2008 around a blunt premise: golfers deserved evidence instead of equipment marketing dressed up as evidence. From Yorktown, Virginia, the company built an independent testing and media operation with roughly 17 million annual readers, 15 years of proprietary testing data, a facility with five simulators, and a community of more than 300 volunteer testers, according to the transaction announcement.

That is a media company on paper. In practice, it is a decision engine sitting close to the moment a golfer chooses a driver, ball, shoe, or launch monitor. Most Wanted and Tested & Approved do not just create content. They compress uncertainty for buyers, and that kind of influence gets very valuable when distribution is expensive and trust is scarce.

Pin High Golf plans to expand direct and programmatic advertising, video, affiliate commerce, relationships with golf manufacturers and retailers, and selective acquisitions across golf media, data, and experiences. David Blitzer’s Bolt Ventures brings sports-investment reach. Graham Sports & Entertainment Partners, led by CEO Shervin Mirhashemi, brings a platform-building thesis. MyGolfSpy brings the thing neither firm can manufacture overnight: earned credibility.

The strategic logic is clean, but the operating challenge is hiding in plain sight. MyGolfSpy’s audience showed up because Beach and the team were willing to tell manufacturers when the numbers did not match the marketing. More monetization can compound that advantage, but only if editorial independence remains more than a sentence in an announcement. Better video can make testing more legible, while smarter commerce can shorten the distance between evidence and purchase.

Adam Beach, the MyGolfSpy team, Graham Sports & Entertainment Partners, and Bolt Ventures now have an ambitious platform to build around a rare asset. The capital is useful, but the next test is whether Pin High Golf can scale the business without sanding down the skepticism that made it worth acquiring.

Takt

Takt raised $9.25M in Series A funding, led by Ballast Point Partners, to turn a warehouse problem into the operating system most teams do not have.

The old model treated labor as the main variable. Measure the person, compare the result with a standard, review the report, and call it management. That logic breaks when people, robots, conveyors, time clocks, and warehouse software influence the same shift but report into different systems.

Takt is building for that mess. Its platform connects warehouse systems, time clocks, robotics, material-handling controls, and applications, then organizes that activity into one model of warehouse work. The goal is to help supervisors act while the shift is moving, not explain yesterday.

A warehouse can have accurate systems and still produce a useless picture. A robot slowdown can surface as a labor variance elsewhere. Indirect time can look harmless in an operations report and become a margin problem later. If every system measures only its own lane, nobody sees the pileup.

Takt says more than 100 warehouses across North America and international markets use the platform. Kenco has deployed it across 19 distribution centers and plans 30 more. ODW Logistics reports a 29% improvement in workforce performance and a 39% increase in employee retention after replacing a legacy labor management system with Takt.

Those are customer-reported results, but the signal is useful: real-time visibility matters when it changes a decision, not when it decorates a monthly review. Enterprise operators expand software across a network because the numbers survived contact with the floor.

Takt began with a proof of concept inside a live distribution center. Glynn LoPresti, Noah MacMichael, and Alex Rhea built alongside the supervisors who would use it, an origin that still shapes its roadmap.

LoPresti leads Takt as CEO, MacMichael serves as Chief Product Officer, and Rhea is CTO. The Series A will fund intraday order orchestration, resource scheduling, broader integrations, engineering tools, agentic decision support, and expansion across the UK, EU, and Asia-Pacific.

Ballast Point Partner Sean Barkman is joining Takt's Board of Directors. This is Takt's first institutional round, so the customer proof carries weight. The capital is arriving after operators tested the product where bad software creates late orders and lost margin.

The warehouse stack has spent years getting smarter while the operating picture stayed fragmented. Takt is betting $9.25M of new runway on one clean idea: if labor, automation, and robotics work together, they should be measured and managed together.

Lemma

Lemma just raised a $2.3M pre-seed because AI agents have learned a familiar trick: complete the task, report success, and leave somebody else to discover the wreckage.

The San Francisco company, founded by Jerry Zhang and Cole Gawin, is building production monitoring for AI agents. Lemma looks beyond whether a workflow technically ran. It compares what the agent did with what it was supposed to do, surfaces silent failures, groups recurring issues, and brings the evidence into Slack and the coding tools engineers already use.

Traditional software usually breaks loudly. Tests fail, exceptions fire, requests return bad codes. An agent can execute every step, produce a clean trace, and still make the wrong decision. The dashboard stays green while the customer gets the digital equivalent of a waiter confidently delivering dinner to the wrong table.

Zhang put the problem plainly in the funding announcement: teams often learn about agent failures from users, or never. Lemma’s pitch is that observability cannot stop at execution. If software is reasoning, monitoring has to evaluate whether the outcome matched the intent.

