The Company You Cannot Run From Your Head Anymore

Adaptive

The most expensive spreadsheet in construction may be the one everybody trusts three weeks late.

A controller can reconcile every transaction in the ledger and still be wrong about the job. The missing facts are outside the accounting system: a superintendent knows the schedule slipped, a project manager approved extra work in an email, and somebody in the office is calling around to learn what the numbers mean.

Matt Calvano, Henry Bradlow, and Francisco Enriquez built Adaptive around that handoff. The company has raised a $30M Series B led by Tidemark, with Emergence Capital, Andreessen Horowitz, Pathlight, Definition, and 3KVC participating. Adaptive says the round brings total capital raised to $57M.

The product does not ask a contractor to replace the ERP or teach every person in the field another dashboard. Adaptive’s Project Accounting Agents pull context from schedules, daily logs, project-management systems, email, text, and voice, then connect it with records in QuickBooks, Sage, Foundation, Acumatica, and Procore.

That context feeds the unglamorous work that decides whether a contractor understands the business in time to act: coding invoices, drafting change orders, preparing billings, assembling WIP, and keeping lien waivers, insurance, payments, and compliance current. A person reviews every action before it reaches the general ledger. The agent handles the chase; the finance team keeps the judgment.

Adaptive reports that more than 750 construction companies now use the platform, from businesses with $5M in annual revenue to operators at $1B, across more than 10 accounting systems. It also works with over 40 accounting firms. Those are company-reported figures, but the shape of the adoption matters: Adaptive is moving from one SMB workflow toward a layer that has to survive different contractors, trades, ERPs, and approval rules.

Tidemark calls this a System of Action. That phrase earns its keep here because the bet is not another record of what happened. It is software that goes looking for the context required to move the work forward, then stops at the line where an accountant still needs to approve the answer.

The Series B will expand agents across job costing, AP, billing, WIP, payments, and compliance while Adaptive grows its Boston and New York teams. Valuation and terms were not disclosed. The operational wager is clearer: a contractor should not have to wait until month-end to discover what the field already knew.

Deeper Analysis 👇 https://devcuration.com/articles/adaptive-raises-30m-construction-accounting-ai

Confido

CPG back offices rarely lose margin with one dramatic crash. It leaks through the handoff between the sales forecast, trade calendar, deduction queue, cash posting, and supply plan.

Every system can work. The business can still spend the week reconciling them.

Confido has raised a $55M Series B led by Insight Partners, with Footwork, Trenches Capital, Watchfire, Barrel Ventures, and Y Combinator returning. Confido reports $77M in total funding, although its prior $20M disclosure and the new round do not fully reconcile, so that cumulative figure remains company-reported.

The sharper part of the story is what Justin Hunter and Kara Holinski chose to connect.

Confido started with cash application and deductions, then expanded into trade promotion, sales forecasting, demand planning, and supply planning. It all reads from the same customers, products, promotions, contracts, and shipments.

That architecture turns a boring back-office problem into a serious operating question. If sales changes the forecast, does supply see it immediately? If a promotion moves, can finance see the margin effect before the money leaves? If a retailer deducts cash, does the evidence travel with the dispute, or does somebody start hunting through portals and email again?

Confido says more than 250 CPG brands use the platform, including divisions of Unilever, Mars, and Nestlé, plus OLIPOP, Simple Mills, and Daisy. It says those brands plan more than $30B in retail sales on the system. The metrics are company-reported, but they show the scale of the coordination problem Insight Partners is underwriting.

The $55M will fund agentic, zero-click workflows, a broader planning cycle, expansion into food service, and hiring across product, engineering, and go-to-market.

“Zero-click” sounds clean. CPG operations are not. An agent filing a dispute or changing a plan needs the right contract, promotion, customer, shipment, permission, and exception history. Connect that context and automation can remove reconciliation. Miss it and one system distributes the mistake faster.

That is the real Series B assignment. Confido has capital to become the operating layer between a brand’s growth plan and the margin it actually keeps. The next evidence will be whether customers expand across modules without building another spreadsheet economy around the platform.

Retail volume creates complexity before it creates clarity. Confido is betting the brands that control the handoffs will keep more of the growth they worked to earn.

Deeper Analysis 👇 https://devcuration.com/articles/confido-raises-55m-series-b-cpg-ai

Ferry Health

The first version of Ferry Health was a spreadsheet 2 brothers built for their father after he left the hospital.

Mehul Mehta and Keyur Mehta had discharge instructions and an insurance directory. They needed a neurologist, cardiologist, and physical therapist who accepted the plan, treated the right condition, and could see a new patient.

They started calling. Retired providers, dead numbers, and long waits turned follow-up care into a family operations project. Andrew Buie joined them, and Ferry grew from that spreadsheet into an AI assistant that does the work instead of handing the patient another list.

