When the Answer Reaches the Line

Hatz AI

Jimmy Hatzell wants MSP employees to become AI builders. Hatz AI is putting more engineering help beside the people who already answer their customers' technology questions, then asking them to take on a more complicated assignment.

An IT provider can know the customer well and still need help turning that knowledge into working software. Which task should an agent handle? Whose information can it use? Who supports the workflow when the business changes? Those questions arrive together, usually with somebody expecting an answer.

The New York company announced a $24M Series A on November 12, backed by Nadia Partners, Vestigo Ventures and The Private Shares Fund. Nadia partner Josh Schertzer led the deal. CEO and co-founder Jimmy Hatzell built Hatz AI with co-founder Aidan Kehoe around a deliberate choice: serve businesses through their managed service providers.

That choice keeps the MSP in the customer relationship. Hatz supplies the platform and specialized support; the provider builds an AI practice under its own brand. The company says it has no competing direct-to-enterprise sales motion. A partner investing in the skills to deliver this service can therefore keep developing the account it brought along.

There is technical work underneath that commercial promise. Hatz brings model access, automatic routing, governance and custom workflows into a platform for multiple client organizations. Its integration documentation makes an important distinction: sharing an agent does not automatically give everyone the creator's external credentials. Someone still has to decide what that agent may reach.

The company reports that its active MSP partner count tripled year over year and that it has issued more than 15,000 AI certifications. Those are company-reported measures of network growth and training activity. They give the expansion some shape without telling us how much value each customer is getting.

Hatzell told CRN the company intends to double its roughly 60-person team, concentrating hiring in engineering, AI research and forward-deployed engineering. He also described helping MSPs train builders internally. The people closest to the client are being asked to develop more of the solution themselves.

That is where this round gets interesting for an MSP owner. Training an employee changes which customer problems the firm can accept, how it staffs the work, and who is available when an AI workflow needs attention after the sale.

Deeper Analysis 👇 https://devcuration.com/articles/hatz-ai-raises-24m-series-a-for-msp-ai-builders

XpertDox

XpertDox charges by the claim, so a clinic waiting to go live is a customer whose volume has yet to reach the meter. Getting another provider connected belongs inside the growth strategy. The implementation work keeps running after the sale.

Sameer Ather, MD, PhD, and Bhupesh Panwar built XpertDox around the work between a clinician's note and a billable claim. The physician co-founders now have a $20M Series A, announced November 11, led by Savant Growth with funding from its investment partner, Kennet Partners.

The Scottsdale company plans to expand onboarding capacity, deepen specialty coverage and grow its engineering and customer teams. For a supplier paid on processed claims, helping the customer start using the product is part of building the business that can support the next deployment.

XpertCoding combines neural networks, ensemble models, symbolic reasoning and payer rules to turn clinical documentation into coded data. The company says it codes more than 91% of received claims without manual intervention, returns them in under 24 hours and keeps coding-related denials below 2%.

Those figures describe different parts of the workflow. Automation measures how much work moves without manual handling; a denial still has a reason somebody needs to understand. XpertDox's BI platform includes an audit trail, documentation feedback and a queue for claims held for review. Staff can see where the software needs attention and take the question back to the clinical record.

Founded in 2015, XpertDox reports doubled revenue over the last 12 months, 126% net revenue retention and 95% customer retention. It serves 7 federally qualified health centers across 6 states, according to the announcement. These are company-reported results, with no absolute revenue disclosed.

A community health center brings medical, dental and behavioral care into the same organization. That breadth follows the documentation into the revenue cycle, giving the coding supplier several kinds of work to understand before a new site can rely on it.

Javier Rojas, Savant Growth's co-founder and managing partner, has joined the board. Bhupesh Panwar, XpertDox's President and COO, is also directing the company's own data and AI workflows toward serving a larger customer base at lower cost per customer.

The next health center needs its claims moving while its clinicians keep seeing patients. Onboarding capacity is where XpertDox's expansion plans meet that working day.

Deeper Analysis 👇 https://devcuration.com/articles/xpertdox-raises-20m-series-a-ai-medical-coding

Imprint

A brand can remember your favorite purchase and still lose the next one to a shopping agent that knows nothing about your rewards. Imprint is putting that missing information within the agent's reach.

