The Room Inside the Moment

Rainmaker Technology

Water managers have paid for cloud seeding with a stubborn accounting problem: the storm arrives, the gauges move, and nobody can cleanly separate what nature delivered from what the operation added.

Rainmaker Technology just raised a $100M Series B to make that uncertainty measurable enough to buy.

Founder and CEO Augustus Doricko is building Rainmaker as an atmospheric research organization, drone operator, and water-infrastructure company at once. NOA VC, Upfront Ventures, DCVC, Lowercarbon Capital, and Dream Ventures joined the round. No lead investor or valuation was disclosed.

The aircraft gets the cinematic part. Rainmaker flies weather-resistant drones into existing supercooled clouds and releases silver iodide so ice crystals can form and fall as rain or snow. The harder business sits behind the flight path: choosing the cloud, measuring what the intervention added, and defending the number after the weather moves on.

Rainmaker says 4 months of 2026 operations produced 82 unambiguous seeding signatures associated with more than 145M gallons of incremental precipitation. During an August campaign in Alaska, the company estimated 45 to 65 acre-feet of additional water, with a mean of 57.6 acre-feet, or roughly 19M gallons.

That Alaska number came from radar and a precipitation-estimation ensemble. Rainmaker's report identifies uncertainty around calibration, beam geometry, evaporation, natural precipitation, and the boundaries around each signature. That disclosure matters because infrastructure earns trust through accounting, not atmosphere.

The Series B will expand Rainmaker's science team, support national-laboratory and university research, and improve targeting and validation. More drone teams can operate in parallel, but field capacity becomes a business only if the evidence scales with it.

Rainmaker wants its cost per acre-foot to fall below most new-water sources during the 2026-2027 season. That forecast still has to survive weather windows, watershed geography, permitting, public trust, and proof that the water arrived where a buyer needed it.

Investors are financing a useful shift in the question. Cloud seeding used to ask whether an operation could influence precipitation. Rainmaker is trying to show how much, where, at what cost, and with enough confidence for someone to put it into a water plan.

The company has capital for more flights. The larger opportunity is a measurement system that keeps earning belief after each storm has left the radar.

Deeper Analysis 👇 https://devcuration.com/articles/rainmaker-technology-raises-100m-in-series-b-funding

Split Pay

The rent is due once. Most paychecks arrive twice. That calendar mismatch has been collecting fees, forcing tradeoffs, and turning stable household income into a monthly scramble.

Split Pay has raised $125M across Series A and Series B rounds to underwrite that gap. Khosla Ventures led both financings, with Thrive Capital, Max Levchin, New York Life Ventures, MetaProp, Alpaca VC, Moderne Ventures, Intuit Ventures, SciFi VC, and other investors participating.

The financing breaks down into a $25M Series A and roughly $100M Series B. It follows a $15M seed round in 2023 and brings Split Pay's disclosed funding to about $140M.

Co-founder and CEO Andrew Borovsky and the founding team built the product around household physics. Rent, mortgages, and car payments arrive as one large obligation. Split Pay pays the landlord or lender in full on the due date, then lets the customer repay the platform in two paycheck-timed installments.

The commercial bet sits inside the underwriting. Split Pay says its Lens AI model evaluates real-time cash flow and behavioral signals rather than leaning on a conventional FICO score. Axios reports the service charges 2% of the full bill plus a $10 monthly subscription, with no interest or late fees.

That structure gives customers breathing room without asking a landlord or lender to change how it gets paid. It also leaves Split Pay holding the timing risk, which means customer acquisition only matters if repayment discipline follows it.

The company reports 70X growth over the past 12 months, more than $1B in bills split, 150,000+ members, and 95% monthly retention. It has also reported a 97.5% repayment rate across $350M in originations. Those are company figures, but they explain why investors are financing scale instead of another checkout feature.

Borovsky told FinTech Futures the capital will accelerate customer acquisition. The product is already moving from rent toward mortgages and car payments, while Axios reports a Visa credit card is planned.

The useful lesson is not that every bill needs another payment plan. It is that credit products can be built around the timing of a specific obligation and the cash flow available to meet it, rather than treating every consumer as a static score.

Split Pay now has enough capital to carry that thesis into larger balances, more bill types, and more households. Each expansion gives the underwriting system more room to learn, and more ways to discover whether better timing can remain better economics.

Deeper Analysis 👇 https://devcuration.com/articles/split-pay-raises-125m-in-series-a-and-series-b-rounds

Softcat / GDT

A five-year search gets expensive the moment the right answer appears.