Ferry has now emerged from stealth with $9M in Seed funding from Andreessen Horowitz, Index Ventures, Avid Ventures, Layout Ventures, SV Angel, BoxGroup, and Asymmetric Capital, plus Jay Desai and founders of Grow Therapy, Hex, and Chapter.

The investor logic sits inside the difference between finding a doctor and securing an appointment. Ferry takes a request by message, checks insurance, location, subspecialty, new-patient status, and availability, then calls offices and books the patient's choice. Online workflows, direct integrations, and human care coordinators meet the exceptions healthcare produces for sport.

Ferry distributes through health systems, insurers, digital-health companies, and care-navigation platforms. The company says more than 1M patients have access across all 50 states, with a target above 5M by year end. Access is the precise word. Active use, booking volume, revenue, pricing, retention, and audited outcomes remain undisclosed.

That boundary matters because healthcare is full of technically completed handoffs that never become care. A referral can exist, a directory can load, and the patient can still remain home.

Ferry is selling completion into that gap. Providers gain a better chance that referred patients arrive, health plans get a more usable network, care managers recover time on hold, and patients get something less glamorous and more valuable: an appointment on the calendar.

The $9M gives the team room to hire, deepen integrations, and extend Ferry into imaging, home health, and other coordination work. The harder proof sits in stale data, insurance nuance, scheduling rules, human preference, and the office that answers on the seventh call.

That spreadsheet is becoming infrastructure. The next proof will live in the appointments Ferry can complete when the directory, the phone system, and the patient's life all disagree.

Deeper Analysis 👇 https://devcuration.com/articles/ferry-health-raises-9m-ai-care-navigation

Teal Health

The Teal Wand can cross the country in a mailer. Making at-home cervical screening routine still requires payers, providers, labs, employers, and follow-up care to move together.

That is the operating job behind Teal Health’s $22M Series A.

.406 Ventures led the round. Emerson Collective, managed by Yosemite, Forerunner, and Serena Ventures returned, while Japan-based MPower Partners joined. Teal says total funding is now $45M.

Kara Egan and Avnesh “Ave” Thakor, MD, PhD, built Teal Health around a stubborn reality: cervical cancer screening works, yet appointments, the speculum exam, time away from work, and limited availability keep millions from staying current.

The Teal Wand changes the collection step. A patient completes a virtual visit, receives a prescription kit, collects a vaginal sample at home, sends it to a lab for high-risk HPV testing, and gets results through a secure portal with follow-up navigation when needed.

The clinical evidence matters because convenience without performance would be a nicer failure. In the peer-reviewed SELF-CERV trial, self-collected samples showed 95.2% positive agreement for high-risk HPV and 95.8% sensitivity for high-grade cervical dysplasia, equivalent to clinician collection.

Teal received FDA De Novo authorization in May 2025 and expanded nationwide in January 2026. The company now reports that it has screened thousands of women, 59% of whom were underscreened. Those are company figures, but they point at the customer the old workflow keeps missing.

The next handoff is financial. Federal guidance is expected to require most health plans to cover self-collected cervical screening and needed follow-up without patient cost sharing for plan years beginning in 2027. Teal is using the new capital to expand payer, health-system, employer, provider, and brand partnerships, grow direct-purchase channels, and hire across clinical, commercial, technology, and operations.

Authorization made the product possible. Reimbursement can make the category routine. Between those milestones sits the unglamorous machinery that decides whether a patient actually gets screened: prescribing, fulfillment, billing, lab capacity, results, and follow-up.

Teal’s Series A is financing that machinery. The company’s next milestone will show up in ordinary handoffs: a prescription issued, a kit delivered, a sample processed, a result understood, and a patient who is no longer overdue because the system finally met her at home.

Deeper Analysis 👇 https://devcuration.com/articles/teal-health-raises-22m-for-at-home-cervical-screening

BackOps

Enterprise software is good at identifying problems. Supply-chain teams get paid for resolving them.

BackOps has raised a $42M Series B led by Insight Partners, with investors Theory Ventures, Construct Capital, Gradient Ventures, and 10VC participating. The round arrived 6 months after a $26M Series A, putting the operating evidence under a brighter light than the announcement.

Sean McCarthy, co-founder and CEO, learned the problem inside Amazon Shipping. Henry Ou, co-founder and CTO, came from applied machine learning at Apple and ByteDance plus ERP integration. Supply-chain work rarely fails because nobody spotted the issue. It fails in the handoffs among ERPs, carrier portals, email, operations, finance, and the person who still knows what the process should do.

BackOps calls the space between identifying a need and completing the outcome the “Resolution Gap.” Relay ingests SOPs, tickets, emails, and workflow history, then deploys purpose-built agents across APIs and old interfaces that still require browser clicks. Approval gates, escalations, and audit trails keep people in control where consequences demand it.

