The Question After the Win

Standard Metrics

Renewal is the ordinary vote a software customer gets. Standard Metrics has persuaded many of its customers to make a second one and finance the company behind the product.

Standard Metrics raised a $20M Series B led by 8VC, with Salesforce Ventures, Spark Capital, January Capital, and others participating. The company has now raised roughly $50M and says most of that capital came from customers.

That relationship matters because Standard Metrics is trying to become the shared operating layer between private-market investors and portfolio companies. The product has to collect inconsistent data, preserve the trail back to source documents, and turn it into reporting, benchmarks, and answers an investment team can trust. The answer still has to survive review.

Co-founder and CEO John Melas-Kyriazi started Standard Metrics in 2020 after seeing the reporting problem from the investor side: spreadsheets in different formats, KPI definitions that drifted, and quarterly updates trapped in email and PDFs.

AI makes that mess more dangerous before it makes it useful. An agent can answer faster, but speed is worthless when the underlying metric is stale, mislabeled, or impossible to audit. Standard Metrics spent its first chapter building the data foundation. This round is financing the move from collecting portfolio information to making it queryable and operational.

The company says it now supports more than 150 investment firms managing over $400B, alongside more than 10,000 portfolio companies. It also reports approximately 20X business growth since its $23.7M Series A in 2022. Those figures are company-reported, but they clarify what 8VC and the syndicate are underwriting: a network with enough structured private-company data to support AI analysis, benchmarking, and deeper portfolio intelligence.

The capital will expand the platform, improve AI document processing, deepen the AI Analyst and MCP capabilities, and broaden performance insights. One customer reportedly cut quarterly reporting from 15 days to 2 by connecting Standard Metrics data to its AI workflows. Nobody buys infrastructure because the demo knew the word agentic. They buy it because the quarter closes, the auditor stops asking for another file, and an investment committee can see what changed.

The open question is whether customer-backed conviction can become a durable standard across firms that disagree on definitions, workflows, and data ownership. The cap table says buyers want it. The next proof is whether the numbers keep agreeing under real operating pressure as the network gets much larger.

Hike Medical

A custom insole can begin with a phone scan, move into a 3D printer, and ship in days. The referral behind it can still spend weeks moving through faxes, missing documentation, payer rules, and systems that were never taught to share the same patient.

Hike Medical is raising around that handoff. The company announced a Series A that brings its total disclosed seed-and-Series-A funding to $22.5M. Max Altman at Saga Ventures led the financing, with Indicator Ventures, Fifth Down Capital, RiverPark Ventures, other funds, strategic investor and commercial partner Orthofeet, and angels including Sam Blond and Jerod Mayo participating.

The accounting matters. Hike did not disclose the Series A check by itself. The $22.5M covers its reported seed and Series A funding together.

Aadi Bhanti, Steven Chacko, and Amit Bhanti built Hike from an industry they knew before it became an AI category. Aadi Bhanti comes from three generations of orthotics and prosthetics operators. The company now connects 3 jobs that usually sit in different queues.

Hike Intelligence handles referral intake, missing records, payer criteria, approvals, and reimbursement work with AI agents. Hike Clinical gives providers one workflow for evaluation, smartphone scanning, documentation, and ordering. Hike Lab manufactures custom devices in-house at the company's Peoria, Illinois operation.

That combination is the investment thesis. Software can make paperwork move. Computer vision can capture a patient's geometry. A factory can turn the model into a device. The patient only benefits when all 3 arrive in the right order and the provider still gets paid.

Hike says its custom-insole process reduced turnaround from weeks to about 5 business days and pushed remake rates from 1 in 15 to 1 in 400. Those company-reported results help explain why Saga Ventures is backing expansion into orthotics, prosthetics, and durable medical equipment.

The new capital will support engineering and sales hiring in San Francisco, manufacturing roles in Peoria, and a broader device platform. It also gives Hike a harder operating problem. Every added device brings different clinical requirements, payer documentation, fit decisions, fabrication steps, and consequences when the chain misses.

Vertical integration sounds clean in a pitch deck. In device care, it means accepting responsibility for the referral, the claim, the scan, the factory, the fit, and the wait between them. Hike now has $22.5M of disclosed backing to carry that responsibility beyond insoles and through the supply chain.

RQD Clearing

A broker can add overnight trading to an app before the operation behind that button is ready to stay awake.

The customer sees an order move. The clearing firm has to turn that order into positions, margin, custody, records, exceptions, and risk that remain accurate after the interface has gone dark. Extend the trading day, add clients across regions, and introduce digital assets, and the quiet machinery underneath the market starts carrying a much louder responsibility.

RQD* Clearing has raised a $74M minority growth investment for that machinery. Bain Capital Tech Opportunities led the financing, with ABN AMRO Clearing Bank and Nyca Partners participating. Bain joins an investor group that also includes Gentree Fund and Belvedere Strategic Capital.

Brian Donnelly, RQD's Chairman and Founder, built the company around a consequential operating choice: own the clearing infrastructure instead of renting the books-and-records core from a legacy vendor. CEO Michael Sanocki and CTO Jon Fowler can change the platform from the inside because the code, data flow, and operating controls sit inside the business they are scaling.

That control matters when the market refuses to keep banker's hours. RQD says its platform gives broker-dealers, RIAs, and foreign financial institutions real-time visibility across clearing, custody, and risk while supporting 24x5 access to U.S. equity markets. The company plans to use the financing to expand across North America, Asia, and the Middle East, deepen product and technology capabilities, and build further into digital assets, tokenization, and custody infrastructure.

RQD is already carrying meaningful volume. The company reports that year to date it processed more than 543M ledger transactions and approximately 515M equity transactions, representing 69.5B shares, nearly $2T in notional value, and about 2.43% of the NMS equities market. It also reports nearly 64.8M options contracts representing $120.7B in premium and $3.93T in notional value.

