The People Who Were Already There

ImpriMed

Two patients can share a diagnosis and hand the same drug completely different futures.

ImpriMed, Inc. built its company around putting the individual patient back into that decision. It has raised a $10M Series A2 bridge to move its live-cell testing and AI platform further into human cancer care.

LB Investment led the round. New investors NAVER D2SF, Samsung Securities, and Alois Ventures joined returning investors BonAngels and Han River Partners. The financing follows ImpriMed's $23M Series A in 2023.

Co-founders Sungwon Lim, CEO, and Jamin Koo, CTO, started ImpriMed in 2017 with a practical question: what if clinicians could test how a patient's living cancer cells respond to available drugs before committing to a treatment?

xCellSense keeps those cells viable outside the body, measures drug response, and combines the results with genomic, immunophenotypic, clinical, and outcome data. AI has something more demanding than a slide deck here: biology that refused to behave like an average.

Veterinary oncology became the first commercial proving ground. ImpriMed reports more than 27,000 canine and feline lymphoma tests. Peer-reviewed retrospective work in relapsed canine B-cell lymphoma associated prediction-matched treatment with better response and survival.

That evidence deserves precision of its own. It is veterinary and retrospective, not proof that the same outcomes will transfer to human patients. ImpriMed's U.S. human predictions remain for research and investigational use only.

The National Cancer Institute is advancing functional precision medicine because living cells can reveal drug response that a mutation report may miss. NCI also names the unfinished work: rigorous evidence, standardized assays, viable tissue workflows, regulatory acceptance, and routine clinical integration.

This bridge is financing that handoff. ImpriMed is targeting FDA clearance and CLIA certification by Q1 2027, developing human blood-cancer and blood-infection products, expanding its biopharma CRO work, and pushing commercialization across the U.S. and Asia. No valuation was disclosed, and the regulatory timeline remains a target.

Veterinary oncology gave ImpriMed years of cases and operating repetition. Human oncology will demand a different standard of proof. The next chapter belongs to whether the company can carry that evidence across the regulatory and clinical handoff without losing what made the platform useful.

Deeper Analysis 👇 https://devcuration.com/articles/imprimed-raises-10m-series-a2-precision-oncology

Heidi

Between a healthcare AI demo and routine use sits an institutional handoff: procurement, integration, training, governance, budgets, and the clinician deciding whether the tool survives Monday morning.

Heidi just financed both sides of that handoff.

The Melbourne company announced $340M in new funding: a $100M Series C led by Blackbird, with Phoenix Court, Point72 Private Investments, and Headline participating, plus a separate $240M growth investment led by General Catalyst’s Customer Value Fund. The equity round values Heidi at $900M.

The Series C is $100M. The larger $240M component is growth capital tied to go-to-market capacity, not more Series C equity. General Catalyst’s published Customer Value model finances customer acquisition against value produced by funded cohorts. Heidi’s complete contract is private, so its economics remain outside the public record.

That split matters because healthcare AI has two different scaling problems.

The first is technical and clinical. Co-founders Dr. Thomas Kelly, Waleed Mussa, and Yu Liu built Heidi around ambient documentation. The platform now includes Scribe, Evidence, Remote, Dictate, workflow tools, and EHR integrations. Thomas Kelly leads as CEO and Yu Liu as CTO. Heidi says the next phase will move into supervised clinical actions while medical judgment stays with clinicians.

The second problem is institutional. A health system has to fit the product into local workflows, privacy rules, procurement, training, clinical governance, and deployment economics. General Catalyst’s capital gives Heidi a dedicated pool for that commercial work while the equity can support product, engineering, quality, and expansion.

Heidi reports 2.8M patient visits supported each week across 190 countries and 110 languages, more than 175M visits in total, and over 67M clinical hours. It says ARR grew from $1M to $50M in two years, reaching that level in April 2026. Those are company-reported measures, not an independent audit, but they explain why existing investors are underwriting another jump in scale.

Named deployments include NHS Midlands, Metro South Health, Health New Zealand, and Beth Israel Lahey Health. Heidi also opened Toronto as its North American headquarters. Every new system adds local configuration, implementation work, regulatory boundaries, and evidence the product must keep earning.

Heidi now has capital for the model and capital for the handoff. The next record will be written inside the health systems that turn both into durable clinical use.

Deeper Analysis 👇 https://devcuration.com/articles/heidi-raises-340m-series-c-growth-capital

Mantic

A forecast only becomes valuable when someone is willing to act before certainty arrives.

That is the difficult handoff Mantic is selling. A probability has to influence a trade, product launch, acquisition plan, supply decision, or policy choice while the outcome can still make the person using it look brilliant, reckless, or both.