The company says a national parcel platform has processed more than 500,000 claims through BackOps since December 2025, with 91% resolved without human involvement. BackOps also reports that the automated resolution rate rose from 87% at launch to 99% today as monthly volume increased 150-fold without added hiring.

A separate retailer operating 13 sites reportedly cut an average billing cycle from roughly 28 hours to 14 minutes. BackOps estimates the change avoided 11,000 staff hours and a planned 6-person billing expansion worth about $660K in projected fully loaded cost.

Those are meaningful production claims, but they remain company-reported. BackOps has not disclosed revenue, pricing, customer count, valuation, ownership, or an independent audit of the results.

The Series B is financing an expansion from workflow automation into operating infrastructure. BackOps plans to support more processes across operations, finance, and customer service while growing product, engineering, and go-to-market teams. Every resolved workflow adds context, decisions, actions, and outcomes to the operating record.

The category gets interesting when software stops finding an exception and starts carrying approved work through the awkward systems and human handoffs between the problem and the result. BackOps now has $42M to prove resolution can become a layer, not another queue.

Deeper Analysis 👇 https://devcuration.com/articles/backops-raises-42m-supply-chain-resolution-ai

Firecrawl

An AI agent can be brilliant and still fail because the source it needed was never connected, licensed, indexed, or even visible. That gap is becoming a business.

Firecrawl has raised a $75M Series B led by Smash Capital, with Altos Ventures, Nexus Venture Partners, Y Combinator, Freestyle, and Offline Ventures participating. The round arrived with Alexandria, a knowledge library that brings official data providers, specialized indexes, custom connectors, and the live web into one retrieval layer for AI agents.

Caleb Peffer, Nicolas “Nick” Silberstein Camara, and Eric Ciarla started with a practical problem. The web was messy, JavaScript-heavy, and full of teams rebuilding the same extraction plumbing. Firecrawl turned that work into search, scraping, crawling, parsing, and interaction APIs.

The company says more than 1.5M users now build with Firecrawl. That scale promoted the team to a harder question: once an agent can read the open web, how does it reach financial data, scientific work, government records, code artifacts, publisher archives, and expertise that requires permission, payment, or a dedicated integration?

Alexandria combines the live web with first-party providers and Firecrawl’s Research, Developer, and Government indexes. Firecrawl already pays sources including Wikimedia Enterprise for direct access. The new capital will support deeper retrieval, more official sources, and payment infrastructure for contributors whose knowledge agents use.

Firecrawl reports that Alexandria improved answer quality by 21% over built-in web tools in an internal test of 845 tasks using the same models and prompts with blind AI judging. That is a company benchmark, not an independent verdict. It still sharpens the thesis behind this round: model quality is only one ceiling. Coverage, freshness, rights, and retrieval discipline can become ceilings first.

The operating burden grows with the ambition. Firecrawl has to make incompatible sources feel consistent, make usage economics worthwhile to providers, and keep agents from confusing access with authority. A library for AI becomes valuable when builders can understand what is inside it, how current it is, and why that source belongs in the answer.

This round gives Firecrawl room to build that trust layer while the market is still deciding what agent knowledge should cost. The important shift is not simply that AI needs more data. The people and institutions holding useful knowledge may finally get a place in the transaction.

Deeper Analysis 👇 https://devcuration.com/articles/firecrawl-raises-75m-series-b-alexandria

Snorkel AI

A useful AI training example used to be something a person could label before lunch.

At the frontier, it can look more like a small company. The model needs tools, documents, policies, hidden constraints, a long sequence of decisions, and a grader smart enough to catch the moment it finds a shortcut instead of solving the job.

That change helps explain why Snorkel AI just raised a $350M Series E at a $3.5B valuation.

Insight Partners and S32 co-led the round. New investors Third Point Ventures, March Capital, Blumberg Capital, Allegis Capital, Frontline, and Standard VC joined returning backers Addition, Lightspeed Venture Partners, Greylock, GV, Prosperity7, Wells Fargo, Walden Catalyst Ventures, and Factory.

Snorkel began as Stanford AI Lab research in programmatic labeling, a way to create training data without labeling every example by hand.

Agentic AI changed the unit of work. Snorkel now combines domain experts with specialized models and agents to build datasets, reinforcement-learning environments, benchmarks, rubrics, and quality controls for high-skill work.

A useful environment has to resemble the real job, expose the right tools, preserve the right constraints, and grade behavior that may unfold over hundreds of steps. The expert is no longer only supplying an answer. The expert is helping define the world in which the answer must survive.

Snorkel says its data-as-a-service business grew more than 18x after launching in September 2025 and crossed a $375M annualized revenue run rate. Reuters reported the run rate as more than $350M. Those are company-reported, unaudited figures, but they explain why investors are treating advanced AI data as infrastructure.

Co-founder and CEO Alex Ratner is scaling a thesis the team began studying in 2015. CTO Jimmy Hillis has to turn expert-dependent work into a repeatable system without sanding off the judgment that makes the data valuable.