Softcat screened roughly 100 U.S. acquisition targets before GDT - General Datatech cleared the bar. GDT brought scale, infrastructure depth, enterprise customers, and a culture Softcat believed it could keep rather than flatten.

The price of that conviction is a $1.05B all-cash agreement, announced September 17. Softcat plans to fund it with balance-sheet cash, new debt, and an equity issue that raised about £354M gross. The transaction has not closed. It still needs HSR and CFIUS clearance, among other conditions, with completion expected by the end of Q1 2027.

That distinction matters because the work starts before ownership changes hands. Graham Charlton is financing a step change from international reach to owned U.S. delivery capacity. GDT Chair and CEO Shawn O'Grady carries the relationships that made the target worth buying.

GDT brings 692 enterprise and upper-mid-market customers, roughly 1,000 vendor relationships, and depth across networking, data centers, cybersecurity, hybrid cloud, and AI infrastructure. Bengaluru adds 232 people and a 24/7 delivery platform. Softcat brings broader procurement, lifecycle, cloud, and international capabilities.

The cross-sell deck writes itself. The customer experience will be harder.

Softcat has said GDT will keep its name, leadership team, workforce, and operating structure after closing. That is sensible. A buyer does not spend five or six years searching for culture and capability only to file both down during integration.

Softcat forecasts GDT at about $240M in gross profit and $80M in EBITDA for the 12 months ending December 2026, implying a 13.1x purchase multiple. It expects high-single-digit to low-double-digit underlying EPS accretion in the first full fiscal year. Those are forecasts, not realized outcomes.

What Softcat is really buying is proximity: to U.S. enterprise customers, to Cisco, NetApp, NVIDIA, and other major vendors, and to the infrastructure work underneath AI adoption. GDT grew from carrier networking into commercial cloud, security, data-center, and AI environments where failure becomes an operating event, not a software demo.

Now the deal has to preserve the exact qualities that justified the premium. The search ended when Softcat chose GDT. The harder handoff runs through regulatory clearance, customer trust, employee retention, and every multinational deployment the combined company has promised it can carry.

Deeper Analysis 👇 https://devcuration.com/articles/softcat-agrees-to-acquire-gdt-for-1-05b

NVIDIA / Brookfield

The most consequential NVIDIA deal in an AI factory may now happen before the first GPU arrives.

Land has to be controlled. Power has to be secured. Data-center shells, cooling, networking and compute have to be financed long before customers turn capacity into revenue. The chip gets the attention. The balance sheet carries the calendar.

Brookfield Asset Management disclosed that NVIDIA is a $2B anchor LP in the Brookfield Artificial Intelligence Infrastructure Fund. Brookfield launched the vehicle in November 2025 with a $10B equity-commitment target and $5B already committed by a group including Brookfield, NVIDIA and the Kuwait Investment Authority.

September 17 brought a new disclosure about NVIDIA's share, not a new $2B fund launch or a completed $10B close. With co-investment and financing, the vehicle may support up to $100B of assets across energy, land, data centers and compute.

Jensen Huang has spent years arguing that accelerated compute is becoming productive infrastructure. NVIDIA is now putting capital beside that argument. A supplier financing the asset class that buys its platform is doing more than waiting for demand. It is helping decide which projects can become supply.

Brookfield's materials show NVIDIA as an LP, technology partner and tenant, NAVER investment partner, and financing partner in a separate initiative designed to mobilize more than $500B of third-party capital.

The $100B describes asset capacity BAIIF and related financing may support. The $500B belongs to broader compute-financing platforms. Neither is cash deployed today or NVIDIA revenue.

Sikander Rashid, Brookfield's Global Head of AI Infrastructure, is organizing the parts of the stack that refuse to move on software time: power, permits, construction, hardware and long-duration capital. NVIDIA brings the systems and demand network. Brookfield carries projects between ambition and utilization.

That handoff will shape more than GPU sales. It will influence who owns technology risk, who absorbs empty-capacity risk and which governments, enterprises and AI clouds can turn a compute plan into an operating asset.

NVIDIA's $2B commitment gives Brookfield a formidable anchor. It puts both firms closer to the moment when demand forecasts must survive project finance. The harder work now moves through power contracts, project underwriting and customers willing to pay for the capacity once the lights come on.

Deeper Analysis 👇 https://devcuration.com/articles/nvidia-anchors-brookfield-ai-fund-with-2b-commitment

Matter Venture Partners

HardTech founders do not get to solve one problem at a time. The technology can work while the supplier slips, the factory misses yield, the customer delays qualification, and the hiring plan discovers that specialized talent already has 4 offers.

Matter Venture Partners built its second fund for that pileup.