Toby Shevlane and Ben Day founded Mantic in London in 2024 to build AI for judgmental forecasting: questions that demand fresh research, contextual reasoning, and calibrated uncertainty instead of a neat historical dataset begging for a trend line.

This summer, Mantic's system finished ahead of all 676 human entrants in the Metaculus Cup, including professional forecasters. One bot, laertes, finished ahead of Mantic.

That last detail belongs in the story. The result applies to a defined tournament, its questions, scoring rules, timing, and field. It shows that an AI can commit probabilities before answers are known, disagree with a strong human crowd, and earn a score worth examining.

Radical Ventures has now led a $25M seed round, with Balderton Capital, Thinking Machines Lab, DRW, FT Ventures, M12, Episode 1, Charlie Songhurst, and Thomas Wolf participating. Mantic previously raised a $4M pre-seed. The valuation was not disclosed.

The company says hedge funds use Mantic across geopolitics, macroeconomics, and business. Fortune 500 customers are using forecasts around M&A and product launches. Reuters reports that companies and government agencies have integrated the technology, although Mantic has not named customers or published measured decision outcomes.

That evidence gap is where the business gets interesting. Tournament accuracy is clean enough to rank. Enterprise decisions are not. A portfolio manager can ignore a model. An executive can use the right probability and still make the wrong choice. Better forecasts still have to earn trust without destroying the economics.

Mantic will use the new capital to hire across research, engineering, product, sales, and operations while expanding the compute and data behind its forecasts. The technical job is to keep improving calibration. The commercial job is to make disagreement useful inside institutions that already have analysts, incentives, committees, and reputations attached to their own view of the future.

Winning a tournament earned Mantic attention. Building a durable company means carrying that measured advantage into live decisions, then staying accountable after reality finally shows up.

Deeper Analysis 👇 https://devcuration.com/articles/mantic-raises-25m-ai-forecasting-metaculus

Ande

The cleanest client dinner is the one nobody inside the company has to discuss after it ends.

The room worked. The right people showed up. The contract cleared, the payment landed, and the customer remembers the conversation instead of the logistics.

Ande has raised more than $52M across seed and Series A financing to turn that workflow into a network. Lightspeed Venture Partners, Redpoint Ventures, Duration Ventures, and Sierra Ventures invested, with Bain Capital Ventures participating. The split between rounds and the valuation were not disclosed.

CEO and co-founder Lohit Sarma spent roughly 2.5 years building supply before taking Ande out of stealth. The company says its network now reaches 93,000 entertainment providers across 90+ cities, including 1,600 hospitality venues and partners.

That sequence matters. A general-purpose agent can suggest a restaurant. It cannot reliably book a private room, confirm live availability, negotiate group terms, route legal approval, execute the contract, enforce policy, pay the venue, and reconcile the expense if the underlying supply data and enterprise permissions do not exist.

Ande is building both sides. Employees get one place to discover and book dinners, offsites, sporting events, catering, and other experiences. Finance and legal get visibility and control. Venues get a direct channel into enterprise demand that historically arrived through relationships, inboxes, and whoever knew whom.

The company reports more than 60 enterprise customers, including Cloudflare, Salesforce, McGraw Hill, Netskope, and Navan. It says more than $400M in annual entertainment spend flows through the network and customers report savings of 12% to 15%. Those are company-reported metrics, not Ande revenue or audited results.

The investor bet is bigger than reservations. Corporate entertainment lives where procurement, payments, hospitality, customer relationships, and executive visibility collide. Everyone wants the experience to feel personal. Nobody wants the controls behind it improvised.

Ande still has to prove the economics, retention, and service quality behind the scale it reports. The company has kept negotiation human-led, which is a useful admission about the category: the software can carry the process, but one bad event can damage a relationship no workflow can refund.

The capital gives Ande room to make the machinery disappear. The opportunity is making sure the relationship is the only part people remember.

Deeper Analysis 👇 https://devcuration.com/articles/ande-raises-52m-corporate-entertainment-ai

Brahma AI

Hollywood taught enterprise AI a lesson the chatbot cycle skipped: content has to survive production.

A generated image can look convincing and still fail the company using it. The source may be unclear. The performer may not have consented. The brand team may have no way to prove which model touched the asset, who approved the result, or where the rights travel next.

That operating mess is the market BRAHMA AI is building for.

The company raised $150M through preferred shares at a reported $2B post-money valuation. Multiples Alternate Asset Management invested $100M and led the round. Brahma AI also disclosed another $100M of investor interest, which is interest, not closed financing.