The $350M gives Snorkel room to hire researchers and engineers, expand enterprise and government delivery, support third-party model evaluations, and enter more industries and modalities. It also raises the operating standard. Growth this fast has to become durable quality across customers whose hardest data problems cannot be solved with a bigger queue of labels.

The next generation of model capability will be shaped inside these engineered environments, where experts decide what good work looks like and machines learn how to practice it at scale.

Deeper Analysis 👇 https://devcuration.com/articles/snorkel-ai-raises-350m-series-e-frontier-ai-data

Harvey

The most expensive legal knowledge may be the judgment a firm has already paid to develop and cannot teach a generic model.

Precedent lives in briefs, deal documents, review comments, partner preferences, client history, and thousands of choices that never became a clean dataset. Access to a frontier model gives every firm similar raw capability. Teaching a system how one institution actually works is where the advantage starts getting difficult.

Harvey has raised $550M at a $15.5B valuation. Lightspeed calls the financing Series H and co-led it with Diffusion. Sequoia, Kleiner Perkins, a16z, Coatue, GIC, and Goldman Sachs Alternatives were among returning investors; Sapphire and Whale Rock were among the new ones.

Winston Weinberg and Gabe Pereyra are putting the capital into people and compute: hiring across product, engineering, research, and go-to-market, building for contracting, litigation, deals, and compliance, and training the models after Harvey Tenet.

Tenet matters because Harvey is moving beyond wrapping someone else's model. The company post-trained an open-weight Kimi K3 base for long-horizon legal work and built LAB to evaluate agents on realistic matters. The results are early company research, but the direction is clear: application companies want more control over cost, capability, and customer intelligence.

That ambition carries weight. Harvey says it has crossed $400M in ARR and serves more than 3,000 customers, including 80% of the Am Law 100, 20% of the Fortune 500, and five of the Fortune 10. Those are company-reported figures, yet they explain why investors are financing a much bigger technical obligation.

Legal agents have to respect permissions, preserve citations, understand firm standards, survive document volume, and behave predictably when client work gets messy. Siva Gurumurthy's engineering organization is scaling that infrastructure while Harvey expands inside high-stakes workflows.

The valuation climbed from $8B in December to $11B in March and now $15.5B. Capital can buy time, compute, talent, and acquisitions. It cannot shortcut the trust review that happens before a firm lets an agent touch a live matter.

Harvey's bet is that legal teams will own more intelligence built from their work instead of renting every answer from the same model layer. If that becomes the buying standard, the durable asset may be the system that makes institutional judgment trainable, testable, and governed without separating it from the people responsible for the outcome.

Deeper Analysis 👇 https://devcuration.com/articles/harvey-raises-550m-series-h-legal-ai

Yardi / Sidero Labs

Yardi did not meet Sidero Labs in a banker’s deck. It met Talos Linux in production.

Yardi knew how the operating system behaved before it bought Sidero, the company behind Talos Linux and Omni. The September 14 transaction gives Sidero the backing of a 10,000-plus-employee software company. The price and structure were not disclosed.

Sidero had a problem good code could not solve. Enterprise buyers told the team it was too small to trust with critical infrastructure. They also feared an acquirer could close the project, redirect the roadmap or treat the community like a line item.

The buyer became the answer to the buyer-risk question.

Talos Linux is built around subtraction. It has no conventional SSH administration or package manager. The host is minimal, immutable and managed through an API, leaving platform teams fewer moving parts to secure across Kubernetes fleets.

Omni carries that model across bare metal, virtual machines, cloud and edge locations. Sidero says its products manage tens of thousands of clusters for organizations including Roche, SNCF, Ubisoft and Nokia. Those are company-reported claims, but they show why continuity matters beyond the cap table.

The licensing promise needs precision. Talos Linux remains open source under MPL-2.0, with the same repositories and contribution process. Omni is source-available under BSL 1.1. The products remain together, but they do not share the same open-source license.

Sidero’s current site lists Steve Francis as CEO, while Yardi’s release calls him president. Talos creator and former CTO Andrew Rynhard left months before the deal. Yardi founder Anant Yardi is now chairman.

The first roadmap consequence is already visible. Talos Hypervisor adds native virtual-machine support, while edge containers can run without forcing Kubernetes onto every single-node site. Sidero plans an alpha demonstration in October and general availability in December.

That expansion arrives while buyers are reconsidering virtualization licensing and the cost of separate systems for VMs and containers. Sidero is betting that the narrow architecture that made Talos useful for Kubernetes can carry a wider estate.

Yardi gives the team scale and a longer enterprise résumé. Sidero gives Yardi an infrastructure product it trusted before it owned. The stewardship test starts now: ship the roadmap, keep the licensing boundary honest and let the community remain strong enough to tell its new owner when the architecture is drifting.

