The Distance After the Breakthrough

Hexagon Bio

Cancer can learn to survive the drug inside a targeted therapy. That leaves an oncologist with a problem a different delivery address may not solve.

Hexagon Bio is working on the cargo. The Menlo Park company closed a $62M Series C on October 23, led by The Column Group, with returning investors Nextech Invest, Deviation Capital, 8VC and CPP Investments. The financing supports HEX-360, a HER2-targeting antibody-drug conjugate, and continued work on additional and dual payloads.

An ADC joins an antibody to a cancer-killing molecule. Hexagon's lead program carries a protein translation inhibitor, intended to interrupt the production of proteins cancer cells need. The company's argument is that a different way of attacking the cell could create another treatment option after resistance to familiar topoisomerase or tubulin payloads. The treatment history comes into the design brief.

CEO and co-founder Maureen Hillenmeyer built Hexagon with Colin Harvey, Yi Tang and Brian Naughton. Harvey remains VP of Synthetic Biology; Naughton is a former CTO. Their starting material sits in microbial genomes, where the company searches for molecules with useful biological effects, then develops them into potential drug payloads. This is a long route from a sequence in a database to a medicine somebody can receive. The molecule has to make that journey with its usefulness intact.

CSO Tara Arvedson, CMO Bill Grossman and SVP of Drug Discovery Victor Cee now sit around that clinical handoff. HEX-360 remains in IND-enabling studies, with a Phase 1 trial anticipated in the first half of 2027. The encouraging activity and tolerability Hexagon presented at AACR came from preclinical experiments. Patients still have to supply the evidence that matters for a treatment decision.

There is an investor choice worth paying attention to here. The Column Group participated in earlier financings and is now leading the round aimed at taking a specific candidate into human testing. The same discovery business needs to keep finding useful chemistry while its lead program takes on the much more particular demands of clinical development.

For a future development partner, the important question reaches past whether Hexagon can find another potent molecule. It reaches into where a new payload might fit after an existing therapy has stopped working, and how a tolerable dose could let that mechanism do its job. Hillenmeyer's team is taking that question toward the first HEX-360 study.

Deeper Analysis 👇 https://devcuration.com/articles/hexagon-bio-raises-62m-series-c-for-adc-payloads

Colare

The recruiter screening a propulsion engineer may never have designed a propulsion system. Colare wants that person to reach the first technical decision with something more useful than a polished résumé.

Co-founders and co-CEOs Esther Thomas and Nain Abdi have launched the San Francisco company publicly alongside a $2.5M pre-seed round backed by Afore Capital, Garage Capital and Acadian Ventures. The money supports a hiring platform built around the work hardware engineers would actually do.

That first decision carries an expensive dependency. The people qualified to assess a mechanical, electrical or embedded engineer are often the same people a company needs building its product. Colare estimates the screening and interview process for 1 hire can consume roughly 40 hours of senior engineering time. That is the company's estimate of the existing workload, rather than a promised saving.

Abdi encountered the problem during engineering work at Ford. At Waterloo, he and Thomas developed the venture through entrepreneurship and co-op programs, turning a familiar frustration into a very specific product question: how much engineering judgment can reach a hiring team before an engineer has to join the call?

Colare maps a role to practical assessments in areas including CAD, PCB design, thermal analysis and finite element analysis. Candidates work through relevant problems; the system assesses their reasoning and output against rubrics informed by experienced hardware engineers. A recruiter gets evidence to review before asking the technical team to spend another afternoon interviewing. The employer approves the assessment and calibrates the hiring bar, keeping the decision with its own team.

The company says its clients range from startups to large manufacturers, with about half in aerospace and defense. Customer counts, revenue and independently measured hiring outcomes remain undisclosed. Those gaps matter for buyers deciding how much responsibility to give an assessment score.

A useful hardware screen has to preserve the tradeoffs inside the work. A design can be easy to manufacture and hard to cool. The hiring team needs enough of the candidate's reasoning to understand the choice, then enough control to decide whether that choice belongs on its own team.

For Thomas and Abdi, expanding the assessment library means carrying that detail into more engineering disciplines. Somewhere in that process, a recruiter will have to decide which candidate earns time with the person responsible for the actual hardware.