The firm has closed Matter Venture Partners Fund II at $450M, following a $300M inaugural fund announced in March 2024. An amended SEC Form D reports the full $450M sold across 151 investors, with $0 remaining. Matter's original filing listed a $350M offering, so the final close landed $100M above that starting point.

Wen Hsieh, Founding Managing Partner; Haomiao Huang, Founding Partner; and Mel Tang, Founding Operating Partner and CFO, are directing Fund II toward early-stage companies across semiconductors, robotics and Physical AI, AI infrastructure, quantum computing, advanced manufacturing, energy, and AI for physical sciences.

That map shares one problem: the product has to survive contact with the physical world.

A semiconductor company may be living inside fabrication schedules, packaging capacity, qualification cycles, power budgets, and customers who need proof before they redesign a system. Robotics adds hardware reliability and factory deployment. Quantum adds technical risk that does not care what the fundraising calendar promised.

Matter's more interesting asset may be the people behind the money. The firm named ASML, Development Bank of Japan, Kleiner Perkins, Nitto Denko, Quanta Computer, Resonac, Sojitz, and TSMC among its strategic LPs. Nitto separately disclosed a $10M commitment.

Those institutions span semiconductor equipment, materials, manufacturing, industrial markets, and institutional capital. Matter says its LP network can help portfolio companies as collaborators, early customers, co-investors, manufacturing partners, suppliers, and entry points into global industry ecosystems.

Production is still difficult. The network changes who can answer the phone when a prototype needs a supply chain and a customer needs confidence.

Matter's portfolio runs from chips and AI networking to robotics, manufacturing, and compute infrastructure.

The $450M close gives Matter more room to back difficult systems early. The LP roster gives those systems more doors into the industries that will have to build, qualify, buy, and scale them.

Fund II now moves from capital formation into those handoffs, where HardTech stops being a promising design and becomes an operating company.

Deeper Analysis 👇 https://devcuration.com/articles/matter-venture-partners-closes-450m-fund-ii

Hines / Rialto Capital

A national vacancy rate cannot tell a lender whether the elevators work, the tenants will renew, the floor plans still fit, or the borrower can carry the next refinancing.

That judgment lives one building at a time. Hines and Rialto Capital just raised $1.1B around it.

Hines Rialto Credit Partners reached its final close on September 2 with $1.1B in investor commitments for U.S. office credit. The managers announced the close on September 14, and a September 4 SEC filing reported $1.079B sold across the vehicle and feeder funds from 126 investors.

The partnership is useful because each manager enters the loan file from a different door. Hines brings nearly 7 decades of owning, developing, operating, and investing in real estate. Rialto brings origination, underwriting, asset management, and special servicing, including the experience that arrives after a capital structure stops behaving politely.

That combination matters in an office market where the averages are improving while the collateral remains stubbornly specific. CBRE reported national office vacancy fell to 18.3% in Q2 2026, with prime vacancy at 12.3%. The Mortgage Bankers Association estimates 17% of office-backed mortgage balances mature this year.

Better leasing does not erase a maturity date. A maturity wall does not make every building broken. Somebody still has to decide which sponsor, tenant roster, basis, submarket, and business plan can support new debt.

Hines Rialto has already shown both sides of the mandate. Reported transactions include a $228.9M bridge loan for Manhattan's Textile Building, nearly $100M of Midtown loan acquisitions, a $91M financing package for One America Plaza in San Diego, and a $58M refinancing for a Short Hills, New Jersey office campus.

Those deals do not prove a portfolio-wide result, and the managers have not disclosed LP names, leverage, fees, returns, total deployment, or remaining dry powder. A Hines affiliate's audited statements also recorded a $2.5B target at the end of 2025; the final announcement does not explain whether that target changed.

The close still places meaningful nonbank capital inside the hardest part of the office cycle: separating a wounded category from an individual asset that can earn another loan. Hines can read the building. Rialto can read the debt. The next chapter gets written where those 2 views agree closely enough to fund the same address.

Deeper Analysis 👇 https://devcuration.com/articles/hines-rialto-closes-1-1b-office-credit-fund

Fluxnium

Three parts per billion is enough uranium to sustain centuries of nuclear demand and little enough to ruin the economics of anyone trying to collect it.

Fluxnium just raised a $7M Seed round to work on that contradiction.

Congruent Ventures led the financing. Constellation Technology Ventures and Active Impact Investments also participated. The capital will fund development, pilot testing, and scale-up of a fiber designed to adsorb uranium from seawater.

PNNL estimates that the oceans hold more than four billion tons of uranium. Collection is harder.