Deeper Analysis 👇 https://devcuration.com/articles/yardi-acquires-sidero-labs-as-talos-moves-beyond-kubernetes

MIND

An AI agent can behave like a power user without ever taking a lunch break.

It can read thousands of files, connect to internal systems, move information, and finish before a security analyst has worked through the first alert queue. That is useful until the wrong data travels with it.

MIND has raised a $72M Series B led by Crosspoint Capital Partners, with existing investors YL Ventures and Paladin Capital Group participating. The company reports $112M in total funding, one year after its $30M Series A.

Eran Barak, Itai Schwartz, and Hod Bin Noon built MIND around a blunt observation: data loss prevention became famous for rules, noise, and operational pain before software started making decisions on its own.

The old model asks a security team to discover sensitive data, write classifiers, tune policies, investigate incidents, clear exceptions, and decide how to respond. MIND's AI DLP Agents take on more of that work across SaaS, GenAI, agentic AI, endpoints, on-premises file shares, and email. The customer sets the strategy and boundaries. The machinery handles more repetition.

Enterprises are trying to adopt AI without casually donating sensitive data to a browser tab, an unsanctioned agent, or a workflow nobody thought to inspect.

MIND says revenue grew more than 17x and customer count grew 8x over the past year. It describes an eight-figure revenue business trusted by dozens of customers, with billions of data events analyzed and hundreds of thousands of endpoints protected. Those company-reported numbers help explain the size of this operating bet.

Crosspoint's role adds some history. The firm includes leaders who helped build and run the first generation of enterprise DLP. Now it is leading a round behind a company arguing that the category needs a different operating model for machine-speed work.

The capital will fund product development, enterprise expansion, technology and channel partnerships, and team growth. Calcalist/CTech reported an approximately $300M valuation and an all-primary round; MIND did not disclose valuation or terms.

The interesting question is not whether security teams want fewer alerts. They have wanted that for years. It is whether autonomous DLP can make thousands of small decisions with enough context to reduce the work while keeping policy ownership where it belongs.

MIND now has the money to push that handoff deeper into the enterprise. Every new agent, connector, file path, and exception will make the answer easier to judge.

Deeper Analysis 👇 https://devcuration.com/articles/mind-raises-72m-ai-native-dlp

Cyera

Permission reviews were built around the comforting assumption that a person would eventually slow the system down.

An employee asks for access, a manager approves it, and somebody has time to notice when context changes. AI agents remove that pause. They call tools and take action at machine speed, often with valid credentials for a decision nobody reviewed.

That is the operating pressure behind Cyera’s $400M Series G extension from Growth Equity at Goldman Sachs Alternatives.

The investment extends the $600M Series G led by Evolution Equity Partners in June, bringing the round to $1B and disclosed funding above $2.7B. Cyera says it is valued above $12B. The interesting part is the security handoff it is trying to collapse.

Yotam Segev, Tamar Bar-Ilan, and Yonatan Itai built Cyera around understanding sensitive data: what exists, where it lives, and who can reach it. The company has expanded across DSPM, DLP, privacy, identity, endpoint, and agent security.

Agent Guardian maps sanctioned and unsanctioned agents, their models, tools, identities, and data connections. Cyera Endpoint follows that activity onto devices. The $1B Oasis Security acquisition adds non-human identity controls and turns Oasis into Cyera Identity.

The architecture is chasing a simple problem with expensive consequences. Data security can tell an enterprise which information matters. Identity security can tell it who or what has access. An agent can satisfy both systems and still take the wrong action because its permission outlived its business purpose.

Goldman Sachs Managing Director Irit Kahan framed the gap as the difference between what agents are trusted to do and what they can reach. AI agents do not need stolen credentials to create risk. Valid access, stale intent, and machine-speed execution are enough.

Cyera says the capital will fund AI Security capabilities, federal expansion, and growth across EMEA and APAC. Its federal offering is FedRAMP In Process, so certification work remains. The platform still has to prove that acquisitions and rapid expansion can become one coherent control plane.

The wager behind this $1B Series G is that enterprise AI will make access decisions continuous. Every tool call, query, and automated action becomes a fresh authorization question, even when the identity looks familiar.

The next evidence will come from security teams trying to let agents move quickly without discovering, millions of actions later, that a valid credential was doing an invalid job.

Deeper Analysis 👇 https://devcuration.com/articles/cyera-adds-400m-series-g-extension-goldman-sachs

Luzern Risk

The software demo begins only after the customer agrees to own an insurance company.

That is the strange bargain inside captive insurance. A business can stop handing every dollar of risk to a commercial carrier, retain underwriting profit, and build surplus inside an entity it owns. It also inherits an insurer's capital, claims, filings, actuarial work, audits, board obligations, and regulator.

Luzern Risk has raised a $45M Series B to make that second part less punishing. Insight Partners led the round, with Trust Ventures and existing investor Caffeinated Capital participating. Caffeinated led an undisclosed 2023 seed and a $12M Series A in 2025, putting Luzern's disclosed funding floor at $57M before counting the seed.