Those figures are company-reported, and the announcement does not disclose valuation, ownership terms, individual check sizes, prior-round amounts, or authoritative total funding. The capital creates a clear operating mandate: expand without weakening the control that made proprietary infrastructure valuable.

RQD is financing the move from a modern clearing alternative into infrastructure expected to travel across regions, products, and market clocks. Michael Sanocki and Jon Fowler now have to make every added hour feel ordinary to the institutions depending on what happens after the trade.

Rundoo

An independent supply store can know every contractor in town and still lose margin because the software behind the counter learned nothing from the last sale.

The owner remembers who pays late, what moves before a storm, and which price turns a quote into an order. That knowledge is valuable. It is also trapped in conversations, spreadsheets, old systems, and the heads of people who have worked the counter for years.

Rundoo is putting a $30M Series B behind a different operating model. Battery Ventures led the round, with existing investors Bessemer Venture Partners and CRV participating, bringing Rundoo's total financing to $48M. Battery general partner Michael Brown is joining the board.

CEO Nick Hershey and CTO Andrew Beckman co-founded Rundoo in 2021 for independent hardware, paint, lawn and garden, farm and feed, and lumber stores. The category looks ordinary until somebody tries to run it.

Thousands of SKUs meet vendor catalogs, contractor accounts, special pricing, receivables, seasonal inventory, and customers who expect the person behind the counter to know what the project needs.

Rundoo begins with one system of record across POS, purchasing, inventory, payments, CRM, accounting, e-commerce, and mobile apps. Dooey, its AI agent, sits on that data so a store can ask what sold below margin, which accounts are past due, or what needs reordering.

That sequence matters. AI without a clean record is another screen asking the owner to do the integration work. Rundoo is betting the agent becomes useful because the underlying software already knows the store.

The company says 500+ stores across the U.S., Canada, and the Caribbean use the platform. The capital will expand engineering in Redwood City, go-to-market in Chicago, and product development.

Battery's logic reaches beyond checkout. Michael Brown previously backed ServiceTitan, software for the trades. Rundoo is working one step upstream, with stores supplying those trades. Contractors need materials, stores need better operating memory, and the software must respect workflows built long before anybody called them data.

Every new store type brings its own purchasing habits, supplier relationships, accounting edges, and definition of service. Rundoo now has the capital to carry more of that complexity inside the product, while the owner keeps the part no national chain can install: knowing the town, the trade, and the person at the counter.

Scan.com

A patient can live 10 minutes from an idle MRI and still spend weeks waiting for a scan.

The machine exists. The referral cannot see it, the schedule does not travel, the price changes across town, and the paperwork still behaves like every imaging center is its own country.

That coordination problem is where Scan.com has closed $220M in combined financing. The package includes a $90M Series C led by Noteus Partners, with Aviva, Concord Health Partners, YZR Capital, Oxford Capital, and others participating. VerisFi Capital and Atempo Growth supplied another $130M in debt facilities for M&A and working capital.

The accounting matters. This is not a $220M equity round wearing a more dramatic headline. The larger share is debt, which gives the company acquisition capacity while keeping the equity story focused on network and technology expansion.

Charlie Bullock, CEO, and Oliver Knight, COO, helped turn a 2017 UK project started by clinicians Dr Khalid Latief and Jasper Nissim into an imaging network operating across the UK and nationwide in the United States.

The product has also moved past a directory. Employers, health plans, workers' compensation systems, TPAs, digital-health platforms, and referring providers connect through one network while Scan.com coordinates availability, quality, price, scheduling, paperwork, subspecialty routing, reports, and billing.

That is a far more demanding promise than putting pins on a map. A directory tells a patient where a scanner lives. Infrastructure has to know which machine is available, whether the study is appropriate, which radiologist should read it, what the payer owes, and how the result returns to the workflow.

Scan.com says revenue doubled over the past year to exceed a $165M annualized run rate, more than 900,000 patients have accessed care through its global network, and results typically return within 48 hours. Those are company-reported figures, but they explain why Noteus sees provider connectivity and proprietary data beginning to compound.

The debt changes the next chapter. M&A can add capacity, contracts, local knowledge, or workflow control faster than integrations alone. But acquired operations still have to become one usable network.

Medical imaging does not lack expensive machines. It lacks a national handoff that can move a patient from referral to the right machine, then return the answer before uncertainty becomes its own cost.

TrustedRouter

The router is supposed to give an AI team more choices. It also creates one more company capable of seeing every prompt that explains how the business works.

That is the contradiction behind TrustedRouter's $1.25M Seed round.

Founder and CEO Joseph Perla built one OpenAI-compatible path to more than 600 models from more than 81 providers at the announcement. Developers can route by price, reliability, region, or privacy posture without rebuilding an application when the model market moves.

The flexibility is useful. The position in the stack is sensitive.

A router can sit between an enterprise and the models reading its code, contracts, customer records, research, and unfinished ideas. Add enough providers and models become easier to swap. The router becomes the place that has to explain why it deserves to stay.

TrustedRouter's answer is unusually direct: do not extend the trust. Verify the path.

The company publishes its gateway code, runs the prompt path through hardware-attested confidential computing, and exposes evidence customers can compare with the source. TrustedRouter says ordinary requests are not retained as prompt or output logs. Provider handling remains route-specific, because one attested gateway cannot give every downstream host the same privacy promise.

That distinction is the product, not a footnote from legal.

The Seed syndicate includes Sam Lessin of Slow Ventures, Bill Tai, Linda Avey, George Xing, and other backers named in the announcement. Nobody was identified as the lead. Their experience across product, infrastructure, privacy, and distribution fits a bet on middleware becoming a control point.