The New York fintech raised $130M at an approximately $1.9B valuation in a November 11 financing led by Khosla Ventures. Hedosophia, Ribbit Capital, Spice and Thrive Capital returned, with Positive Sum joining. The company is expanding payment products and AI capabilities around a relationship retailers have spent years paying to build.

Founder and CEO Daragh Murphy co-founded Imprint with Gaurav Ahuja. With CTO Will Larson, the team operates technology spanning applications, risk, servicing and rewards. Owning those connections gives Imprint room to change how a brand's card works, beyond changing the offer printed above the application button.

Its released Model Context Protocol server lets AI agents access members' rewards, virtual cards and transaction history, with tools for members to review and authorize agent-initiated purchases. A reward can become part of the shopping decision even when the customer starts somewhere other than the retailer's app. The price comparison becomes more personal when it includes value the shopper has already earned. That creates a new job for loyalty teams: make the benefit understandable to the software doing the comparing.

Imprint says it serves more than 2M members and launched programs this year with Albertsons Companies, Fanatics, Kroger and Shell. It reports that total cardholder spending more than doubled between 2025 and 2026. Those figures describe the company's existing reach; adoption and transaction volumes for the agent tools were not disclosed.

Imprint Access opens another route into that relationship. Eligible customers new to credit or rebuilding it can use a refundable deposit to access the same ongoing rewards as the brand's unsecured card. Responsible use can help them work toward an upgrade. A customer already buying from a retailer may finally have a way to join its card program.

The $130M follows a $150M Series D in December 2025. Separately arranged lending capacity finances card balances, giving the platform's product budget and its receivables different sources of capital. As more shopping moves through assistants, Imprint's partners have to decide which benefits deserve to travel with the customer, and how to earn the next purchase once those benefits become easier to compare.

Deeper Analysis 👇 https://devcuration.com/articles/imprint-raises-130m-brand-loyalty-ai

Duet

Nursing school did not come with an MBA. Duet is building around what happens when the nurse practitioner becomes the person responsible for keeping the clinic open.

The patient needs an appointment. The owner needs credentialing, clean claims and an insurer contract that makes the numbers hold together. Being good at primary care creates plenty of work before anyone gets paid for it.

Founder and CEO Jonathan Goldberg has raised a $15M Series A for Duet, led by Kairos HQ with Lerer Hippeau participating. Announced November 12, the financing brings total funding to $22.5M and backs a network the company says already spans more than 130 independently owned clinics across 21 states.

That ownership detail gives the business its shape. Duet supplies technology, practice advisors and operating support around clinicians who retain their own practices. Its services run from billing and coding to patient recall, bookkeeping and payer negotiations. An owner can need help collecting the money and understanding where it went.

The software is close to the clinical day. Duet describes tools that find patients who have drifted from care, help get them scheduled and surface issues before an appointment. Its insurer offering also brings utilization, quality and patient-attribution information together across practices. That administrative work helps organize care.

Insurers including Anthem, Cigna, Centene and Molina contract with Duet as a network. A solo practice gets access to arrangements that would be difficult to win alone, while Duet says it takes financial risk in many of those contracts. The shared capacity comes with obligations.

The company reports a 30% increase in preventive screenings and 20% fewer ER visits among participating practices. Those are company-reported results, with no independently audited comparison supplied in the announcement. They explain the direction Duet is pursuing without promising the same result for every clinic.

The Series A will support more clinics, deeper insurer partnerships and engineering hires. Duet also plans a nurse practitioner-led Medicare ACO for 2027, giving its ownership model a larger shared-savings assignment.

For the NP who wants to stay independent, the next contract will reach well beyond the reimbursement for today's visit. Duet is asking those owners to build enough collective capacity to participate in that negotiation while the name on each practice's door remains their own.

Deeper Analysis 👇 https://devcuration.com/articles/duet-raises-15m-to-support-independent-np-owned-clinics

Tristar AI

Tristar AI is putting part of its new funding into the people who help a factory use the software. Customer success and deployment teams belong in this $12.4M Series A story as much as the computer vision models.