The round includes Matrix, Y Combinator, Liquid 2 Ventures, several other funds, and angels affiliated with OpenAI, xAI, Meta, and DoorDash. No lead investor was identified because precision beats decorating a cap table with assumptions.

Lemma says it monitors more than 1M agent traces per day, a company-reported metric. That scale and the company’s active hiring show where this round lands: inside a team trying to turn agent reliability from a postmortem into an operating system.

The larger signal is sharper than another AI funding headline. Model access is getting easier, and agent frameworks are multiplying. The expensive part is moving from “the demo worked” to “the system behaved correctly across messy production runs,” right before somebody asks why the machine did what nobody intended.

Lemma is betting that the next valuable layer in the agent stack will not be the loudest model or the flashiest assistant. It will be the infrastructure that notices when successful execution produces a failed outcome, explains why, and keeps the same mistake from becoming a customer-support ritual.

Jerry Zhang, Cole Gawin, and the Lemma team now have $2.3M behind that assignment. The test is whether reliability can become continuous enough that agent failures stop looking like weather: inevitable, mysterious, and discussed after the damage. If agents perform real economic work, what evidence should operators demand before trusting the green check mark?

Palona AI

Palona AI just put a hard number behind a big operating thesis: $28.315M in new equity financing for AI that has to perform where business gets loud, messy, and expensive.

Proactive AI Lab, Inc., the legal entity behind Palona, reported the full offering sold to 29 investors. The first sale was July 20, 2026, and the Form D was filed August 4. The filing does not name the investors, disclose a valuation, explain the use of proceeds, or label the round a Series A. Accuracy matters, especially when the headline is carrying somebody else's money.

What Palona is building is easier to understand from the restaurant floor than from an AI conference stage. Its system answers calls, takes orders, handles catering demand, connects with restaurant software, and turns operating signals into actions. Voice, text, images, and video must work together because customers care whether the order is right, not which model handled it.

Palona says its technology exceeds 98% order accuracy through internal testing. In a company-published Cali BBQ case study, it reported $26K in Father's Day sales across 268 orders, up 18% from the prior year, while an AI agent answered 150 calls without a miss. Those company-reported results point to the real scoreboard: captured demand, correct execution, and staff time returned to hospitality.

CEO Maria Zhang led engineering and AI organizations at Google and Meta and served as CTO of Tinder. CTO Tim Howes co-invented LDAP and helped build Loudcloud, Opsware, and RockMelt. Chief Scientist Steve Liu brings academic and applied AI experience from McGill, Samsung AI Center Montreal, and Tinder. The team knows scale and failure.

Palona previously announced a $10M seed round in January 2025. Add the latest filing, and it now has $38.315M in disclosed financing. The new investors remain undisclosed, so the useful question is what the capital now has to prove.

The restaurant industry is projected by the National Restaurant Association to reach $1.55T in U.S. sales in 2026 while operators face costs, uneven traffic, and labor pressure. AI that merely talks is interesting. AI that catches missed revenue, executes accurately, and helps humans deliver better service is a business system.

For Maria Zhang, Tim Howes, Steve Liu, and the Palona team, the next test is not whether the technology can sound human. It is whether it can make a complicated physical business run better without sanding away the hospitality that made the place worth calling in the first place.

Trajectory

Trajectory just raised $40M at a $300M valuation in a round led by Sequoia Capital, according to Axios, citing The Information. The interesting part is what investors are financing: a system designed to make deployed AI improve from the messy evidence of actual use.

Trajectory calls this continual learning. The platform captures what users correct, retry, edit, and accept, then turns those signals into a workflow for understanding model behavior, training improvements, and deciding what gets deployed. Trajectory is betting the valuable trick is helping AI stop repeating the same mistake in production.

The company launched in May with a $15M seed at a $115M post-money valuation led by Conviction. Bessemer Venture Partners, Radical Ventures, BoxGroup, Jeff Dean, and Fei-Fei Li also participated. Less than three months later, Sequoia is leading a new $40M round at a reported $300M valuation. That is a fast change in altitude for a company WIRED introduced as an 11-person team.

Ronak Malde, Michael Elabd, and Arjun Karanam built Trajectory after work spanning Windsurf, Google DeepMind, Apple, robotics, and Vision Pro. Ronak Malde serves as CEO. The pedigree gets attention, but operators should keep the thesis: intelligence that cannot learn from its operating environment becomes expensive repetition with a polished interface.

Trajectory's public examples include Clay, Decagon, and Harvey. Sales, customer service, and legal work all produce valuable, complicated feedback that one generic benchmark will not capture.