TrustedRouter had been in public beta since May. Joseph Perla told Axios the platform crossed 1B routed tokens in a day, and the company says it is bringing on hundreds of customers. Those are company-reported signals, not audited revenue or retention.

They still point toward a purchasing question. Once every task can move to a different model, the value no longer lives only in choosing the smartest endpoint. It lives in deciding who can touch the traffic between them.

$1.25M is early capital for a large claim. TrustedRouter now has to make verifiability survive procurement, provider policy, failure handling, and the moment a security team asks for evidence instead of another promise.

The model market will keep changing names and prices. The route carrying a company's private work has to remain understandable while all of that moves around it.

Odyssey Energy Solutions

A solar installer can sign the customer, win the site, and still be unable to order the panels.

Suppliers want cash before equipment ships. Customers release cash after construction milestones. That timing gap can turn a healthy order book into a waiting list.

Emily McAteer and Piyush Mathur built Odyssey Energy Solutions inside that calendar mismatch. The company has announced $74M in new financing, split between $27M in equity and $47M in debt, to expand across Africa, Asia, and Latin America.

Broadscale Group, FMO, and Al Mada Ventures are new equity investors. Union Square Ventures, Equal Ventures, Abstract Ventures, Twelve Below, FJ Labs, MCJ, and Transition Ventures returned. British International Investment, BIO, the Facility for Energy Inclusion represented by Cygnum Capital, and the Energy Entrepreneurs Growth Fund represented by Triple Jump supplied debt.

Equity can expand the company. Debt can finance equipment orders moving through it. The capital stack is doing two jobs because the customer problem has two clocks.

Odyssey aggregates purchases across smaller engineering, procurement, and construction companies, giving them supplier pricing and embedded credit that would be difficult to negotiate alone. Latitude Media reports that installers can place a deposit, buy on extended terms, and repay as project cash arrives. Odyssey charges equipment fees and retains ownership until payment.

Odyssey says 6,000+ renewable-energy companies use the platform across 50+ countries, with access to $3.6B in capital. Its procurement product has unlocked 1.5 GW of projects. Those company-reported numbers describe the system being built: smaller projects become financeable when procurement, underwriting, and repayment travel together.

The financing follows a $5.34M Seed round and $15M Series A. No new round label, equity lead, or valuation was disclosed. This package gives Odyssey both corporate growth capital and lending capacity.

Odyssey now has to underwrite installers across different markets, currencies, suppliers, and customer payment schedules without letting deployment speed outrun credit discipline. Every panel that reaches a project earlier gives the model evidence. Every repayment teaches the platform whether the next installer can move sooner.

Distributed energy will be built project by project. The financial infrastructure deciding which projects can start together is beginning to look like infrastructure of its own.

Guickly

Prashant Jalan once traced cycles and bytes through AI systems serving billions. Now he is tracing something most enterprises cannot explain: where their AI money went.

Guickly has launched with $4.2M in Seed funding led by Engineering Capital, with Converge VC, Neon Fund, and angel investors participating. The September 1 announcement follows SEC filings showing the offering near $4.2M by May.

Software procurement trained finance teams to count seats. Enterprise AI arrived with subscriptions, APIs, tokens, agents, models, and tools employees can adopt before anyone adds them to a central inventory. The bill grows across departments while ownership of the answer stays politely homeless.

Guickly is building an independent measurement layer across that mess. The platform is designed to identify sanctioned and shadow AI, attribute usage and spending, set budgets, flag unused licenses or expensive models, and show executives whether the activity is producing a business result.

The customer is not buying another assistant. The customer is buying a shared answer that finance, engineering, IT, and security can read without reconciling four different spreadsheets.

That is familiar territory for Prashant Jalan. During more than 8 years in applied AI at Google, he worked on the Google Maps speed-limit feature and built a profiler to optimize TPU performance. Both jobs begin with the same instinct: make a complex system visible enough to measure before trying to improve it.

Engineering Capital led the round. Converge VC and Neon Fund joined it, backing Jalan's technical depth against a problem that has reached the executive budget conversation.

Guickly says prompts, responses, and source code remain on premises, and that it integrates without code changes or an SDK. No customers, audited savings, or verified ROI were disclosed. It still has to prove it can see enough of a fragmented environment to become the trusted ledger beneath it.

McKinsey's 2026 State of AI survey found that only 37% of respondents attributed any enterprise-level EBIT impact to AI, while about 20% said AI operating costs constrained use. Enthusiasm is not the shortage. The missing piece is a clean handoff between activity and accountability.

Guickly is entering that handoff while the budget is still moving. If it works, the most useful AI dashboard may be the one that makes an executive ask fewer questions because engineering and finance finally started from the same number.

Visko

Video AI has learned to make a beautiful moment. Keeping that moment coherent after the user touches it is where the illusion starts filing bug reports.

A prompt change can alter the character, break the room, shift the color, or quietly erase what happened 20 seconds ago. The demo looks cinematic. The world has no memory.

Visko Platform is betting that the product should behave less like a clip generator and more like software that stays awake. The Sunnyvale AI research company has raised a $10M Pre-Seed round led by Llama Ventures and opened public access to Orbis, the foundation model it calls a Live Model.

Orbis keeps generating while a user changes the prompt, carrying the scene forward instead of rebuilding after every request. Qing Yin, Visko's founder and CEO, built the company around the idea that a model should run in step with the world rather than return an answer and stop.

Visko says Orbis streams interactive 4K video at 24 FPS and can sustain hour-scale generation without evident visual-quality or color drift. Its company-authored technical report describes persistent multi-scale memory, streaming super-resolution, and a latent world model that scores possible futures for physically plausible motion.