Fluxnium plans to suspend high-surface-area fibers from offshore longlines, expose them to seawater for roughly 30 to 60 days, then retrieve and process them into yellowcake. The lines are meant to be reused, creating a physical model that looks more like seaweed farming than conventional mining.

That replaces the geology question with an operating system.

Adsorption capacity, selectivity, reuse degradation, marine operations, recovery chemistry, and processing all land in the delivered cost.

Those variables decide whether a technically vast resource becomes commercial inventory.

U.S. nuclear operators bought 46.9 million pounds of uranium in 2025, according to the EIA. Ninety-three percent came from foreign sources. Domestic supply accounted for 7%.

Constellation's participation deserves careful reading here. Its venture arm gives Fluxnium an industry connection to a major nuclear operator, but the announcement explicitly says the check is not a fuel-purchase commitment. That line matters. Strategic interest can sharpen the product roadmap without pretending revenue has already arrived.

Founder and CEO Jeff Green previously built NanoH2O, a membrane company acquired by LG Chem. Co-founder and Executive Chairman Jim McDermott has backed industrial climate companies. Congruent partner Kevin Kopczynski joins the board.

Fluxnium says its system can reach cost parity with conventional mining while avoiding mine tailings and groundwater disruption. Public research has already shown that seawater uranium recovery works. Older lab economics were still above market uranium prices, and Fluxnium has not disclosed a current independently verified cost per pound.

So the $7M is not paying for the discovery that uranium is in the ocean. It is paying for evidence that surface area, time at sea, reuse, recovery yield, and processing can become a supply chain.

That is a demanding pilot. It is also the right proof point.

Deeper Analysis 👇 https://devcuration.com/articles/fluxnium-raises-7m-in-seed-funding

Marel Power Solutions

The factory floor gets the final vote on power density, because a brilliant converter architecture still has to be built repeatedly, cooled reliably, sourced flexibly, and priced for the system around it.

Marel Power Solutions now has $1.9M to put that vote on the record. The Plymouth, Michigan company received a Michigan Supplier Conversion Grant to purchase equipment for a pilot production line, beginning with electric-vehicle applications.

The accounting matters. This is a manufacturing grant, not an equity round. Marel is one of 11 Michigan suppliers sharing $21M through a state program supported by the U.S. Department of Energy.

Amrit Vivekanand, CEO, is building around a constraint that tightens as systems grow. EVs need lighter traction inverters. Defense platforms need more power without giving back space or weight. Data centers need more electricity inside footprints already claimed by compute.

Marel Power Solutions, Inc. attacks the layer between the semiconductor die and the finished system. Its modules and stacks integrate electrical and thermal paths, giving customers a platform they can adapt instead of funding another bespoke converter design from zero.

The company says its architecture can cut size, weight, semiconductor count, and cost, depending on the configuration and comparison baseline. Those are company-reported claims. The grant has a more immediate job: buy equipment that can show whether the design survives repeatable manufacturing.

That work sits on a real technical base. CTO Ian Byers and COO Gary Miller are named inventors on Marel's granted air-cooled power-converter patent.

The commercial idea is direct. A system builder selects a configuration, keeps supplier flexibility, and reuses the architecture across programs instead of treating every power-conversion requirement like a fresh expedition.

The obligation is harder. Marel still has to prove pilot-line repeatability, performance under customer conditions, production economics, and a path from EV applications into defense and data-center programs. The grant gives the team the machinery required to answer those questions in Plymouth.

That is the uncomfortable middle of hardtech, where engineering claims meet fixtures, yields, suppliers, cycle times, and customers who need the same result more than once. Marel's next advantage will be measured in the output of that line, as each power stack carries the density argument into a vehicle, rack, or platform with no patience for laboratory excuses.

Deeper Analysis 👇 https://devcuration.com/articles/marel-power-1-9m-michigan-grant

Health Force

Europe's hospitals can be rich in software and short on people to move the work through it.

That is the buyer problem behind Health Force's $4.8M Seed round. The Barcelona company builds AI agents that log into the systems hospitals already use, complete administrative workflows, record each action, and return exceptions to staff with the relevant case prepared.

LUMO Labs led the round, with Step Venture, Next Tier Ventures, Target Global, and Calm/Storm participating. At the European Central Bank's September 17 reference rate, the financing equals approximately $4.8M.

Fadi Haddad, Co-Founder and CEO, and Dr. Juan Suárez, Co-Founder and CPO, chose the hard version of enterprise AI. CTO Andrew Kochura and the team are sending agents into workflows where legacy systems, national rules, and human accountability meet.