A camera can find a defect. Someone on the line still has to understand the alert, change the process and keep the next batch from inheriting the same problem. Selling across several plants means earning that response more than once.

CEO and co-founder Salem Karani grew up around his family's plastic-film plant, where an error caught too late could consume a month's margin. Co-founder Jack Liu and CTO Justin Au-Yeung are building with him around a familiar manufacturing frustration: discovering what went wrong after the expensive part has already happened.

The Cambridge, Massachusetts company announced the round on November 12, led by Isogon Ventures. Blackhorn Ventures, Ford Street Ventures, ISAI Cap Venture, Capgemini's VC fund managed by ISAI, Las Olas VC, NextView Ventures, Seraph Partners, SNR Ventures and TenOneTen participated.

Tristar inspects products during production and sends operators live feedback. Its platform connects with existing cameras, PLCs, sensors and manufacturing systems. The buyer gets to discuss scrap, throughput and customer claims without first agreeing to replace the factory's technology estate.

There is a useful example in Tristar's steel-stamping case study. The company describes an alert arriving within 10 seconds of a defect spike during a night shift, giving operators time to stop the press before more than 10,000 defective parts were produced. Those are company-reported results from a particular deployment, with people taking the action that made the warning valuable.

Tristar says it now operates across more than 500 production lines in North America and Europe. The new capital also supports its AI vision and edge-computing platform, entry into additional manufacturing sectors, and sales and marketing.

That expansion gives the deployment team a demanding commercial assignment. A plant already using the system can introduce it to another facility, but the next site's products, processes and operators still need attention. Customer success starts carrying part of the engineering promise. Support has to reach the people working that shift.

For Salem Karani's team, another plant means another group of operators deciding whether the next alert deserves an interruption.

Deeper Analysis 👇 https://devcuration.com/articles/tristar-ai-12-4m-series-a-factory-inspection

Databento

Christina Qi had a profitable market-data business and a $97M Series B behind her when customers started asking Databento to expand faster. Four months later, the co-founder and CEO has taken another $100M to bring more of that work forward.

The Series C values Databento at $1.1B. Fifth Down Capital, which invested before the company had revenue, led the round. Fitch Ventures joined as a strategic investor, putting a financial-information institution alongside a company built to make exchange data easier for developers to buy and use.

Christina Qi and co-founder Luca Lin came out of quantitative trading. They understood how a research idea could get buried under the work of obtaining data, cleaning it and making it usable in production. A self-service API changes that first encounter with the supplier. The engineering behind the answer still has to happen somewhere.

Databento delivers real-time and historical data across futures, options, equities and indices. That means physical servers, specialized hardware and systems capturing market activity before an application can ask a question about it. The convenience at the customer's end carries a substantial assignment at the provider's end.

That customer is changing. Databento says it has enterprise contracts with 5 of the 6 leading frontier AI labs, alongside the trading firms, brokerages, asset managers and commodities businesses it already serves. The labs remain unnamed in the announcement, but the demand brings machine consumption directly into the financing story.

The company reports that revenue has more than doubled since the Series B while it remains profitable. It closed this round with 39 employees. Those figures put an interesting obligation on a small team: extend a physical data network while preserving the precision that persuaded demanding customers to use it.

NEA, DRW Venture Capital, Tribe Capital, Redpoint Ventures, Hedosophia, Trousdale Ventures, Cross Creek, Operator Collective, Alumni Ventures and TGVP also followed on. Existing investors are helping finance a wider version of a business they have already watched operate.

Databento says buildouts at 9 data centers are nearly complete. Its ambition reaches toward the largest commercially available historical market-data archive, with more markets and asset classes still to cover. Each addition gives another developer more history to work with, while giving Christina Qi and Luca Lin more of the world's trading activity to capture faithfully.

Deeper Analysis 👇 https://devcuration.com/articles/databento-raises-100m-in-series-c-funding

Delphina

At Substack, letting executives ask their own data questions gave the analytics team something useful in return: a view of what leadership wanted to know. The questions became part of the work, instead of disappearing into a private exchange.