The report places Orbis first overall in its long-form human-preference and temporal-stability comparisons, with leading results on several DOVER and VideoAlign measures. The evidence is serious, but it was produced by the team selling the system. No independent replication, customer data, revenue, pricing, or cost per running world has been disclosed.

That last gap matters because a live world changes the unit of work. A clip ends. A simulation, companion, game, classroom, or robot-training environment keeps consuming memory, inference, and operational attention for as long as somebody expects it to remain alive.

Visko reports 16 people with experience from Apple, Google DeepMind, Meta, Amazon, and Tesla. Its advisers include UC Berkeley's Michael I. Jordan and Columbia University's Steve WaiChing Sun. Reactor handles distribution, while the commercial range reaches entertainment, education, robotics, and simulation.

The $10M buys Visko time to turn liveness from a memorable demo into a reliable product discipline. Every intervention now becomes a test of memory. Every extra minute becomes a test of infrastructure. And every customer who stays inside the world long enough will teach the company whether continuity is a feature or the market itself.

Senticell

A blood sample can contain the signal a diagnostic needs and lose it during the first processing step.

That is the bet behind Senticell, which closed an oversubscribed Seed round approaching $7M to develop a liquid-biopsy platform built around red blood cells, or RBCs. It is asking whether the part routinely removed from plasma workflows has carried useful disease information all along.

Neha Shah, co-founder and CEO, and Nilam Mangalmurti, co-founder and Chief Scientific Advisor, built Senticell around research from the University of Pennsylvania. The underlying finding is deceptively simple: RBCs can bind and preserve nucleic acids circulating in the blood. If tumor-derived signals attach to those cells, separating plasma from RBCs may also separate a test from biology it could have measured.

TheVentureCity and Capital Factory backed the round alongside Scalpel Ventures, Evoce Capital and high-net-worth investors. The money gives Senticell room to move from an elegant scientific premise into the less forgiving arena of prospective clinical evidence.

About 300 patients are already part of that test. One Penn study is enrolling roughly 200 people with lung and pancreatic cancers and is expected to complete in Q1 2027. Another is enrolling about 100 patients with head-and-neck and cervical cancers across Penn and Penn-Botswana, with results expected in Q4 2026.

Both studies will compare signals captured from RBCs with plasma-based liquid biopsy and tissue analysis. A diagnostic platform earns its place by recovering clinically useful information consistently, not merely by proving that the information exists.

Senticell licensed the platform from Penn, where foundational work also explored RBC-bound nucleic acids in sepsis and pneumonia. The biology opens a wider question: how much diagnostic information has sample preparation trained itself to ignore?

Liquid biopsy has spent years making plasma analysis more sensitive. Senticell is taking the contrarian route by reopening the sample after everyone else decided which part mattered. Nearly $7M now sits behind the possibility that the overlooked fraction is not laboratory waste. It may be an archive.

The next proof will not come from a sharper pitch deck. It will come from those 300 patients, where the red-cell signal either adds something plasma and tissue comparisons can defend, or it does not. That is the useful tension in this round: capital has funded the question, but clinical evidence still owns the answer.

Norbert Health

A robot can reach every room on a nursing floor and still be useless when it gets there.

Navigation handles the hallway. Care begins with knowing what to notice, which question to ask, what belongs in the chart, and when a small change deserves a nurse's attention.

That gap is where Norbert Health has raised a $14M Series A. And the Brooklyn company is building a physical AI control layer that runs on partner robots, giving the hardware a library of healthcare-specific skills.

Alex Winter, co-founder and CEO/CTO, describes the product as the clinical brain and voice behind an autonomous nursing assistant. Norbert-powered robots can make rounds, capture contactless vitals, assess mobility and cognition, handle requests, document encounters, and route alerts.

William A. Marino with Cardinal Group supported the round. Exor Seeds and CareIT participated, alongside Alexis Lê-Quôc, Olivier Pomel, Saeju Jeong, and Thomas Clozel. Norbert reports $19M raised to date.

The capital will deepen its clinical-skill library, advance regulatory clearances, and expand deployments. Work includes post-fall neurological evaluations, pressure-injury prevention, cognitive screening, and fall-risk reassessment.

The operating evidence is promising and still company-reported. Norbert says its robots have been live in skilled-nursing facilities since August 2025, rounding on hundreds of patients daily. It reports 96% patient acceptance and 82% monitoring compliance, nearly 2x its stated manual baseline. No independent deployment audit accompanied the announcement.

That distinction matters because the product remains investigational pending FDA clearance. Healthcare does not award points because a robot completed a lap. The work must be safe, repeatable, documented, and useful enough that patients and clinicians let the machine return.

MedPAC reported 46% median nursing-staff turnover at skilled nursing facilities in 2024. The pressure is real, but replacing a tired human with an unproven machine misunderstands the problem.

Norbert is making the more disciplined bet. Let nurses keep the judgment, relationships, and hands-on care. Give robots the recurring rounds, measurements, structured assessments, documentation, and escalation rules.

Every new clinical skill can travel across another robot platform. It also brings another clear obligation to validate what the machine sensed, wrote, and escalated.

Light

The most durable revenue line in home electrification is the one most hardware companies hand away every month: the electricity bill.

A solar installer can own the panels. A battery company can manage the electrons behind the meter. An automaker can sell the charger. Then a retail provider inherits the recurring customer relationship and the chance to make that equipment more valuable after installation.

Light raised a $46M Series A to change that handoff. Matrix led, Activate Capital joined, and existing investors Spark Capital, Mischief, Gigascale Capital, MCJ, and BoxGroup participated. Announced funding now totals approximately $60M. Light's capital base exceeds $100M only when a credit facility is included, because credit and equity are different fuel.