The company says its agents handle work behind more than 200,000 patient cases a year. Three of every 4 finish end to end without human touch. The remainder is prepared for staff review.

Those figures are company-reported, but the deployment context is concrete. Health Force names Humanitas and Gruppo San Donato in Italy and supports hospital quality reporting in Portugal. It says production can begin within 6 weeks because the agents operate existing software.

That product choice carries the business thesis. Hospitals have spent years buying systems that record admissions, authorizations, claims, reports, and trials. Health Force wants to become the operational layer that completes the work those systems describe.

The round will expand the agent catalogue into quality reporting, clinical-trial operations, and resource optimization. Health Force also plans to grow engineering and delivery while moving into Germany, France, and Spain.

WHO/Europe estimated this year that the region could face a shortage of nearly 1M health workers by 2030. Software will not recruit a nurse or settle every governance question. It can change how much administrative volume each hospital team carries without adding another screen to the shift.

LUMO Labs and the participating investors are backing a change in what hospital software gets paid to deliver: completed, traceable work inside the systems a hospital already owns.

The expansion now carries that promise across borders, where every new hospital will test whether the same agent model can respect different systems, rules, and people without losing the accountability that earned the first deployment.

Deeper Analysis 👇 https://devcuration.com/articles/health-force-raises-4-8m-hospital-operations-ai

Corridor

A 20-person company can demand nearly the same brokerage labor as a large employer while producing a fraction of the commission. Small businesses have been living inside that mismatch every renewal season.

Corridor has raised a new $16M seed led by Bain Capital Ventures (BCV), bringing total funding to $25M after a prior $9M pre-seed. BoxGroup and Definition Capital participated alongside healthcare and AI operators.

The funding math matters because Corridor's operating thesis starts with the math traditional brokerages cannot make work. Quoting plans, comparing carriers, supporting enrollment, coordinating with insurers, and helping an employee untangle a surprise bill do not shrink in proportion to the account's commission.

Nikhil Aggarwal, CEO, and Jason Dong, COO, came from healthcare distribution and company building. Jackson Wagner, CPO, and Eric Qian, VP of Engineering, came from AI and data infrastructure after building clinical agents around musculoskeletal care. The 4 co-founders moved upstream to the health plan that determines how most Americans enter the system.

Corridor pairs licensed advisors with AI agents. The humans learn the team's needs, explain tradeoffs, and stay accountable to employers and employees. The agents organize company and plan data, gather carrier options, build proposals, support enrollment, and coordinate the repetitive work that can turn a broker's calendar into a filing cabinet with notifications.

That division of labor matters. KFF found that workers at firms with 10 to 199 employees contributed an average $8,889 toward family coverage in 2025, compared with $6,227 at larger firms. Their average single-coverage deductible was $2,631, versus $1,670 at larger employers.

Corridor says its clients save 20% on average without reducing benefit quality. That is a company-reported result, and the company now has to prove it can hold service quality as volume, carrier complexity, and renewal pressure rise together.

Bain Capital Ventures is backing the administrative leverage. Corridor is betting that licensed advice becomes more valuable when software removes the work that never needed human judgment in the first place.

Small employers have spent years being told their account is too small for serious attention while every premium, deductible, network decision, and claim lands at full size on the people involved. Corridor's $16M gives the team room to make the service model fit the consequence.

Deeper Analysis 👇 https://devcuration.com/articles/corridor-adds-16m-seed-to-launch-with-25m-total

Feldera

The most expensive row in a data warehouse may be the one that did not change.

Data teams still rerun enormous SQL workloads because a sliver of new information arrived. The warehouse recomputes yesterday so the business can learn what happened a moment ago. Freshness becomes a luxury while the cloud bill grows with accumulated history.

Feldera has raised a $15.4M Series A led by Inovia Capital, with Costanoa Ventures and Battery Ventures participating. The company says it has raised $21.5M across the Series A and its earlier Seed.

CEO and co-founder Lalith Suresh and CTO and co-founder Leonid Ryzhyk are part of a five-founder systems team with Mihai Budiu, Ben Pfaff, and Gerd Zellweger. Their work became DBSP, the research foundation beneath Feldera's incremental view maintenance engine.

The mechanism sounds academic until the invoice arrives. Feldera watches how data changes and updates only the affected results, even with hundreds of joins, recursive logic, and state larger than memory. Compute follows the change rather than the whole dataset.

Feldera reports customers cutting warehouse compute costs by 95% or more while moving freshness from hours or days to seconds. One company case study moved hundreds of thousands of lines of SQL from a 70-node Spark cluster to one or two Feldera nodes after backfill. The outcomes are company-reported, but the buyer's problem is specific.