Read the full story at DevCuration.

Bedrock Robotics

A contractor can own the excavator, have the next project waiting, and still struggle to put that machine to work. The missing capacity sits in the relationship between the equipment, the crew and the hours available to run it.

Bedrock Robotics is building inside that relationship. Its autonomy system retrofits the machinery contractors already own, starting with excavators. The customer gets a way to extend the usefulness of an existing fleet while experienced people concentrate on work that needs their judgment.

On October 26, Bedrock announced a $600M Series C at a $4.1B post-money valuation, co-led by CapitalG and Emergence Capital. New investors US Innovative Technology Fund, Bessemer Venture Partners and Temasek joined returning backers including 8VC, Georgian, Valor Equity Partners, Xora, Eclipse, Perry Creek Capital and NVIDIA.

CEO Boris Sofman and CTO Kevin Peterson founded the San Francisco company with engineering leaders Tom Eliaz and Ajay Gummalla in 2024. Sofman, Peterson and Gummalla bring Waymo experience; Eliaz brings the engineering background of Segment. That combination now has to work on ground that changes every time the bucket moves.

Bedrock says its fully autonomous excavators began paid work on active commercial sites in August. It also reports more than doubling its customer volume since 2025. Those are company-reported milestones, with revenue and the size of the operating fleet still undisclosed in the announcement.

The adoption choice is unusually practical for an AI company. A contractor has already paid for the iron. Bedrock can enter through that installed equipment, then earn its place in how the crew plans and performs the job. Emergence describes working with the team on pricing and packaging around utilization, throughput, safety and schedule certainty.

The Series C takes cumulative funding to nearly $1B, following the $270M Series B and the earlier $80M Seed and Series A package. The stated use now includes the field, commercial and operating infrastructure behind wider deployments. Amy Jones Satrom joins as VP of Operations and Lando Juarez as General Counsel as that footprint expands.

For the superintendent bringing autonomy onto the next site, the relationship continues after installation: machines working alongside people, support available when needed, and another day's earthwork fitting into the contractor's plan. The next deployment brings a different crew and a different patch of ground.

Deeper Analysis 👇 https://devcuration.com/articles/bedrock-robotics-600m-series-c-autonomous-construction

Emvolon

Landfills and commodity trading desks rarely share an engineering brief. Emvolon is giving them a reason to: turn methane at the waste site into a liquid a fuel buyer can actually move.

That distance matters. Gas scattered across landfills, dairies and wastewater facilities can be expensive to gather into a conventional chemical plant. Emmanuel Kasseris, co-founder and CEO, and Leslie Bromberg, co-founder and CSO, built Emvolon around bringing smaller plants to the gas.

The Woburn, Massachusetts MIT spinout announced a $22M Series A on October 27, with MassVentures, VX Ventures, Wagner Carbon and others participating. Federal and state grants accompany the financing announcement; the public materials do not break out the equity and grant components. The funding supports its first commercial-scale biomethanol system and expansion of the team.

The machinery has an unexpected résumé. Emvolon repurposes mass-produced engines as chemical reactors, running methane fuel-rich to make intermediates for methanol synthesis. Standardized units can sit alongside the feedstock and add capacity in parallel. Familiar manufacturing gives the founders a way to pursue small chemical plants without designing every installation from scratch.

Getting methanol out of the machine also starts another business: gathering product, storing it and reaching customers. The liquid can travel in standard isotankers. Emvolon's August agreements with Freepoint Commodities put that work beside the reactor. Freepoint has a first purchase option for biomethanol and other liquid fuels under a 7-year framework, plus project-level equity and debt participation rights.

The purchase option carries a potential $450M-plus value. It is an avenue to future sales, with volumes dependent on deployment, rather than revenue already earned. The separate Montauk Renewables collaboration provides a pilot-to-commercial path and a plan for up to 50,000 metric tons a year across multiple sites.

Eric Bensaude is also joining the board, bringing experience building Cheniere's global LNG trading platform. Commodity commercialization belongs in this company's engineering conversation because a distributed plant needs an organized route to market.