Co-founders Baker Shogry and Adam Compain built Light so another company can offer a branded electricity plan without spending years becoming a retail-power operator. Light acts as the regulated provider, handling licensing, wholesale procurement, commodity risk, billing, customer support, grid integration, and Virtual Power Plant participation.

Light says companies can launch tailored plans in as little as 2 weeks. It reports that its network reaches more than 30% of U.S. residential solar sales and 500,000 homeowners, while run-rate revenue grew 10x in 12 months. Those figures are company-reported.

Expansion now moves beyond Texas. Light has joined PJM and plans to enter New Jersey, Pennsylvania, and Illinois. It is also broadening its battery offer and launching an EV product combining subscription charging, solar buyback, and conventional electricity service.

Baker Shogry previously led product at Plaid, so the embedded-finance comparison fits. Payments became more useful when companies placed them inside the product. Light is applying that logic to a market with more regulation, physical infrastructure, price volatility, and consequences when the system gets cute.

A partner that controls the electricity plan can shape pricing around the equipment it sold, keep the customer inside its experience, and coordinate devices for grid value. It stops appearing once at installation and starts participating in the energy relationship that follows.

PJM will test whether that model travels. Every market brings different rules, incumbents, and customers who will judge the electricity plan as part of the partner's brand. The hardware has reached the home. Light's $46M question is who owns what happens every month after that.

Aranya

A GPU rack can be worth millions and still be waiting for somebody to turn it into a business.

The chips arrived. The purchase order cleared. Then networking, storage, orchestration, monitoring, scheduling, permissions, and failure handling begin negotiating over who gets to ruin the launch first.

Aranya Inc. has announced $11M across 2 rounds to work on that conversion. First Round Capital led a $9M Seed with BoxGroup, Vermilion Cliffs, and Asylum Ventures participating. Asylum Ventures led the earlier $2M Pre-Seed alongside Founder Collective, Parable VC, and Uncommon Ventures.

That accounting matters. This is a $9M Seed plus a prior $2M Pre-Seed. Aranya's July Form D reports $11.31M sold, including converted securities, which explains the small difference from the rounded announcement.

Christian Bhatia Ondaatje, Sasivarnan Kanaghasalam Sathyapriya, and Aryamika Bhatia Ondaatje founded Aranya around a problem the GPU market made expensive enough to stop ignoring: bare metal is inventory until an operating system can make heterogeneous hardware behave like one production environment.

Aranya's open-source clusterdOS engine builds on Kubernetes and GitOps, pulling storage, networking, observability, inference tooling, VMs, and Slurm environments into one operating layer across facilities.

Aranya says it can move bare metal into a workload-specific production cluster in under 48 hours. Before its first anniversary, the company says it is managing more than $500M worth of GPUs. A Hydra Host case study reports 1,700+ GPUs and a 90% reduction in issue-resolution time.

Those are company-reported results, and the next stage has to show how well they travel across more customers, facilities, workloads, and teams. Infrastructure software earns its reputation on the ugly Tuesday when a network fault, thermal event, or storage problem arrives before the expert does.

The funding will expand engineering and go-to-market hiring and support an AI-native multicluster interface. Natural language may simplify the controls, but the contract underneath is tougher: diagnose the right problem and return expensive compute to work without creating a second incident.

The GPU race started with access. It is moving toward utilization. Data centers can own the racks, AI companies can own the demand, and the economic value still waits in the handoff between them. Aranya now has $11M behind its attempt to make that handoff disappear into the operating layer, one very different cluster at a time.

Clipto

The file was never lost. It was sitting exactly where somebody saved it, which becomes the same thing once the library reaches terabytes.

A creator remembers the shot but not the filename. A lawyer remembers the answer but not the deposition. A researcher remembers the finding but not where the evidence went.

Clipto.AI is turning that gap between storage and recall into a business. The company disclosed $15M in all-equity funding at a $250M post-money valuation from HSG, GL Ventures, EnvisionX Capital, Palm Drive Capital, Hans Tung, Lu Zhang, and 522 Ventures.

Hongwen Henry Kang founded Clipto in 2023 after spending roughly 2 decades circling the same problem. At Carnegie Mellon, Kang worked on robots that remembered where objects were. Later came Zenvideo, acquired by Tencent in 2020.

Clipto indexes video, audio, images, meetings, and documents so users can search in natural language and return to the source moment. Its MCP integration lets AI tools request results from selected folders without open-ended machine access.

Mac and Windows are designed around local-first processing, with source files staying on the device by default. Mobile and web use cloud-powered workflows. This is controlled choice, not a promise that every byte lives under the desk.

Clipto says more than 30M people have used its products, it has hundreds of thousands of paying subscribers, and it reached $15M in ARR while remaining profitable on a net-income basis. Those are company-reported figures, not audited results, but they help explain how a team of just over 20 people raised at this valuation.

Clipto began with video creators. Kang told TechCrunch that creators now represent roughly one-quarter to one-third of users. Lawyers, doctors, researchers, marketers, and students make up much of the rest. The archive problem escaped the editing room because knowledge workers produce media faster than folders can explain it later.

The capital will deepen models and computing infrastructure for consumer hardware, expand AI-agent integrations, and support global growth.

Adobe, Apple, and Google already search libraries they control. Clipto is betting users will pay for memory that crosses file types, tools, and local archives. That position is harder to defend than another AI feature, and more interesting to build.

Reasoning gets the headlines. The durable advantage may belong to the system that recovers the right source, shows its work, and remembers how much of the archive it was allowed to see.

Blue Voice

The highest-stakes AI feature may be the one that tells the model what it is allowed to know.

A police officer asking a question on the side of the road does not need a confident tour of the internet. The officer needs the current statute, the department's policy, the local procedure, and a source that can be opened before a judgment becomes somebody else's emergency.