That detail, described in Delphina's customer story, gets at the business Jeremy Hermann and Duncan Gilchrist are building. Faster answers change who participates in a decision. They also expose how much agreement has to sit underneath an ordinary number.

Delphina announced $7.5M in new funding on November 12, taking total funding to $15M. Radical Ventures and Costanoa co-led, with Tenacity Capital participating alongside angels including Fei-Fei Li and Guido Imbens. The latest round was not given a public series label.

Jeremy Hermann, co-founder and CEO, built Uber's Michelangelo platform and co-founded Tecton. Duncan Gilchrist, co-founder and President, led global pricing data science at Uber. They have seen what happens when people across a business can get close to the numbers, and how much engineering it takes to keep those numbers useful.

Their Context Factory puts agents to work building and maintaining the knowledge behind AI analytics. Trusted tables, metric definitions, the exception an experienced analyst remembers: somebody has to keep that understanding current as the company changes.

When the system finds uncertainty or conflicting definitions, it investigates, drafts a correction and brings the relevant expert into the decision. Approved knowledge becomes part of the context used for later answers. The expert's judgment can travel further than the original conversation.

Delphina says business teams now ask 73% of platform questions, and CEOs at more than half its customers are active users. Those are company-reported figures. The customer list spans Substack, LATAM Airlines, Gen Digital, Medely and Coursera, putting very different operating questions around the same requirement for dependable meaning.

The funding goes toward scaling Context Factory and expanding go-to-market and customer engineering. That last function belongs in the story. An enterprise already has its own vocabulary, permissions and people who know why a definition looks strange.

At Substack, shared questions give the data team a chance to catch assumptions early. As Delphina reaches more desks, the person approving a definition is helping shape a conversation across the business they may never be in.

Deeper Analysis 👇 https://devcuration.com/articles/delphina-raises-7-5m-context-factory

VulcanForms

Martin Feldmann and John Hart kept the machines. VulcanForms Inc. makes customers' metal parts inside its own factories, taking on the industrial bill that a printer manufacturer can hand to the buyer.

That choice has grown into a $500M financing agreement led by J.P. Morgan and funds managed by Apollo Global Management, Inc., announced November 12. VulcanForms plans 3 new factories alongside its 3 existing manufacturing sites. The Massachusetts company says multibillion-dollar commercial programs are driving the expansion across medical devices, consumer products, aerospace, defense, space and industrial markets.

The customer gets a different purchasing decision. Instead of buying a metal printer and assembling the rest of the production system around it, an engineering team can bring VulcanForms a component to manufacture. Somebody still has to connect materials, printing, machining, inspection and delivery. VulcanForms has chosen to own that work.

CEO Kevin Kassekert has spent years around the consequences of those decisions. His background includes Tesla's factory infrastructure and the COO role at Redwood Materials. CTO Mike Kenworthy brings experience from Relativity Space and Seurat Technologies. Hart remains a co-founder, board member and MIT professor, keeping a connection between the research and the production floor.

Founded out of MIT in 2015, the business combines metal additive manufacturing, precision machining, automation and software. The new financing also supports deeper physical-AI integration. An interesting machine becomes a much larger business when the customer can keep ordering the part it produces.

This follows a $220M financing announced in January, led by Eclipse and 1789 Capital. VulcanForms now reports more than $1B in funding from lenders and investors combined. The latest announcement describes a financing agreement, with no valuation or detailed instrument terms disclosed.

There is a useful commercial distinction inside those big numbers. Company-reported programs describe demand for future production; they are not a public revenue statement. The factories have to turn that demand into components a customer can accept and use.

JPMorganChase places the transaction inside its Security and Resiliency Initiative. For VulcanForms, that national ambition arrives as a very local assignment: equip another factory, connect its processes and make room for the next customer's production run.

Deeper Analysis 👇 https://devcuration.com/articles/vulcanforms-500m-financing-three-new-factories

Mirendil

Choosing what an AI model reads is still somebody's job. Mirendil's newest data role puts a researcher in charge of the training mixture, with agents running the experiments that help decide what belongs in it.