The most revealing example sits inside authorization. A joint Feldera and Auth0 case study reports a 7.1B-plus Permissions Index updated on one node with average latency under 200 milliseconds. AI agents can generate thousands of checks while retrieving private data, so stale access logic can put the wrong context in front of the wrong model at machine speed.

The new capital will support the core engine, scalability, lakehouse integrations, a fully managed experience, growth, and hiring. Inovia is leading the Series A while Costanoa returns from the Seed and Battery joins the syndicate.

That investor choice is a bet on more than a better benchmark. Enterprise infrastructure wins when the hard math disappears behind dependable operations, clean migrations, support, and a product that survives the night shift.

Feldera has proved that complex SQL can be maintained without repeatedly buying the same computation. The Series A now funds the slower conversion: making that idea ordinary enough for enterprise teams to stop treating waste as the price of fresh data.

Deeper Analysis 👇 https://devcuration.com/articles/feldera-raises-15-4m-to-cut-sql-recomputing-costs

Terragrit

A utility can approve a capacity project in one room and discover the consequences in five others. Engineering sees load. Operations sees downtime. Finance sees capital. Reliability sees what remains after the lights blink.

That is the decision problem Terragrit is financing with a strategic investment from National Grid Partners. The undisclosed amount keeps the useful story away from the funding scoreboard and closer to why these organizations belong together.

Terragrit gives teams a shared environment for modeling changes to physical operations before they reach the field. A utility can compare a substation upgrade, transmission-line addition, or capacity reroute, then inspect the cost, timing, resilience, and work-sequencing effects. The platform is built for the people who understand the operation, not only simulation specialists.

Founder and CEO Sam Nagar is building around a lesson enterprise infrastructure keeps teaching: a technically correct answer can become an operational mistake when the teams carrying its consequences work from different versions of the problem. CTO Zaid Al-Ali owns the simulation engine, AI layer, and platform reliability behind making the model accessible without making it shallow.

National Grid Partners brings more than an investor logo. Arun Chetty and the NGP team work where load growth, aging infrastructure, reliability obligations, and data-center demand collide with slow planning cycles. Its NextGrid Alliance connects leaders from more than 170 utilities, giving Terragrit access to procurement realities and deployment constraints that product demos rarely volunteer.

Utilities buy evidence that a change can survive engineering review, operating conditions, budget pressure, safety requirements, and the handoff to the crews living with it. Terragrit reports more than 80 Fortune 500 customers and a 150-person team. Both figures are company reported.

The investment will support product development and commercial growth, with energy and utilities becoming a deeper focus. It also puts an unforgiving market beside the roadmap. If Terragrit shortens the distance between an operational question and a defensible decision, NGP can carry that value across a utility network.

The grid does not care which department had the cleanest slide. It carries the combined decision after everyone leaves the room, and Terragrit now has National Grid Partners helping bring more of that room into the model before work reaches the field.

Deeper Analysis 👇 https://devcuration.com/articles/terragrit-adds-national-grid-partners-for-utility-simulation

PtEverywhere / Sheridan

Andrew Shofner joined PtEverywhere as CEO in October 2025 with a growth investor and a new leadership team. Less than a year later, he is staying to run the company under Sheridan Capital Partners' ownership.

That handoff is fast, but the investment logic has been developing inside physical therapy clinics for years.

A patient visit may last an hour. The business behind it keeps moving through scheduling, intake, clinical notes, home exercises, messages, billing, payments, collections, and follow-up. Every broken handoff asks a clinician or office manager to become the integration layer.

PtEverywhere was built to stop that job from bouncing between systems. The platform brings practice management and integrated payments into one workflow for outpatient PT and rehabilitation providers, with a particular focus on small and mid-sized cash-pay and hybrid clinics.

The problem is expensive enough to attract serious capital. The American Physical Therapy Association reported that 90.8% of respondents to its 2025 survey linked administrative burden to burnout. Three-quarters of facilities had added nonclinical staff to handle it. More than half said the burden had pushed their practice out of a payer or network.

That is an operating-model problem with a measurable payroll consequence.

Sheridan completed its investment on September 17. Terms were not disclosed. Andrew Shofner remains CEO. Jacob Greenberg, Aman Singla, and Noah Benson led the deal for Sheridan, with Sidley Austin advising the firm and Goodwin Procter and Kaizen Equity Partners advising PtEverywhere.

The sharper sentence in Sheridan's announcement was the plan to support organic growth and strategic add-on acquisitions.