Blue Voice Inc. has emerged from stealth after raising $6M in venture capital over 3 years. SignalFire and Las Olas Venture Capital led the financing, with Govtech Ventures and PAR Capital Management also backing the company. That accounting matters: Blue Voice disclosed $6M accumulated across its first 3 years, not one freshly closed $6M round.

David Roosevelt Lawrence, CEO; Amit Patankar, CTO; and Michael Gropman, Chief Business Officer, founded the Boston company. Their backgrounds span law and government, Google Brain, and more than 30 years in law enforcement, including command experience during the Boston Marathon bombing.

That mix produced a useful product decision. Blue Voice does not search the open web for a plausible answer. Each department supplies the law, policy, procedures, forms, maps, and operational documents the system is allowed to use. The response returns quoted source text and a path back to the governing document, while the trained officer keeps responsibility for the decision.

This is vertical AI with an unusually honest boundary. The value is not that the software sounds authoritative. The value is that an agency can inspect where the answer came from, control which material enters the system, and update the source when law or policy changes.

Blue Voice says more than 225 public-safety agencies across 25 states now use the platform. It reports more than 350,000 questions answered, roughly 1 every minute, and 10X growth over the past year. Those company-reported numbers put field evidence before the familiar funding announcement.

The difficult work now lives inside procurement, trust, and maintenance. Every new agency brings another jurisdiction, another policy library, another security review, and another set of edge cases where traceability has to survive contact with a live call.

Blue Voice is entering that room with capital, reported adoption, and a product built around a controlled source boundary. The next chapter will be written department by department, wherever speed, policy, and public authority meet before anyone has time to open the binder.

HiBob

The org chart was never designed to become machine-readable infrastructure. That is exactly where a $166M investment is pointing.

Salesforce is leading the financing in HiBob, with Farallon Capital Management participating. Bloomberg reports that Salesforce supplied most of the capital and that the round values HiBob above $3.2B. HiBob did not assign the financing a Series label or confirm the valuation in its announcement.

AI agents can retrieve information and trigger actions. They still need to know who owns a decision, what that person can approve, where a team has capacity, and who remains accountable when software moves first.

That context usually lives in the least glamorous corner of enterprise software: the people system.

Ronni Zehavi and co-founders Israel David, Andy Bellass, and Amit Knaani started HiBob in 2015. The company built Bob around HR, payroll, finance, performance, compensation, workforce planning, and the relationships connecting employees to teams and managers. HiBob now says more than 5,500 customers use the platform across 170+ countries.

The company is turning that installed position into what it calls Organizational Intelligence. The idea is to let trusted workforce context move into collaboration, CRM, finance, operations, and AI workflows instead of remaining trapped inside an HR application. HiBob's expanding Slack integration and MCP work give Salesforce a direct view into that possibility.

Salesforce has spent years teaching software to understand customers and commercial workflows. HiBob holds another layer: roles, reporting lines, skills, permissions, capacity, and accountability.

Farallon led HiBob's $150M financing in 2023, when HiBob reported 3,500 customers, more than $100M ARR, and $574M total capital. HiBob says the new money will accelerate product, technology, AI, and innovation; CTech reports acquisitions and platform expansion may also be part of the plan.

The opportunity is larger than adding a chatbot to HR software. As agents take on work across departments, companies will need a reliable way to decide which tasks belong to people, which belong to software, what each actor can access, and where responsibility returns when the handoff breaks.

HiBob is betting that the people system can become that operating context. Salesforce is now putting serious capital behind the same idea. The next chapter will be written inside the messy organizational details that every polished AI demo politely skips.

iPipeline / Origo

A pension transfer can be digital at both ends and still spend its worst days trapped in the handoff between them.

One provider has the policy record. Another has the investment platform. An adviser has the client instruction. The customer has a retirement deadline that does not care which system forgot to call back.

iPipeline has completed its acquisition of Origo after receiving what it describes as all required regulatory approvals, including UK Competition and Markets Authority approval. The purchase price and transaction structure were not disclosed.

The transaction brings together 2 companies circling the same problem from different sides. iPipeline builds digital workflows across insurance, annuities, wealth, pensions, and investments. Origo connects UK providers, platforms, advisers, and customers so the data inside those workflows can move.

Pat O'Donnell, CEO of iPipeline, is buying more than another software product. Origo was established in Edinburgh in 1989 by pension providers and investment-management firms to remove shared friction. Anthony Rafferty, CEO of Origo, now takes that mission inside a larger Roper Technologies business.

Origo says its Transfer Service handles about 95% of UK defined-contribution pension transfers, while Unipass services are used by 8 in 10 UK financial advisers. Those company-reported figures explain the asset iPipeline wanted: infrastructure that already sits between institutions.

This was not a blind date arranged by an investment banker. In 2023, the companies connected Origo services with iPipeline's SSG Digital platform. The acquisition puts ownership around an integration the market had already started testing.

Financial services modernization has a habit of making the screen prettier while leaving the handoff where it was. A faster quote does not fix a broken transfer. A modern adviser portal does not help if identity, policy, account, and valuation data still have to be rekeyed.

iPipeline says Origo and Unipass will remain available, connected, and partner-friendly. That promise carries the integration work. Origo became valuable as shared infrastructure across a fragmented market. Bringing it inside one vendor works only if customers and partners still trust the connections running through it.

The next proof will arrive quietly: fewer missing instructions, fewer manual chases, cleaner transfers, and less time explaining to a customer why 2 digital systems still needed a human to carry the message.

Numeral

The sale is the clean part. Sales tax begins after the customer pays, then follows the business into every threshold, registration, filing calendar, exemption certificate, and letter from a revenue department.