That division of labor runs through Mirendil's $1B Series A, announced November 12. Kleiner Perkins and Clearlake Capital Group led, with Battery Ventures, Doerr Capital, Gil & Co, Google, M13, Pear VC and S32 among the participants. Mirendil had announced a $200M seed in June, co-led by a16z and Kleiner Perkins.

Co-founder and CEO Behnam Neyshabur and co-founder and CTO Harsh Mehta built the company with Shayan Salehian and Tara Rezaei. Their earlier work spans Google, Anthropic, xAI and OpenAI. Mehta had already built an early automated research platform inside Anthropic, giving this team some experience with the machinery it now wants other labs to use.

Mirendil says one researcher, working with agents, pretrained its first coding model in 3.5 months with less than $10M in compute. The researcher chose directions and decided which results to trust. That is a company-reported experiment, with a skilled person very much inside it.

Fern adds up to 10M tokens of context. Lor, now in private beta, coordinates agents that write code, run experiments and keep working across machines for days. Together, they give Mirendil a way to sell more of the research process than a model response alone.

The commercial work is already taking shape. A founding GTM vacancy asks someone to design customer evaluations, agree on success criteria, work through procurement and turn pilots into paid use. The finance vacancy says Mirendil is starting to bill enterprise customers while making large compute commitments.

Those responsibilities belong together. A promising experiment can justify another run before anyone has decided how much the customer's problem is worth solving. Someone has to connect the research ambition, the machine time and the price.

Google occupies an interesting seat here, participating in the round while Google Cloud supplies research infrastructure. Its partnership with Mirendil spans TPUs and NVIDIA systems, bringing the supplier relationship into the same story as the investment.

As Mirendil opens Lor to more researchers, each lab brings a different judgment about useful data, acceptable evidence and experiments worth another GPU-hour. Those choices will help shape the product the next customer is asked to buy.

Deeper Analysis 👇 https://devcuration.com/articles/mirendil-raises-1b-series-a-ai-research

Q-CTRL

Losing GPS gives a navigation supplier a demanding customer. The aircraft still needs to know where it is, and nobody aboard gets to pause the flight while the software catches up.

Q-CTRL is building for that dependency. Founder and CEO Michael J. Biercuk, CTO Robert Love and Chief Scientist Michael Hush are taking quantum control technology into vehicles that have to operate outside a laboratory. The Sydney company announced a $290M Series C on November 12, led by Salesforce Ventures, with significant participation from Bullhound Capital and Affinity Partners.

Ironstone Opal uses quantum sensing for navigation when satellite signals are unavailable. Q-CTRL reported safety-of-flight qualification under RTCA DO-160 in July. That milestone matters: a system intended to travel with the customer has to survive the customer's environment.

Computing takes another route. IBM offers Q-CTRL's Dynamics Simulator through its quantum platform, letting researchers specify a supported physical model while Fire Opal handles circuit optimization and error suppression. Putting that capability inside an existing workflow changes how a researcher can encounter the product.

One company is therefore selling into two clocks. Aerospace integration has aircraft, qualification work and operational requirements attached to it. Computing adoption has researchers choosing a problem, testing a function and deciding whether the output is useful enough to keep working with. The technology has to earn its place in both.

The investor mix brings industrial relationships into that assignment. EDGE disclosed a $50M investment within the round. MPower Partners is backing expansion in Japan, while Australia's National Reconstruction Fund Corporation is supporting Australian R&D and workforce growth. Those relationships put geography, engineering capacity and potential routes to market inside the financing.

The capital follows a Series B that reached $113M in 2024, including a $59M extension. Q-CTRL now has more resources to turn a sensitive instrument into something an organization can depend on. Its customers will still make that decision through their own engineering teams, procurement processes and operating requirements.

For Biercuk and the team, scaling means staying close enough to those teams to carry the technology into the next aircraft integration, navigation program and research workflow. The person signing off on its use belongs inside the growth story.

Deeper Analysis 👇 https://devcuration.com/articles/q-ctrl-290m-series-c-quantum-software

UFORCE

Moving a drone factory abroad also moves it farther from the people teaching the drone what has changed. UFORCE's expansion puts that relationship inside a $700M financing decision.