That turns PtEverywhere from a clinic-software company into a potential platform. The wager is that scheduling, documentation, billing, payments, patient communication, and rehab-specific workflows are dense enough to support more capability without becoming another stitched-together healthcare stack.

Private equity can fund the product roadmap and buy adjacent pieces. It cannot make integration painless by declaration. Clinic owners will judge the outcome in fewer duplicate tasks, faster collections, better support, and more time that reaches the patient instead of the back office.

Sheridan now owns the layer between the clinician's work and the business that must collect for it. PtEverywhere's next chapter will be written in how little of that machinery the clinician has to notice.

Deeper Analysis 👇 https://devcuration.com/articles/sheridan-takes-ownership-of-pteverywhere-pt-software

AGV America

When a manufacturer takes 90 days to pay, the integrator still has payroll, steel, controls and technicians moving this week.

That timing gap is the machinery behind AGV America’s $1.25M invoice factoring facility from Gateway Commercial Finance. The San Antonio integrator did not sell equity or announce a venture round. It replaced a bank line with financing tied to eligible receivables.

AGV America’s work is physical long before the customer’s payment becomes cash. A project can require engineering, in-house fabrication, controls, installation and commissioning across AGVs, AMRs, conveyors, cranes and collaborative robots. The customer may be large and creditworthy. The people and materials building the system still need to be paid on today’s calendar.

That is where factoring becomes less exotic than the label sounds. A strong customer makes an invoice valuable, but the invoice can remain stubbornly illiquid for 30, 60 or 90 days. Gateway is using those eligible receivables to create more working-capital availability while AGV America keeps current jobs moving and pursues larger ones.

Eric Jones has built AGV America around an equipment-agnostic model spanning 15+ suppliers. The company designs systems, fabricates steel in San Antonio and carries projects through installation and commissioning. Its largest reported deployment involved 100 vehicles, which means equipment choices, shop work, field labor and customer acceptance can all move before the final invoice clears.

The facility connects AGV America’s financing more closely to that operating cycle. Gateway founder and managing director Marc J. Marin is underwriting eligible receivables that the prior bank structure did not fully convert into availability. The announcement does not disclose the advance rate, fees, recourse terms or how much was drawn, so $1.25M is capacity, not cash already spent.

For project businesses selling into large enterprises, customer credit quality can lower collection risk without shortening the wait. Growth can improve the order book while making cash conversion more demanding when every larger job mobilizes more labor and material before payment.

AGV America now has a financing structure built around the work it has already billed. The value will show up in the handoff between signed opportunity, fabricated system, commissioned line and collected invoice, where growth either keeps moving or starts asking operations to finance the customer.

Deeper Analysis 👇 https://devcuration.com/articles/agv-america-adds-1-25m-receivables-facility

Peaky

The month-end report has a cruel sense of timing. It explains the customer who left, the salesperson who missed, the hiring plan that slipped, and the currency move that changed the margin after the business has already spent weeks living with the consequences.

That lag is the market Peaky is financing.

The Ghent fintech has announced a $1.3M (€1.1M) Seed round led by JK INVEST, with imec.istart, Super Capital, and about 15 Belgian angels participating. The company says the round closed earlier this year; the public announcement arrived on September 16.

Ser Cappelle spent years in finance, including work as a fractional CFO for scale-ups. The frustration kept repeating: operations and sales could see change immediately, while finance inherited the explanation when the next reporting cycle finally caught up.

Ser Cappelle built Peaky with Guus De Graeve, Willem Vos, and John-Alexander Kolmus. The four co-founders have worked across Netlog, Twoo, Ablo, and Tinder, and now hold the CEO, CTO, COO, and CDO roles respectively.

Peaky connects accounting, CRM, ERP, payment, and other operating systems to the financial plan. It brings in external signals such as currency movements, identifies when an assumption drifts from reality, and shows the possible financial impact. AI agents support reporting, analysis, forecasting, and scenarios, while the finance team keeps the decision.

That last part matters. A company does not need another dashboard announcing that something went wrong. It needs the assumptions behind the plan connected to the people and systems changing them.

Peaky names Aikido, Tout Bien, Mobility Plus, The Harbour, and Proplanner among its customers. It also won Best FinTech Startup of the Year at the 2026 Digital Finance Awards Belgium. Those are useful signs of early relevance, not audited proof of revenue, retention, or product outcomes.

JK Invest's lead fits its early-stage B2B SaaS focus. Returning investor imec.istart, Super Capital, and the angel group are backing a product that sits inside a sensitive handoff: operating data becomes financial judgment, then management action.

The capital will fund more developers, product work, marketing, and European commercial expansion. Peaky says it wants to become the European standard for companies scaling without surprises.