There is also a control question hiding inside the excitement. Trajectory says customers decide what data trains, every update runs through evaluation and approval, and deployments remain auditable. That matters because continual learning without governance is not a product advantage. It is an incident report practicing its opening paragraph.

The round is a vote that the next AI infrastructure layer may live after the base model, where real usage becomes training signal and product teams decide what improvement means. Bigger models, better harnesses, and evaluations still matter. But companies that turn their operating data into safer, repeatable gains may build an advantage that cannot be rented from the same API catalog as everyone else.

The formal round label and basis of the $300M valuation were not confirmed, so the clean fact pattern is a new $40M financing led by Sequoia. The larger takeaway is visible: capital is following teams trying to make AI systems learn from work instead of merely talking about it.

VastAdvisor

VastAdvisor says it has closed a $1M SAFE round, and the number is almost the least interesting part of the announcement.

The wealth management industry has spent decades building machinery for managing assets. Client acquisition still runs on referrals, rented leads, and disconnected campaigns. VastAdvisor is building for that gap: an AI-powered Growth OS designed to help RIAs and enterprise wealth platforms turn acquisition into a governed, repeatable system.

The company says Dani Fava of Carson Group, Jason Pereira of Woodgate Financial, and Sally George of Convergency Partners led the SAFE financing. That is practitioner capital in the literal sense. Each brings proximity to the buyer, the operating model, and the regulatory reality VastAdvisor has to navigate.

That proximity matters because wealthtech does not get to cosplay as ordinary SaaS. If compliance arrives after the creative is finished, automation has simply built a faster route to the review queue. VastAdvisor's pitch is different: audience intelligence, campaign orchestration, compliance automation, CRM-connected data, and continuous optimization inside one system.

The company plans to use the new capital to accelerate its product roadmap and support go-to-market execution. It has also been adding operating weight. CEO Ian Karnell is joined by COO Phil Gale, CTO Eli Gassert, and Chief Product Officer Jeremi Karnell as VastAdvisor expands across RIAs, broker-dealers, and enterprise wealth platforms.

A SAFE is not a priced round, and the announcement does not disclose a valuation cap, discount, conversion terms, investor allocations, or prior-funding total. Those blanks should stay blank until somebody with the documents fills them. Private-market storytelling gets expensive when adjectives start doing the accounting.

What can be said is more useful: $1M is focused capital for a company attacking a narrow, expensive problem. Many advisory firms still do not have a durable system for finding, converting, and learning from the next client. If VastAdvisor can make growth data compound while keeping governance inside the workflow, the value will not come from more campaigns. It will come from making the next decision better than the last one.

Ian Karnell, Phil Gale, Eli Gassert, Jeremi Karnell, and the VastAdvisor team now have more room to prove whether wealth management's next operating layer sits before the portfolio, at the point where trust becomes pipeline and pipeline becomes accountable growth.

Wispr

Wispr raised $280M in a Series B at a $2B post-money valuation. Menlo Ventures led the round, turning a voice-dictation company into a test of whether AI's next power center sits where humans tell software what they want.

People speak faster than they type. Wispr Flow turns the pauses, corrections, half-sentences, and verbal clutter of speech into polished text across Mac, Windows, iOS, and Android.

The hard part is not transcription. That technology has been around long enough to collect a pension. The hard part is making voice reliable enough to become a habit across email, documents, code editors, and AI prompts without leaving cleanup behind.

Tanay Kothari and Sahaj Garg have pursued that interface since founding Wispr in 2021. Kothari serves as CEO and Garg as CTO. Wispr began with silent-speech hardware, shifted toward software, and found the sharper wedge in Flow.

The capital curve now looks almost as fast as the product promise. A $30M Series A led by Menlo arrived in June 2025. Another $25M led by Notable Capital followed in November, when Wispr said it had raised $81M and was valued near $700M. Now the reported post-money valuation is $2B.

The Series B includes Acrew, Forerunner, Goodwater, Peak XV, Together Fund, Plus Capital, Livvy Dunne, Shaun White, Dak Prescott, and DK Metcalf, alongside existing investors Notable Capital, NEA, Neo Ventures, 8VC, and MVP Ventures.

That mix says plenty. This is enterprise software, consumer habit, and cultural distribution sharing one cap table. Wispr wants voice to feel normal wherever a cursor appears, then move from words toward actions, context, and interfaces assembled around intent.

Traction needs attribution. Sensor Tower counted more than 2.5M global downloads from October through April, while Wispr said in March that Flow had hundreds of thousands of daily users. Reliability, privacy, and retention now have to scale with the check.

Apple, Google, and Microsoft own surfaces where typing happens. Wispr has to become the preferred layer before better voice input becomes another bundled feature people accept by default.