Founder and CEO Prabhu Narasimhan assembled Brahma AI from technology and teams connected to DNEG, Prime Focus Technologies, and Metaphysic. Co-founder and CTO Jo Plaete leads technology and R&D. Together, they are trying to join the parts of enterprise audiovisual work that vendors usually sell separately.

Brahma AI Core makes archives searchable and turns content into usable intelligence. Brahma AI Studio handles creation, including ATMAN digital humans and VAANI multilingual voice localization. Around them sits the work enterprise buyers have to defend: security, identity, consent, provenance, governance, localization, and distribution.

The company names Warner Bros., the NBA, and Mayo Clinic as anchor customers, with Google, Hakuhodo, and DNEG among its strategic relationships. Those are company-reported relationships, not a substitute for audited adoption, but they show where the platform is aiming: institutions with enormous content libraries, valuable identities, and very little appetite for losing control of either.

Multiples is financing the shift from a creation tool to an operating layer. The proceeds are expected to support R&D, global go-to-market, talent and process development, sales capacity, and a significant Silicon Valley presence. Brahma AI is also preparing interactive digital humans and a model-agnostic platform for new enterprise use cases.

The hard part begins after the demo. Brahma AI has to make inherited technologies behave like one dependable system, give legal and security teams evidence they can trust, and move across industries without becoming a custom-services maze.

Enterprise audiovisual AI will be judged by what it can create. It will be bought, renewed, and governed by what the organization can still control after creation.

Deeper Analysis 👇 https://devcuration.com/articles/brahma-ai-raises-150m-for-enterprise-content-ai

Rising Tide

Property-management firms carry a second ledger that never reaches the balance sheet. It holds the owner who calls one person when a water heater breaks, the vendor who trusts the invoice will be paid, and the employee who stayed through the lean years.

That ledger is hard to buy and painfully easy to damage.

Rising Tide has closed an undisclosed Seed round led by Left Lane Capital to build a federation of independent residential property-management companies. Left Lane incubated the Chicago company alongside founder and CEO Blake Mohseni around a thesis that acquisition math often misses: the local name, team, and relationships are part of the asset.

Rising Tide wants to centralize the work that benefits from scale while keeping local judgment where it was earned.

The interesting part is how Rising Tide is trying to earn the right to scale. Its Lab embeds with an operator for 60 to 90 days and builds automation inside the workflows already consuming the week: tenant communications, leasing, unit turns, collections, maintenance, and owner reporting. The company says the program is free and currently live with three firms. Operators keep using what is built, while their specific business data stays protected by the engagement terms.

Rising Tide can see where software removes coordination work and where a person still needs to exercise judgment. In property management, confusing those categories is how an efficiency plan becomes a service problem.

Left Lane's investment logic fits the model. Managing Partner Vinny Pujji says the firm has invested more than $100M across AI companies modernizing traditional services. Property management gives that thesis a difficult proving ground because the work touches owners, residents, vendors, employees, local rules, physical assets, and emergencies that ignore the product roadmap.

The financing amount, valuation, other investors, and names of federation members were not disclosed. Neither were audited productivity gains. Rising Tide is early, and the company says so plainly. The current design is specific: shared systems behind the scenes, local accountability at the door.

Blake Mohseni now has to show that the second ledger can compound instead of disappearing into the integration plan. The answer will arrive inside ordinary moments: a maintenance handoff that moves faster, an owner report that no longer eats the week, and a local team that still recognizes the business after the new infrastructure shows up.

Deeper Analysis 👇 https://devcuration.com/articles/rising-tide-closes-seed-round-for-property-management-ai

Topdog

Real-time multiplayer has an unforgiving empty-chair problem. The second player has to arrive before the first one loses interest.

That pressure sits underneath Topdog's $2.5M Seed. Co-founders Aman Agarwal and Jatin Narang built real-money skill games where people compete live and adjust to one another instead of submitting separate scores.

The difference sounds small until the product has to work at 11:47 on a Tuesday morning. A live game needs the right opponent, at a compatible skill level, in the same entry range, ready at the same moment. Matchmaking, player liquidity, payments, fraud detection, and live operations become part of the game design.

Boston Seed Capital led, with The Raine Group, Bullpen Capital, Versus Ventures, Spoondrift Capital, and Permit Ventures participating. An earlier $1.6M Lightspeed pre-seed brings disclosed funding to $4.1M.

The investor mix carries useful memory. Boston Seed Capital Managing Director Peter Blacklow previously led WorldWinner's skill-games business, and the syndicate spans sports, gaming, consumer products, and marketplaces. This group understands why an empty lobby can kill a clever game.