That ambition now travels through every integration, forecast, and conversation where a finance leader decides whether Peaky saw the change early enough to alter the decision.

Deeper Analysis 👇 https://devcuration.com/articles/peaky-raises-1-3m-for-live-financial-planning

Euka AI

Jeremy Yaoxin Ding watched a six-to-seven-person affiliate team at PatPat manage TikTok Shop by hand. Creators had to be found, pitched, sampled, briefed, tracked, and paid while every post and order created another loose end.

That workload became the starting point for Euka AI, which CEO Ding co-founded with CTO Kevin Wang in 2024. Euka has disclosed $5M in previously raised seed funding led by Susa Ventures and Creator Ventures. September 17 was the disclosure date, not a proven same-day close.

The financing surfaced beside a bigger product move. Euka is extending its TikTok Shop creator system to Instagram, with Shopify-connected fulfillment and YouTube planned next.

Creator marketing loves discovery because discovery photographs well. The operating bill arrives later. A brand working with thousands of creators has to know who received product, who posted, who owns the content, which link produced the order, what commission applies, and whether the same person exists as one relationship or four platform accounts.

Euka wants to be the system that keeps those handoffs from splitting apart.

The platform covers creator search and vetting, automated outreach, CRM, samples, affiliate links, content workflows, contests, attribution, and payouts. Euka also lets brands move licensed creator content into Meta advertising and query program data through MCP tools.

The scale is material, according to Euka: $4B+ in GMV, 17,000+ brands, 4M+ indexed creators, and 55,000 creators using its app. The company says it powers 6 of the top 10 TikTok Shop brands and won a 2026 Catalyst Award for driving the highest incremental GMV increase for brands.

Those figures show reach. They remain company-reported and do not independently prove durable retention, margins, or that a TikTok-native workflow will transfer cleanly to every platform.

That transfer is the seed-stage bet. Instagram, Shopify, TikTok Shop, Meta ads, and YouTube create different rules for content, rights, identity, attribution, and checkout. Brands do not need 5 dashboards that agree after finance closes the month. They need one operational record that survives the handoff.

The capital gives Euka room to turn platform expansion into operating continuity while brands discover how quickly a creator program becomes infrastructure. The next chapter sits between a creator post, a Shopify order, a usage-rights approval, and a payout, where campaign cleanup starts deciding whether creator commerce can scale.

Deeper Analysis 👇 https://devcuration.com/articles/euka-discloses-5m-seed-for-omnichannel-creator-commerce

Instinct

Every text Instinct turns into a booked flight, paid bill, or canceled subscription consumes infrastructure the user never sees.

The interface is a message thread. Behind it sits an agent reading context, navigating sites, using credentials, and spending compute on work that continues past one answer. That hidden bill is now carrying a large financing conversation.

Instinct is reportedly in talks to raise $1B at a valuation of about $10B. The Information says Sequoia Capital and Benchmark have discussed leading, while Coatue Management has also discussed a lead role. Nothing has closed, and the terms can still change.

This would arrive less than a month after the company raised a $250M Series B at a $2.25B pre-money valuation, co-led by Index Ventures and Benchmark. The round put Instinct near $2.5B post-money and total funding at $350M. The speed points to the economics of personal agents.

Noah Shinn built Instinct around an almost invisible interface. Users text or call. The assistant works across email, messages, calendars, screens, location, and connected services to book, buy, cancel, follow up, or coordinate. The company reportedly has more than 100,000 users, and The Information has reported capacity warnings as that usage grows.

A chatbot can stop after generating an answer. An agent inherits the rest of the job: permissions, retries, website quirks, payment details, confirmation rules, and the cost of a wrong action. That is where a simple interface becomes an infrastructure company and a trust company.

Instinct's own terms authorize the assistant to act across connected services and enter transactions on a user's behalf. Its privacy policy acknowledges access to sensitive communications, account credentials, and payment information when users grant permission. Early testers have already pushed the company on data retention, confirmation, and unintended actions.

Those questions do not erase the adoption. They explain what the capital must help Instinct survive. More users mean more compute, more edge cases, more third-party behavior, and more moments when reliability has to feel boring.

A $10B valuation discussion is a bet that personal agents become a daily operating layer before the cost structure and business model have finished introducing themselves.

Instinct has made delegation feel like sending a text. Noah Shinn and the investors around the table are deciding how much capital it takes to keep that promise when 100,000 people begin sending one back.

Deeper Analysis 👇 https://devcuration.com/articles/instinct-in-talks-for-1b-at-roughly-10b-valuation