That operating mess is what Numeral turned into a company. In September 2025, the tax-compliance platform announced a $35M Series B led by Mayfield at a reported $350M valuation. Benchmark, Uncork Capital, Y Combinator, and Mantis participated. Disclosed funding reached $57M.

Sam Ross knew the pain before building the product. The co-founder and CEO had run ecommerce businesses generating more than $50M in revenue, learning how quickly one transaction could become obligations across dozens of states. Co-founder and CTO Matt DuVall brought engineering experience from Stripe, Notion, and Box to a problem where a clean interface has to survive government portals and finance-team scrutiny.

Numeral monitors nexus, handles registrations, calculates tax, files returns, remits payments, manages exemption certificates, and routes government mail. Customers are not buying a prettier calculator. They are buying an operating layer for work that keeps multiplying after checkout.

At the time of the Series B, TechCrunch reported more than 2,000 customers and 3.5x year-over-year revenue growth. Mayfield reported more than $5B in processed volume and more than 150,000 returns filed. Numeral's current materials say the company now serves more than 3,000 businesses across 70+ countries.

Those are company and investor-reported scale signals, not an independent audit of accuracy, retention, or economics. That distinction matters because tax automation carries a different consequence than most workflow software. A missed task is not just an annoying notification. It can become a penalty, an audit issue, or a market-expansion delay.

Numeral answers that pressure with software, tax experts, and a guarantee that covers penalties and interest when a filing is late. The model pulls the company toward something more demanding than SaaS convenience: accountable execution.

The $35M gives Numeral room to deepen that model while indirect-tax complexity keeps spreading across products and borders. The open question is whether a hybrid of agents, experts, and guarantees can absorb more responsibility without rebuilding the service-heavy economics software promised to escape.

Checkout keeps getting easier. The obligations arriving after it are learning how to scale too.

Principal / Beam Benefits

Small businesses rarely experience employee benefits as a product catalog. They experience the wait between asking a broker for a quote and getting something an owner can explain to employees without opening 3 portals and a spreadsheet.

That handoff is what Principal Financial Group acquired when it completed its purchase of Beam Benefits on September 1. Terms were undisclosed. The operating footprint was not: more than 25,000 small-business customers, over 200 employees, digital quoting and onboarding, and approximately $175M in 2025 premium volume moving into Principal's Benefits and Protection business.

Amy Friedrich, president of Benefits and Protection at Principal, has framed the deal as a way to accelerate growth with small and midsized businesses. Principal says Beam can help Specialty Benefits premium and fee growth reach at or above the high end of its 5%-9% medium-term target range in 2027. That is a company target, but it explains why this purchase reaches beyond another product shelf.

Tolithia Kornweibel became Beam CEO in 2024 after the company spent more than a decade turning a connected toothbrush idea into a national ancillary-benefits platform. Beam raised more than $200M along the way and expanded into 46 states and Washington, D.C. It learned that the buyer of benefits technology is often trying to remove work, not add software.

That history makes the integration delicate. Principal is buying Beam because the smaller company made a slow category move faster. The acquisition works best if Principal's distribution, balance sheet, and product breadth amplify that speed without surrounding it with enough process to recreate the problem Beam was built to remove.

For brokers and customers, the companies are promising continuity while they work toward a unified offering. For the employees who transitioned, the work now sits inside a 146-year-old financial institution serving more than 82 million customers. Those facts create the opportunity and the pressure together.

Ancillary benefits are becoming an operating layer where the carrier that makes quoting, enrollment, administration, and service easier can win the small-business relationship before price comparison gets the final word. Principal now owns more of that layer. The value will appear in ordinary moments: a faster quote, one fewer manual handoff, and Beam's digital habits surviving long enough inside Principal to change how the larger company moves.

SciFin

The forecast review begins with a number. Then the room spends an hour negotiating whether anybody believes it.

The CRM has a version. The call transcript has another. A representative remembers that the customer changed direction. Finance is looking at a forecast assembled before that happened. RevOps becomes the human adapter between systems sold as one source of truth but delivered as partial truths.

Mohit Aron has lived inside that reconstruction work from both sides of the table. The early Google engineer co-founded Nutanix, founded Cohesity, and spent 2 companies making scattered information behave like a system. His third company, SciFin, points that instinct at revenue teams.

SciFin emerged from stealth with $44M in disclosed funding. Madrona describes the latest seed as $38M, co-led with Altimeter. Foundation Capital, S32, Zetta Ventures, and others participated. The official release calls the full $44M seed funding, while Aron calls it the total including the latest seed. The footnote belongs in the story.

The company calls the problem the Context Gap: the distance between what systems report and what is happening across accounts, deals, forecasts, territories, conversations, documents, reviews, and institutional memory. SciFin is building a maintained context layer across that mess. Pixie turns the current picture into answers, reports, and recommended actions. Madrona describes a live context graph with historical snapshots, conflict resolution, and role-based access control. Those are investor-described capabilities, not independently audited outcomes.

Enterprise AI has spent 3 years showing how quickly a model can answer. Revenue organizations are about to spend the next chapter asking whether the model had enough current business truth to deserve the authority. A faster answer built from yesterday's deal reality is still late.

SciFin says the capital will support product development, go-to-market expansion, and customer growth. It has not disclosed customers, revenue, valuation, or an independently measured result. This is a large early bet on founder history and a problem scaled commercial teams recognize.

The commercial handoff is unforgiving. SciFin has to earn access to the calls, documents, systems, and judgment that explain how a deal is moving, then keep that context current enough that managers use it before a forecast slips. Value arrives when the organization stops treating SciFin like another dashboard to reconcile.

Aurenity / Great Hill Partners

Software can clear a submission queue. It cannot decide whether a carrier should accept the risk waiting at the other end.