Topdog offers 21 Hustle, Basketball Hustle, Pinfall, Tilt, and 21 Royale. The games are synchronous and built around decisions that change in response to an opponent, making competition feel less like posting a score and more like reading the person across the table.

The company reports real-money availability in 45 states, more than $10M paid to players, and a 4.6 App Store rating across 1,097 ratings. Those are company and platform-reported signals, not audited proof of retention or unit economics; revenue, active users, and acquisition costs remain undisclosed.

That missing data is where this round becomes interesting. Topdog says it will spend on user acquisition and expand its sports, card, and strategy catalog. More titles give its live-player thesis more chances to prove itself. In a synchronous marketplace, acquisition is also inventory: each active player can make the product more useful for the next one.

More games create more reasons to return, but they also divide the player pool across formats, skill levels, entry amounts, and time windows. Topdog has to grow the catalog and the matching density together while keeping contests fair, payments reliable, and access compliant with state-by-state rules.

The $2.5M gives Topdog more chances to make the second seat appear quickly enough that nobody notices how much machinery it took to fill it.

Deeper Analysis 👇 https://devcuration.com/articles/topdog-raises-2-5m-real-time-skill-gaming

EnergyCAP

Utility spend crosses finance, facilities, sustainability, and operations, which is a polite way of saying every team can touch the data while nobody sees the whole decision.

EnergyCAP has spent more than four decades fixing that handoff. The company now has a strategic investment from LLR Partners, with existing investor Resurgens Technology Partners remaining alongside it. The amount, valuation, ownership stake, and transaction structure were not disclosed.

That missing number does not make the operating signal smaller. More than 750 organizations use EnergyCAP to centralize utility, energy, and emissions data. The investment announcement says those customers track more than $50B in utility bill value each year across government, education, healthcare, and commercial portfolios.

Those bills carry rate changes, meter exceptions, usage patterns, emissions inputs, budget pressure, payment workflows, and surprises that grow expensive inside a spreadsheet.

CEO Shawn Lankton is building the next phase around faster answers and fewer manual handoffs. CTO Chris Carney leads the engineering and technology strategy behind that work, including AI-driven Bill Capture. EnergyCAP has also added Watts AI, Watts Chat, and Insights, while its 2022 acquisition of Wattics expanded the platform into real-time energy analytics.

The customer evidence is company-reported, but specific. EnergyCAP's 2026 survey gathered about 130 responses across more than 100 organizations. 94% said the platform meets or exceeds ROI expectations, 98% reported more confidence in their utility data, and customers reported receiving about 3x the value they pay.

LLR is reading that through its Industrial Technology thesis: physical spaces are getting harder to manage while the systems around them remain fragmented. LLR Vice President Alexandra Van Arkel and Partner Jack Slye are joining a company Resurgens first backed in 2021. Resurgens Managing Director Adi Filipovic is staying invested after five years of product expansion.

Founder Steve Heinz began this work when moving energy information off paper was the breakthrough. The next chapter asks whether EnergyCAP can turn utility data into a live operating system for the people paying the bill, running the building, defending the budget, and explaining the emissions.

The next proof lives in the meter accounts, rate tables, payment queues, and cross-department decisions where customers can catch one expensive mistake before it hardens into the monthly close.

Deeper Analysis 👇 https://devcuration.com/articles/energycap-secures-llr-investment-utility-data-growth

Falcon Gases / Encore

Industrial gas distribution punishes anyone who mistakes physical inventory for the whole business. The cylinder matters. So do route density, fill capacity, technical support, and the local operator customers call when supply cannot wait.

Falcon Gases is putting capital behind that operating truth. Odyssey Investment Partners has taken a strategic majority stake in the packaged-gas platform as Falcon acquired Encore Gas & Supply as its first operating business. Both values were undisclosed.

That combination matters more than a sponsor logo. Falcon now has a Southern California distributor with three showrooms, a specialty and industrial gas facility, and customers buying packaged gases, bulk and microbulk supply, welding equipment, and technical services. Encore turns an acquisition thesis into an operating company with trucks, inventory, employees, and promises that have to be kept.

Falcon formed in late 2025 with a pitch to independent distributors: take liquidity and platform equity, keep local relationships, and add operating and acquisition support. Tailwind backed the formation. The new announcement makes Odyssey the majority investor but does not explain Tailwind's current interest.

The people make the model legible. Falcon CEO Pete Jeffe has scaled industrial businesses. CCO Chris Granger and VP Projects Steve Morton bring decades inside industrial gases, while CFO Eric Friedman adds finance and specialty-distribution experience. Encore President Ben Anderson and owner Kelly Park built Encore from a 2014 startup into a regional supplier. Odyssey reports more than 50 platform investments and approximately 300 add-ons.