Emvolon targets commercial methanol production in 2027. As those units reach waste sites, Kasseris and Bromberg will be connecting the pace of engine manufacturing to the cadence of fuel collection, with Freepoint waiting downstream for the liquid.

Deeper Analysis 👇 https://devcuration.com/articles/emvolon-22m-series-a-biomethanol

Re:Build Manufacturing

By the time a manufacturer receives a finished design, several expensive decisions may already belong to somebody else. Re:Build Manufacturing wants its engineers in the conversation while those decisions can still move.

CEO and co-founder Miles Arnone has built the company around that earlier seat at the table. Jeff Wilke and William Barker, his fellow co-founders and co-executive chairmen, are backing a business that stays with the product from development into production. A beautiful drawing still has to become something a factory can make repeatedly. Its industrial customers span aerospace and defense, electrification, medical, energy, robotics and automation.

Re:Build announced $2.1B in new funding on October 27, led by General Catalyst and JPMorganChase's Strategic Investment Group. The money will expand U.S. manufacturing capacity and support new industrial companies. General Catalyst already invested $120M in 2024, giving this financing a returning partner with some history inside the project.

The customer proposition starts before the purchase order reaches the factory floor. Re:Build's Manufacturing Constrained Design process brings cost targets, available processes and production capacity into product development. A manufacturing engineer can influence the design while the product team still has choices, rather than inherit an argument everyone else has already finished.

There is a fairly physical version of that strategy in Rock Hill, South Carolina. Earlier this month, Re:Build announced an approximately $20M plan for a 130,000-square-foot facility with a clean room, cold storage and added machining and finishing capabilities. The company projects 150 new jobs by 2028. Those plans put equipment and people behind the language of industrial expansion.

Arnone's letter about the financing makes the time commitment explicit: manufacturing develops through long feedback cycles, and capital has to accommodate the work. That matters to an industrial founder deciding how much of a product's future to entrust to a supplier. Production support needs to survive the revisions, not just the first enthusiastic meeting.

Re:Build is also incubating companies, so its engineers and facilities serve ambitions on both sides of the customer relationship. As the platform grows, Arnone's early seat at the design table comes with a longer assignment: carry those decisions into the production runs customers will eventually depend on.

Deeper Analysis 👇 https://devcuration.com/articles/rebuild-manufacturing-raises-2-1b-us-manufacturing

Loamy

Selling strawberries to a juicer can be a good decision made too late. Loamy wants the sales team to have that conversation while the fruit still has somewhere useful to go.

The Los Angeles company announced $6M in Seed funding on October 27, led by Zeal Capital Partners, with Slauson & Co., Supply Change Capital and Amplify.LA participating. Loamy reports $10M in total funding, backing a business built around the expensive interval between spotting a quality problem and doing something commercially useful about it.

Charles Julius, co-founder and CEO, and Jason Pesek, co-founder and COO, came through operational software at Motive. Charles Julius also built Procore's app marketplace. They founded Loamy in 2024, bringing that experience into a supply chain where the inventory keeps changing even when nobody touches it.

A buyer's requirements follow the produce through the journey: size, condition, paperwork and the day it needs to arrive. Lose track of one of those requirements and the sales team can discover the mismatch after the shipment has already committed everyone to an expensive answer.

Loamy connects food safety, quality, inventory, orders and traceability. The company describes satellite and infrared analysis that helps flag field problems earlier, alongside photo-based condition checks and software that reads supplier documents. Field crews can use phones or tablets, including offline, according to its announcement.

The commercial attraction is the handoff. A quality warning becomes useful to the person choosing a customer, a destination or a processing route. Giving that person more time can preserve an option that disappears once a load reaches the wrong dock.

Loamy says customers have cut product rejections by 50%. The release also reports more than 30x ARR growth over the past year, but supplies no absolute revenue figure or customer cohort. Those are company-reported results; the rejection claim gives a buyer a concrete outcome to examine against their own loads.

The financing will support engineering, go-to-market teams, the AI roadmap and additional food categories. Expansion means carrying that relationship between quality and sales into more operating environments, with different products and buyer requirements.