That distinction sits underneath Great Hill Partners' strategic investment in Aurenity, a West Hartford-based managing general agent built for the excess and surplus insurance market. Financial terms were not disclosed. Founding investor Agman and Aurenity's management team will retain significant equity.

Aurenity launched in 2022 with a $10M seed investment from Agman and a pointed premise: technology only helps when it sharpens the judgment behind the quote.

The company now operates 6 core E&S programs across primary, lead and excess casualty, public entity and religious risks, and property. CEO Nick Davies, CUO Douglas Trainor, COO Patrick Safino, and their line leaders have assembled the specialist underwriting bench behind that expansion.

Great Hill's capital is intended to add underwriting talent, launch specialty programs, and deepen the systems and AI supporting the platform. Matt Vettel, Nick Cayer, and Bob Anderson will join Aurenity's board.

That people-and-systems combination is the transaction. The E&S market exists for risks that do not fit neatly inside standard insurance appetites. More unusual exposures create more room for specialist underwriting, but they also punish any platform that confuses faster intake with better selection.

Aurenity has already reported one useful operating result. Its work with Heron doubled submission capacity and reclaimed more than 1,000 hours annually without adding headcount. The metric matters because it shows where automation can earn its keep: cleaning and routing the work so experienced underwriters spend more time deciding which risks deserve capacity.

The wider market is large enough to reward that discipline. NAIC says U.S. surplus lines direct premium reached roughly $131B in 2024, up 12.2%. As more complex risks move outside admitted markets, MGAs can become important operating bridges between wholesale distribution, underwriting expertise, and carrier capital.

Aurenity is now adding fuel to both sides of that bridge. New programs will expand the surface area. New systems will increase the speed. The value will be decided in the quieter moment when a carrier sees the portfolio, understands the judgment behind it, and chooses to provide capacity again. The machinery can move faster. The judgment still carries the loss.

Palo Alto Networks / Console

Every autonomous security demo reaches the handoff: the model knows what should happen, then a human opens 4 systems to make it happen.

Palo Alto Networks has acquired Console, the AI-native service-management company built to turn operational intent into governed action. The companies did not disclose the price, valuation, or consideration. Console will deepen agentic capabilities in Cortex, helping teams investigate signals, prioritize work, and take action across the enterprise.

Security platforms already see endpoint activity, identity behavior, cloud posture, traffic, alerts, and threat intelligence. Acting on that view requires another layer. Which identity, device, policy, and approval chain applies? What can an agent change, who must approve it, and what gets logged?

Andrei Serban, Co-Founder and CEO, and Neal Chandra, Founder and CTO, built Console around that context. Natural-language Playbooks connect people, apps, devices, tickets, and policies so agents can resolve requests, investigate issues, execute configured actions, or escalate the work. Integrations do not grant write access by default. Approvals and audit records remain inside the system.

Founded in 2024, Console disclosed a $6.2M Seed from Thrive Capital and a $23M Series A from DST Global Partners and Thrive Capital. Console says customers including Databricks, Webflow, Scale AI, and Flock Safety automate more than 50% of support requests without human intervention. Those company-reported results explain why Palo Alto Networks saw an operating layer instead of a prototype.

For Nikesh Arora, Chairman and CEO of Palo Alto Networks, Console fits a larger assembly. Koi added agentic endpoint security, Portkey added AI gateway governance, CyberArk expanded identity, and Chronosphere expanded observability. Console adds the workflow layer connecting a signal to a controlled response.

An autonomous SOC does not earn trust by sounding confident. It earns trust by knowing the organization well enough to act inside its boundaries. A brilliant recommendation that still lands in a ticket queue has only moved the waiting room.

Palo Alto Networks brings more than 70,000 customers and $9.10B in Next-Generation Security ARR. Console brings evidence that enterprise agents can move beyond chat into operations.

As Cortex absorbs Console, the consequential work will happen between a recommendation and a permission, where identities, policies, approvals, and real systems begin moving together.

Metal

A founder can spend months raising capital and discover that most of the conversations should never have happened.

The investor looked right in a database. The stage, sector, geography, check size, or timing did not. By the time the mismatch becomes obvious, the founder has burned the introduction and another week that belonged to the company.

That wasted motion is what Usman Gul is building Metal around. Metal announced a $4.5M Seed financing backed by a16z speedrun, Y Combinator, Pioneer Fund, Rebel Fund, Gaingels, Indus Valley Capital, Team Ignite Ventures, Phaze Ventures, and others.

Metal calls itself an AI-driven operating system for founders raising venture rounds. The useful part is not another list of investors. It is an investor memory layer designed to understand which firms fit a company, why they fit, which relationship paths may work, and how that context should shape the round.

The platform ranks investors, surfaces warm-introduction paths, supports research and collateral, manages pipelines, and automates more round administration. Founders still own the financing decision and the relationship. Metal is removing the hours spent rebuilding context before every email and meeting.

That distinction carries Usman Gul's operating history. Before Metal, he says he raised $120M across 6 Airlift rounds and spoke with roughly 300 investors. About 80% were not a fit for the company's stage, sector, geography, or check size. The scar tissue became a product requirement.

Metal reports that paying customers raised $542M in FY2026, more than 100 paying users close rounds each year, and more than 1,000 founders raise through the platform annually. Y Combinator says 100+ YC founders have signed as paying customers. Those are company-reported figures, but they explain why this round backs more than an investor search box.

Last month, Metal launched Investor Relations, its first product meant to stay after the active raise. The bet is that fundraising, investor relations, ownership, and company-investor communication are stages of one relationship whose context usually gets dropped between systems.

Static databases made private-market information easier to find. Metal is aiming at the judgment after the search: who belongs, what the relationship knows, and which action should follow.

If that memory keeps improving across rounds, the founder's next capital decision may begin with the context the last one created instead of another empty pipeline and a fresh spreadsheet.