Gabriel Weiss, CEO and co-founder, is building for a market that has lived comfortably inside the Fortune 500 and awkwardly elsewhere. The economics can work for a middle-market company, but operating a captive still means coordinating actuaries, accountants, lawyers, reinsurers, claims administrators, and regulators.

Jonathan York, CTO and co-founder, is not trying to automate those professionals out of the room. Luzern's platform organizes financials, claims, policies, documents, filings, and workflows around them. Axios reports that actuaries, regulators, and accountants retain their sign-off roles, while Luzern generally uses a third-party claims administrator.

Insurance technology loves a clean interface. Insurance ownership still has consequences.

Luzern charges an annual management fee based on a captive's size and complexity. It works with owners, brokers, fronting carriers, reinsurers, and advisers. The company has not disclosed customer counts, revenue, audited savings, or named clients here.

The capital will expand the platform and AI, systemize operations, reduce turnaround times, and add alternative-risk options. Sam Espinosa, CMO and co-founder, is positioning that infrastructure for buyers who want control without building a captive back office.

NAIC estimates about 8,000 captives operate globally and roughly 90% of Fortune 500 companies use them. Luzern says captives write about $240B in annual premium. The more interesting number may be the companies outside that club whose risk has become expensive enough to own and complicated enough to need a better operating system.

The Series B gives Luzern more room to build that system. Every client has to accept the obligations of an insurer in exchange for control over the economics of one.

Deeper Analysis 👇 https://devcuration.com/articles/luzern-risk-raises-45m-captive-insurance-platform

H Clinical

Ben Joers and Eddie Hannush are entering H Clinical through two doors at once. Emerald Square Ventures is taking a majority stake, while its founders become H Clinical's CFO and CCO.

Mitchell Parrish remains CEO of H Clinical. Ben Joers will lead financial strategy, planning and infrastructure as CFO. Eddie Hannush will own commercial strategy, client relationships and service growth as CCO.

That org chart fits the thing H Clinical sells.

A sponsor can choose Latin America for its patient populations and still spend months coordinating the people and physical systems behind a trial. Sites need coordinators and nurses. Home visits create sample, shipment and regulatory work. Equipment needs to be sourced, stored and moved under country-specific rules.

H Clinical says its in-country teams connect that work across 22 Latin American countries. The company provides clinical-research staffing, community-based recruitment, in-home and hybrid trial support, plus procurement, depots, equipment rental and courier logistics.

The product is operational ownership of the handoffs.

FDA guidance recognizes in-home visits as decentralized trial elements, while qualified personnel, protocol discipline, training and consistent data collection still have to travel with the visit.

Emerald Square is betting that a regional operator can make those obligations easier for sponsors and CROs to manage. Joers brings finance and M&A; Hannush brings enterprise sales and partnerships. Both are leaving the investor lane and taking day-to-day responsibility inside the company they acquired.

The investment will expand site and project staffing, patient recruitment, home and hybrid delivery, procurement and logistics. H Clinical did not disclose hiring targets, new countries, acquisitions, revenue, valuation or audited performance gains.

That boundary matters because integration is a promise until a study feels the difference. Fewer vendor handoffs should create a cleaner operating line, but clinical research still measures the result in enrollment, retention, protocol quality, timelines and the experience of patients who may live hours from a traditional site.

Parrish keeps the continuity. Joers and Hannush take the financial and commercial controls. Their shared job is to scale a 22-country system without separating it from the local people, relationships and physical infrastructure that made the majority investment worth pursuing.

Deeper Analysis 👇 https://devcuration.com/articles/h-clinical-majority-backing-latin-american-trials

CorePower Magnetics

Sam Kernion and Paul Ohodnicki spent years working on a metal ribbon most people would never connect to the AI infrastructure story.

The material promised smaller, lighter, cooler, more efficient inductors and transformers. It also had the manners of a potato chip: brittle, difficult to process, and poorly suited to industrial heat and punishment.

That gap between a brilliant material and a dependable component is where CorePower Magnetics built its business.

The Pittsburgh company has raised a $10.6M oversubscribed equity financing co-led by Engine Ventures and Material Impact. Evergreen Climate Innovations, Carnegie Mellon Catalyst Fund, and Baruch Future Ventures participated. CorePower reports $13.6M in equity funding, plus $27M in separate non-dilutive support from U.S. energy and defense programs.

CorePower works across alloy formulation, rapid-solidification casting, annealing, component design, and production. Its inductors, transformers, motors, and amorphous steel sit where electricity changes voltage, frequency, or form before reaching a data center, grid asset, industrial system, or vehicle.

That integration is the bet. Power-electronics customers have spent decades accepting custom components, long design cycles, heat limits, and supply-chain friction. CorePower is trying to standardize more of the category without losing the materials advantage.