For Charles Julius and Jason Pesek, the next customer conversation can start with a very physical question: which fruit still has a buyer, and who learns that soon enough to make the call?

Deeper Analysis 👇 https://devcuration.com/articles/loamy-6m-seed-food-supply-chain

EDGE Sound Research

The basketball has its own audio feed. Getting it there means teaching software which part of a roaring arena belongs to the ball.

That assignment has carried EDGE Sound Research from live entertainment into a much wider question about how machines understand their surroundings. The Riverside company announced an additional $2M led by Stormbreaker Ventures on October 27, bringing its reported total funding to $9M.

Co-founder and CEO Valtteri Salomaki and co-founder and CTO Ethan Castro, PhD came at sound from different directions. Ethan Castro studied hearing and physical vibration at UC Riverside. Valtteri Salomaki brought an information-systems perspective to the signals moving through a space. Their shared problem now sits inside a broadcast engineer's working day.

A microphone records what reaches it. A production team needs the sound that belongs with the action on screen, while the crowd, the coach and the rest of the building continue contributing their opinions. More capture creates more material for somebody to sort.

EDGE's Virtual Sound Engine connects audio with tracking and other sensor information to help separate those overlapping sources. Shure has described how its DCA901 microphone array works with the software to identify useful sound objects and put them within reach of the engineer making the mix.

That relationship deserves attention. Shure became a minority investor earlier this year, and its capture equipment feeds the software doing the interpretation. A supplier already involved in collecting the signal has a direct interest in making that signal more useful downstream.

For the broadcaster, the purchase has a practical shape: find the relevant sound and keep it attached to the moment. The wider physical-AI ambition grows from that same need for context. Activity can matter even when a camera has no view of it, and a machine needs some way to connect the sound with the rest of the scene.

The new funding supports further sports deployments, more multimodal sensor data and development toward robotics, autonomous systems and intelligent environments. Those broader uses remain development work. The commercial sports business gives the founders a place to keep building while customers have an immediate reason to want better audio.

Stormbreaker is funding the stretch between those two markets. Back inside the venue, the engineer still needs the right sound as the play moves across the floor, with all the other channels open.

Deeper Analysis 👇 https://devcuration.com/articles/edge-sound-research-raises-2m-in-funding

Nord Quantique

DARPA is paying to read the parts of a quantum roadmap that usually stay between the engineers. For Nord Quantique, that puts the risks of building a useful machine inside the funding relationship.

The Sherbrooke company announced a US$10M extension of its Quantum Benchmarking Initiative Stage B agreement on October 27. The money supports continued development and evaluation of its hardware-efficient approach to fault-tolerant quantum computing.

CEO and co-founder Julien Camirand Lemyre, co-founder and CBO Philippe St-Jean, and CTO Marc-Antoine Lemonde are pursuing a design choice with consequences well beyond the laboratory: put error correction directly into the qubit architecture, and reduce the extra hardware required to keep a calculation useful.

That overhead eventually needs somewhere to live. It has to be built, controlled and supported, so a decision about correcting errors becomes a decision about the infrastructure a future customer may need to operate.

Nord Quantique uses superconducting circuits and bosonic codes. The company describes a 1:1 logical-to-physical qubit ratio and a path toward compact systems that can integrate into existing data centres. Those are architecture claims and development goals, with the full commercial machine still ahead.

The useful detail in DARPA's program is what Stage B asks for: an R&D plan, the risks in that plan, the steps to reduce them, and the prototypes needed to make those steps credible. Somebody outside the company gets to examine the assumptions before the promised computer arrives. That gives public R&D support a second job: helping an evaluator understand where the design could fail while the engineers still have room to change it.

Nord Quantique has been working at that level of detail. In July, it reported combined state-preparation and measurement errors below 0.1% in its grid-state qubit research. Getting a computation started correctly and reading its answer accurately are part of the engineering assignment too.

The extension continues Stage B. DARPA lists Nord Quantique there in its October 8 roster; Stage C carries a separate verification and validation assignment. The program's utility-scale ambition is a computer whose computational value exceeds its cost by 2033, while Nord Quantique's own roadmap targets 256 logical qubits by 2030.