This round matters because capital is moving toward the AI interface layer, where distribution, trust, and daily habit can matter as much as model performance. The contest is not simply who understands speech. It is who turns spoken intent into work without making the user manage the machinery.

Tanay Kothari, Sahaj Garg, and the Wispr team now have $280M of new capital behind a larger assignment: make the keyboard optional without making trust optional.

Terra Industries

Terra Industries just added $18M and closed its seed round at $52M. The more interesting story is what the company intends to build with it: an African manufacturing base for autonomous security systems, connected by software and protecting critical infrastructure across the Global South.

The financing includes returning investors 8VC, Silent Ventures, Nova Global, Belief Capital, and SV Angel, plus new investor Norleo Space Investments and angel investor Grant Gordon. Terra named no lead for this extension, and none should be inferred.

Nathan Nwachuku and Maxwell Maduka founded Terra in 2024. Nwachuku serves as CEO and Maduka as CTO. They are betting that power plants, mines, borders, and maritime assets should not depend entirely on systems designed for different terrain, supply chains, and security realities.

Terra connects autonomous drones, an interceptor drone, sentry towers, and unmanned ground vehicles through ArtemisOS, its command layer for detection, tracking, mission planning, and coordinated response. The company reports that its technology protects approximately $11B in critical assets across multiple African countries. That metric is company-reported, but it gives this round more shape than the usual big market, bigger ambition routine.

The capital is earmarked for a first international office in London, expanded manufacturing, faster deployments, and team growth. Manufacturing remains in Africa. Terra's 15,000-square-foot Pax-1 facility is in Abuja. Its planned 34,000-square-foot Pax-2 facility in Accra is targeted to open in Q4 2026 and reach annual capacity of 50,000 aerial systems by 2028.

Those factory and capacity numbers are targets, not trophies. Defense manufacturing is unforgiving. Hardware must survive the field, supply chains must hold, and governments move on their own clocks. Capital buys the chance to prove the system. It does not do the proving.

Still, the round shows where attention is moving. Sovereign manufacturing, autonomous systems, and data control are becoming one connected thesis. The Global South is not merely a place where imported defense technology gets sold. It is becoming a place where defense technology is designed, built, deployed, and financed around local conditions.

Nathan Nwachuku, Maxwell Maduka, and the Terra Industries team now have $52M of seed conviction behind the next test: turning capital into repeatable production, trusted deployments, and an industrial platform that can travel without losing the local advantage that made the thesis credible.

VERO

VERO's 2025 growth-equity round was less about putting a louder number on the cap table and more about putting sharper operators around a messy corner of multifamily housing. Fifth Wall and Sunriver Capital Partners co-led the undisclosed financing, with Rebuild Capital participating.

The amount stayed private. Undisclosed is not an invitation to let imagination do the accounting. The round followed VERO's $9M Series B, and Fifth Wall told Commercial Observer the new financing was significantly larger.

The timing was the real headline. Property teams were sorting through fragmented identity checks, income documents, credit files, criminal-background data, and prior-residency verification. Generative AI made fake pay stubs and synthetic identities cheaper while the cost of a bad approval stayed painfully real.

VERO's answer was one operating layer for that risk. The platform combines screening, verification, fraud analysis, and protection services, with integrations across Yardi, Entrata, RealPage, and Rent Manager. VERO said it had screened hundreds of thousands of applications and flagged tens of thousands of fraudulent ones.

The business lesson sits inside that model. VERO is not adding another point solution. It is trying to own the decision layer where speed, fraud risk, and renter experience collide. For operators, that layer touches occupancy, bad debt, and net operating income.

Capital came with a leadership reset. Jamey Rosamond became CEO, Travis Gibson joined as CTO, and Jeff Cate remained CFO. Founder Louis "Lou" Baugier moved into a senior-advisor role, giving VERO continuity without pretending the next phase needed the first one's operating shape.

That combination makes the investor roster interesting. Fifth Wall knows built-world technology. Sunriver Capital Partners brings growth capital and operating experience across real estate, construction, and insurance. Rebuild Capital works where promising companies need more than a clean venture narrative.

The use of funds was straightforward: accelerate product development, strengthen go-to-market execution, and expand VERO's place in the property-technology stack. Multifamily operators are no longer buying screening as a checkbox. They are buying confidence between an application and a signed lease.

VERO, Jamey Rosamond, Travis Gibson, Jeff Cate, and Louis Baugier are building the next chapter with evidence behind it. The round needed no disclosed valuation to make the signal visible: when fraud gets smarter, the infrastructure deciding who gets through has to get smarter first.