The capital will add Pittsburgh production capacity and equipment, expand hiring, and move multiple product lines toward volume production. Reed Sturtevant of Engine Ventures and Carmichael Roberts of Material Impact are joining the board, putting manufacturing judgment beside the materials science.

The IEA expects global data-center electricity use to roughly double from 485 TWh in 2025 to 950 TWh in 2030. NREL says U.S. distribution-transformer capacity requirements could rise by as much as 260% by 2050. Every new megawatt still has to survive the trip through physical equipment.

CorePower reports paid engagements with major companies and multiple product lines moving toward volume production. Customers, revenue, margins, yields, and production targets remain undisclosed. A September SEC filing reports $14.36M sold, above the announced round and stated equity total, a difference the public record does not explain.

The next phase belongs on the factory floor, where the alloy has to become the same reliable component again and again while demand keeps climbing outside the building.

Deeper Analysis 👇 https://devcuration.com/articles/corepower-magnetics-raises-10-6m-us-manufacturing

Vaulted Deep

The next carbon-removal site may be financed by the customers who agreed to buy its future tonnes.

Vaulted Deep has secured a $35M debt facility from Mediobanca, arranged by CFP Energy, to build new U.S. waste-disposal sites and invest in the technology that moves a location from geological possibility to operating infrastructure.

That is a different funding milestone for a company launched in 2023. Equity pays a team to prove a model. Debt asks whether contracts, assets and performance can carry repayment while the company keeps building.

Vaulted Deep has more than carbon-credit demand behind the loan. Municipalities, farms and industrial operators pay it to manage organic waste that needs a reliable destination. The company turns that waste into slurry and stores it deep underground, creating verified carbon removal alongside the disposal service.

The facility is supported by those waste agreements and long-term offtakes with Frontier buyers. Every future tonne begins as an infrastructure obligation. Someone still has to identify the geology, secure the permit, source the waste, build the site, run the injection operation and prove delivery.

Co-founder and CEO Julia Reichelstein and co-founder and Executive Chairman Omar Abou-Sayed are building from operating history. Vaulted Deep spun out of Advantek Waste Management Services and uses injection technology with a record reaching back to Los Angeles operations in 2008.

The numbers give Mediobanca something sturdier than a forecast deck. Vaulted Deep says it delivered more than 20,000 tonnes to Frontier buyers in the first half of 2026, above all of 2025. Axios later reported more than 27,000 tonnes for the year. The company says weekly waste volume has increased 6x since 2023.

Debt leaves permitting, construction and delivery risk in place. It gives a lender contracted demand to underwrite while waste-service revenue keeps the model connected to an existing market.

The financing terms beyond $35M were not disclosed. Vaulted Deep calls it the largest publicly disclosed U.S. commercial debt deal in durable carbon removal secured by long-term purchase contracts. The sharper point is visible: carbon buyers are influencing what can be built, not only what can be credited later.

Mediobanca is financing the handoff from signed tonnes to new sites. Every permit, waste agreement, injection run and verified delivery will show how much of carbon removal's future can be built on contracts banks are willing to underwrite.

Deeper Analysis 👇 https://devcuration.com/articles/vaulted-deep-secures-35m-debt-facility

MTTI

A mouse tumor can disappear, and a drug program can still be years away from treating a person.

That gap is where Molecular Targeting Technologies, Inc. now has work to do. The West Chester, Pennsylvania radiopharmaceutical company announced a five-year, $3.17M National Cancer Institute grant to advance 177Lu-EBRGD for non-small cell lung cancer.

The accounting deserves precision. MTTI describes a $3.17M five-year program. NIH RePORTER currently shows $637,529 for the first budget period, running through July 31, 2027, with the full project scheduled through July 31, 2031. This is an R01 research grant, not an equity round with a valuation and investor syndicate attached.

Koon Yan "Chris" Pak, MTTI's Chairman and CEO, is the contact principal investigator. The job in front of Pak's team is bigger than extending a promising experiment.

177Lu-EBRGD targets integrin αvβ3 on tumor cells and tumor blood vessels. MTTI's EvaThera approach adds an Evans Blue-derived albumin-binding component designed to keep the molecule circulating longer and improve tumor exposure to the lutetium-177 payload.

Published mouse-model research gives that design a real scientific starting point. In NSCLC patient-derived xenografts, 177Lu-EBRGD accumulated more strongly than the non-albumin-binding comparator. One dose eradicated high-integrin-expression tumors during the observation period, while a higher dose delayed growth in low-expression models.

The words "mouse-model" are carrying weight there. The findings do not establish safety, efficacy, dose, or clinical benefit in people.

The NCI grant finances the less cinematic work that determines whether the molecule ever earns a human test: additional preclinical studies, pharmacokinetics, dosimetry, normal-tissue toxicity, cGMP manufacturing, regulatory documentation, an IND submission, and planning for a study in metastatic NSCLC after first-line systemic therapy.