For the team in Sherbrooke, the compact computer now has a particularly demanding reader following the route from the qubit to the machine around it.

Deeper Analysis 👇 https://devcuration.com/articles/nord-quantique-receives-us10m-darpa-stage-b-extension

Manicule

Naman Bansal says he shut down a documentation business doing $50K in monthly revenue. Customers were asking what their invoices actually bought, and faster pages were becoming an awkward answer.

At Manicule, that question widened the assignment. Naman Bansal, co-founder and CEO, and Shreyans Jain, co-founder and CTO, are building Agent Relations: helping software companies get discovered, recommended and used by AI agents. The San Francisco company announced $1.3M in Seed financing on SAFEs on October 27, co-led by Y Combinator and Broom Ventures, with strategic technology angels participating.

The customer is a developer-tool company whose product has to survive somebody else's explanation. An AI assistant can describe the wrong use case, find an outdated example or produce code that fails against the live API. By then, the vendor's carefully written positioning has another author.

Manicule puts more of that journey inside its scope of work. Its team handles strategy, technical documentation, content, original benchmarks and distribution. The company's website describes checking snippets against live APIs and running coding agents through tasks. Accuracy becomes part of the route into the product, with engineers responsible for what the content enables.

Jain previously worked as a founding engineer at Supermemory, which Manicule also names among its customers. That connection gives the work a specific setting: technical information has to help a developer or an agent accomplish something inside software. A polished explanation can still leave the integration unfinished.

There are 2 ways to buy the current offering. Companies can hire Manicule's team or use its visibility software to track how ChatGPT, Claude and Gemini describe their brands. Keeping both options matters. A customer may have the people to act on the findings; another needs somebody to take responsibility for the work afterward.

The financing is earmarked for core engineering, internal automation and technical account capacity. That last item belongs in the story: expanding the promise to customers also expands the conversations, judgment and follow-through the business has to deliver.

Naman Bansal's account of the change makes the commercial pressure clear. Producing more content with AI gives a services company more output to sell, while the buyer can keep asking which part helped somebody choose the product. Manicule has brought that question into the engagement, where the next invoice will have to live with the answer.

Deeper Analysis 👇 https://devcuration.com/articles/manicule-1-3m-seed-agent-relations

Blue Arbor Technologies

Ottobock is financing work that starts on the surgical side of the prosthetic hand. That puts an established manufacturer of advanced limbs close to the question its customer eventually has to answer: how do I tell this thing to move?

Blue Arbor Technologies, Inc. closed a $10.5M Series A on October 28, led by Ottobock, to advance its RESTORE neuromuscular interface. The Grass Lake, Michigan company is developing the connection between a person's peripheral nervous system and a robotic prosthesis, where the instruction to move has to become something the machine can use.

The approach brings reconstructive surgery into the control system. Electrodes record signals from muscles connected to the patient's nerves; a sensing unit processes those signals and sends movement commands to the prosthesis. RESTORE is designed to support independent and simultaneous finger, wrist and elbow control. The ambition is movement that follows the wearer's intention more naturally.

President and co-founder Paul S. Cederna, CMO and co-founder Theodore Kung, and CTO and co-founder Alex Vaskov bring the clinical and engineering work together. CEO Tod Borton leads the company through the medical-device development work surrounding that interface. A better signal still needs a system a surgical team can implant and a person can use.

The current External Sensing Unit sits outside the body, connected through the skin to implanted electrodes. Blue Arbor plans to finish feasibility work and pursue an FDA Investigational Device Exemption for U.S. clinical studies, while beginning development of a fully implantable Internal Sensing Unit. Two versions, at different stages, with different work ahead.

Blue Arbor reports 2 human implantations so far, including a second procedure at the Medical University of Vienna. Those early experiences belong to the development record. RESTORE remains investigational, without commercial approval or clearance; its FDA Breakthrough designation and TAP participation support the development path.

Ottobock began this Series A with a $5M investment announced in February. The $10.5M is the round's announced close, so adding the 2 figures would count part of the financing twice.

For Ottobock, the investment reaches beyond the machine it supplies and into the biological signals that let somebody operate it. The clinical work now being financed will help determine how that relationship feels in a wearer's hand.