The Ukrainian-origin, London-headquartered company announced its Series A on November 12 at a $6B valuation. Valor Equity Partners and the UK National Wealth Fund co-led, with the U.S.–Ukraine Reconstruction Investment Fund joining the round.

CEO Oleg Rogynskyy is building across 2 demands that arrive on different clocks: supply Ukraine now and develop the manufacturing capacity allied countries want closer to home. Co-founder and board chairman Oleksiy Honcharuk brings the Ukrainian institutional and operational experience into that conversation.

The portfolio spans MAGURA maritime platforms, NEMESIS aerial systems, LIUT ground vehicles and PREDATOR turrets, alongside autonomy and command-and-control software. Hardware gets the photograph. Keeping products useful as conditions change requires people who can carry operational feedback back into engineering.

Rogynskyy described that exchange in his Tectonic interview: listen to Ukrainian users while building trust with Western governments and customers. Expanding the factory map means sustaining both conversations, with more production teams and more customer requirements in between.

The National Wealth Fund is putting £113M, which it reports as $150M, into the round. Its plans include hundreds of skilled UK jobs and more than 70% British content in each UK-built MV series boat. Hulls, engines, electronics and fit-out give domestic suppliers a concrete place in the story. The production plan extends into the businesses supplying the boatyard.

Those are expected outcomes. A financing announcement cannot tell a supplier when an order will arrive or tell an operator when a system will be delivered.

Menlo Ventures, Xantium and Hiro Capital also participated, while Shield Capital, Ballistic Ventures, Lakestar and Squadra increased their stakes. Retired General Christopher G. Cavoli, a Valor partner and former NATO Supreme Allied Commander Europe, is joining the board.

UFORCE's July memorandum with ReconCraft already set out a route toward U.S. maritime production. The new capital can help extend that industrial work while Ukrainian manufacturing continues under attack.

For the engineer adapting a design in Britain, the relationship with the Ukrainian user remains part of what the factory has to preserve as it grows.

Deeper Analysis 👇 https://devcuration.com/articles/uforce-raises-700m-to-expand-allied-drone-production

Agile Robots

Hitachi has 2 ways into Agile Robots SE's next factory deployment: through its investment and through the industrial AI it plans to sell with the company. The investor list helps explain the business being built.

Munich-based Agile Robots has closed a $500M Series F led by SoftBank Group, with NVIDIA, Hitachi, Hitachi Ventures, T.Capital and others participating. The November 12 announcement puts its valuation at $8.5B and total invested capital at $2B. The money supports US expansion, more production capacity including Germany, and entry into new markets.

Founder and CEO Zhaopeng Chen and co-founder Peter Meusel came out of the German Aerospace Center. They founded Agile Robots in 2018, putting robotics research to work across industrial automation. The portfolio now reaches from arms and hands to mobile robots, humanoids and the software coordinating them.

A factory buying into that portfolio has several jobs to get done. Moving material, handling a component and adjusting to a changing task ask different things of a machine. The value sits in getting useful work through the production process, with people able to manage what the robots are doing.

Hitachi's September partnership gives that ambition a commercial route. The companies plan to combine their AI robotics technologies and industrial knowledge, with Hitachi offering the resulting systems globally through HMAX Industry. An investor can help fund the engineering while also bringing the application closer to the customer.

Google DeepMind adds another relationship. Its March research partnership pairs Gemini Robotics models with Agile hardware, connecting deployment, data collection, training and iteration. Experience gathered in operation becomes material for the next round of development. That feedback loop has to work across different factories.

NVIDIA is involved beyond the financing, too. Agile ONE SN, announced in October, brings its Isaac GR00T software and Jetson Thor compute into a compact research humanoid scheduled for early 2027. Researchers get a development environment aimed at carrying skills toward industrial use.

Agile says its installed base exceeds 20,000 robotic solutions. Its Fürstenfeldbruck site has stated capacity for up to 10,000 humanoids annually; that figure describes capacity, not deliveries. As Chen takes the business further into the US, the partners around the table will also have work to do where each factory decides how intelligence earns its place in production.

Deeper Analysis 👇 https://devcuration.com/articles/agile-robots-raises-500m-series-f-physical-ai