Radiopharmaceutical programs can look beautifully simple in a diagram. Find a tumor target, attach an isotope, deliver radiation. The company still has to make the drug consistently, understand where it travels, measure what healthy tissue absorbs, select patients intelligently, and give regulators an evidence package that survives contact with the details.

$3.17M will not answer whether 177Lu-EBRGD helps a patient. It will pay for the manufacturing, regulatory, and scientific work required to ask that question without pretending the mouse already answered it.

Deeper Analysis 👇 https://devcuration.com/articles/mtti-3-17m-nci-grant-lung-cancer-radiopharmaceutical

KKR / Akrapoint

An equipment lender can read the balance sheet correctly and still misunderstand the machine keeping the borrower in business.

A vacuum truck, specialty trailer, tow rig, or production line is more than collateral. Its working life, resale market, service network, and daily utilization help determine whether the operator earns enough to carry the financing. That is the knowledge gap Akrapoint Commercial Capital was built to close.

Funds managed by KKR have committed $350M through the firm's Asset Based Finance strategy to launch Akrapoint. The Denver company will finance vocational assets, specialty trailers, and industrial equipment for small and middle-market businesses across manufacturing, power, waste services, construction, transportation, and logistics.

The capital accounting matters. Akrapoint did not raise a conventional $350M venture round, and the commitment is not customer volume already deployed. It is balance-sheet capacity for a new lending platform that still has to originate carefully, price risk, and earn repeat business.

The Wall Street Journal reports that Akrapoint expects to finance equipment in the $250K-$5M range, with average loans around $500K-$600K. Those are awkward deals for a lender that wants everything standardized and meaningful deals for operators whose growth may depend on one expensive piece of equipment arriving on time.

Nate Smith is leading Akrapoint as CEO and co-founder after nearly a decade at Trans Lease, where KKR says he ran credit, portfolio management, funding, and compliance while expanding the capital-markets function. Gary Shivers will chair the board after building Navitas Credit Corp into a national equipment-finance platform with more than $1.8B in assets.

That pairing explains the wager. KKR brings long-duration capital and an asset-based-finance business with more than $91B under management. Akrapoint's operators bring the pattern recognition to distinguish a useful machine from an expensive problem wearing fresh paint.

Equipment finance is already a $1.3T U.S. market, and ELFA says July 2026 produced record new-business volume among surveyed lenders. Demand creates room. Higher funding costs still punish sloppy underwriting.

Akrapoint now has to turn equipment knowledge into faster, better credit decisions without letting speed soften discipline. Every truck, trailer, and machine it funds will test whether the lender understood both the asset and the operator expected to put it to work.

Deeper Analysis 👇 https://devcuration.com/articles/kkr-launches-akrapoint-with-350m-for-equipment-finance

Portage

When Broadridge and Fifth Third Bank join a fintech fund as strategic LPs, the relationship carries information in both directions. The fund gets institutions that understand how financial technology is bought; the institutions get a closer view of the companies rebuilding it.

Portage has closed Portage Ventures IV at approximately $600M. The fourth flagship venture fund keeps the firm's seed-to-Series-C focus on insurance, consumer and SMB finance, wealth and asset management, and the infrastructure that makes those businesses work.

The headline number deserves accurate company. Portage Ventures III reached a $616M interim close in March 2022, then finished at $655M that July. Fund IV is therefore a substantial successor vehicle, but it is not larger than the prior final close. Capital accounting gets less applause than a round number. It also keeps the story honest.

Adam Felesky and Paul Desmarais III co-founded Portage in 2016 around a specialist view of financial services. Portage now reports $7B in assets under management, more than 140 portfolio companies, and a platform spanning venture, growth equity, and secondaries. Stephanie Choo, General Partner and Co-Head of Portage Ventures, leads the North American investment team.

That is where the strategic LPs become more than a fundraising footnote. Fintech founders rarely lose months because nobody understands the product. They lose them across risk reviews, compliance questions, integration work, security requirements, budget owners, and the quiet institutional habit of moving carefully even when the market is not waiting.

Strategic LP status does not guarantee contracts for Portage companies. Still, their participation puts financial institutions inside the fund's network as Portage evaluates companies serving the same industry. That proximity can sharpen diligence, expose buyer constraints, and improve conversations between founders and institutions.

Portage says its value-creation team works across go-to-market, technology and cybersecurity, commercial partnerships, business acceleration, and M&A. Those capabilities matter because fintech is full of products that can clear a technical test and stall at the institutional handoff.

The $600M gives Portage another vehicle. The more interesting work begins when a founder with a credible product meets an institution with a real problem, a complicated buying process, and enough context on both sides to keep the conversation moving.

Deeper Analysis 👇 https://devcuration.com/articles/portage-closes-600m-ventures-iv-fintech