Deeper Analysis 👇 https://devcuration.com/articles/blue-arbor-raises-10-5m-series-a-prosthetic-control

Fragment

Graybar's finance team helped teach Fragment how its business works. That contribution deserves a place in the financing story.

The electrical distributor appears among the customers named in Fragment's launch announcement, alongside CNX. Graybar describes bringing its operating knowledge into the work with the startup's AI team. That is a demanding kind of software relationship: the customer has to explain the judgment that keeps purchasing moving, then trust the product to carry it forward.

Co-founders CEO Pranav Mulgund and Naveen Kashyap are building for direct procurement at large physical-goods businesses. A changed order or invoice discrepancy can send somebody through purchasing records, contracts, receipts and supplier conversations. The answer depends on what those records mean together, including the rules an experienced employee already knows.

Fragment's Autonomous Context Engine connects that company-specific information so agents can investigate exceptions and execute authorized resolutions across existing systems. The company says it works with the infrastructure already in place. For a manufacturer, retaining that investment makes the adoption conversation considerably easier to start.

The San Francisco company launched October 28 with $56M in total funding, including a Series A led by 8VC and a Seed round co-led by General Catalyst and A*. The amounts allocated to each round were not disclosed. Named angels include Roelof Botha, Nikesh Arora, Jeff Wilke, Eric Wu and Dennis Cinelli.

The money will support further ACE development, broader source-to-pay coverage and more enterprise deployments. Fragment describes a business model built around work its agents complete, putting commercial weight behind the word resolved. Its public announcement leaves the pricing unit unspecified.

That makes the customer's expertise valuable twice. Procurement people use it to handle the exceptions arriving today; Fragment needs to make it usable by software handling the next ones. Supplier relationships and company policies become part of the product's working material, with permission to act still belonging to the customer.

That knowledge can include who owns an approval and which supplier needs a follow-up.

There is a substantial implementation job inside that promise. As Fragment expands, more buyers will be showing its agents how a purchasing problem gets from a discrepancy on a screen to a resolution their colleagues and suppliers can actually use.

Deeper Analysis 👇 https://devcuration.com/articles/fragment-raises-56m-total-procurement-ai

Park Road Partners

A first fund asks limited partners to do something unusually personal with institutional capital.

They have to decide whether an investor's history can survive outside the institution where it was built.

Park Road Partners closed Park Road I at $695M after beginning with a $375M target and later raising that target to $450M. The final close landed approximately 85% above the original plan and 54% above the revision.

That is a remarkable fundraising result for a firm established in 2025. It is also a very specific kind of result. LPs have validated their appetite for the strategy before Park Road has had time to produce a realized fund record of its own.

The fundraising math captures the speed of that belief. Park Road moved from its initial Form D to a final close in roughly five and a half months while the amount investors were willing to commit kept moving higher.

Kyle Peterson founded Park Road after more than 14 years at New Mountain Capital, where the work included healthcare, software, business services and financial services. Park Road carries part of that sector discipline into a new platform focused on technology-enabled services and software across Healthcare & Life Sciences and Financial & Insurance Services.

Matthew Dubbioso serves as Partner. Dino Verardo is Partner, CFO and CCO. William Blair appears in Park Road I's Form D as placement agent.

The regulatory timeline shows how quickly the fundraising moved. Park Road I filed its initial Form D on May 15. At that point, the filing listed an indefinite offering and $0 sold. By October 29, Private Equity International reported the $695M final close.

The capital gives Park Road far more capacity than its original plan anticipated. It also raises the operating obligation attached to the close.

Healthcare and financial services reward specialization because the software is rarely allowed to remain only software. It has to move through regulation, reimbursement, risk, legacy infrastructure, institutional buyers and the human relationships that make those systems function.

Park Road has not publicly identified its LPs, disclosed its check-size plan or shown how the vehicle will be constructed across those markets. The close tells us investors were willing to back the team's ability to make those decisions. The portfolio will begin showing us what that conviction looks like once it has companies, boards and operating consequences attached to it.

Deeper Analysis 👇 https://devcuration.com/articles/park-road-partners-closes-695m-debut-fund-above-target