Independent gas distributors can have deep local reputations and succession questions at the same time. Private equity can offer liquidity, acquisition capacity, systems, and capital. The difficult part is adding those resources without sanding away the response time, product knowledge, and continuity that made the distributor valuable.

Encore is where Falcon's partnership claim becomes measurable. Customers will experience the strategy through availability, delivery, technical help, and whether familiar people still solve problems. Employees will experience it through investment, opportunity, and the daily decisions that determine whether a local business still feels local.

Falcon's next acquisition may add geography. Encore will show whether the platform can make a local distributor stronger while customers still recognize the business answering the phone.

Deeper Analysis 👇 https://devcuration.com/articles/falcon-gases-odyssey-investment-encore-acquisition

Ekai

An AI agent can know the public definition of revenue and still get the company's number wrong. The expensive part is usually hiding in the exceptions: which returns count, when a contract is recognized, and whose definition finance will defend at quarter close.

That is the problem under Ekai's $1.7M pre-seed round. Misneach led the financing, with C10 Labs participating, to back a Cambridge platform that captures business meaning from the people responsible for it and turns that knowledge into governed semantic models, transformation code, and validation rules.

Moatassim (Mo) Aidrus, co-founder and CEO, and Hussnain Ahmed, co-founder and Chief AI Officer, built ekai with co-founder and CCO Tero Miikki after careers around enterprise data and AI systems. Their bet is pointed: dashboards and query histories show what a company has done, but they do not automatically explain what the business intended.

Ekai starts with domain experts. It profiles warehouse data in place, asks the people who own each definition to declare what is true, converts those answers into machine-readable artifacts, and reconciles the result against the underlying data before publication.

That sequence matters because enterprise AI is moving from generating drafts to making decisions. A wrong summary is inconvenient. A wrong definition of revenue, active customer, inventory, or risk can travel through an agent, a dashboard, and an operating meeting before anyone notices that the answer was precise and useless.

The company calls its method forward-engineering. Ekai says early engagements have compressed work that historically took 3 to 6 months into as little as 6 hours, but that is a company-reported benchmark, not an independent performance study.

Ekai runs inside a customer's environment and says data is read in place, not copied out. That deployment model matters for regulated and security-conscious buyers as much as generation speed.

The capital will support product development, go-to-market expansion, and deeper platform integrations. It also gives Ekai a harder commercial assignment: prove that companies will pay to author business meaning deliberately instead of asking another model to infer it from the exhaust of yesterday's systems.

Enterprise AI keeps getting better at producing answers. Ekai is financing the institutional work required to decide which definitions those answers may use, who owns them, and how they survive the next system change.

Deeper Analysis 👇 https://devcuration.com/articles/ekai-raises-1-7m-verified-enterprise-ai-context

Tempo Therapeutics

A graft cannot solve the wound beneath it before that wound is ready.

When skin cancer surgery leaves exposed bone in the head or neck, the surgeon needs a vascularized bed capable of receiving the next reconstruction. Waiting can mean pain, infection risk, and repeated care after the cancer has already been removed.

Tempo Therapeutics has won an NIH Small Business Innovation Research award worth up to $2.25M over 2 years to test whether its MAP Wound Matrix can shorten that wait.

The accounting matters. NIH RePORTER records a $1.42M FY2026 award from the National Cancer Institute. Tempo says the potential total is approximately $2.25M, subject to progress and available funds. This is staged federal R&D support, not a $2.25M payment on day 1.

MOSAIC II will be led by Stephanie Deshayes, Tempo's VP of R&D. The Direct-to-Phase II SBIR funds cGMP manufacturing, regulatory-grade chemical characterization, and a randomized multicenter study in 20 adults with complex wounds after non-melanoma skin cancer resection.

Patients will be randomized 1:1. The primary endpoint is time to a graft-ready wound bed, defined as more than 75% granulation tissue through blinded photo review. MAP has to help a surgeon reach a defensible clinical decision sooner, not merely look impressive in a materials-science diagram.

Tempo earned the next study with a 40-patient first-in-human trial. It met its safety endpoint. MAP-treated wounds reached a favorable Wound Bed Score about 14 days earlier and showed better scar outcomes at roughly 6 months. Overall closure time was similar, so the signal is wound-bed quality, not a blanket claim that every wound heals faster.

Co-founder and CTO Westbrook Weaver helped build MAP as a fully synthetic, flowable scaffold that supports cells and blood vessels moving through it, then gradually resorbs as the patient's tissue takes over. Interim CEO Eric Richman now carries that science into a regulatory and commercial program.

The timing creates a clear division of labor. Tempo announced a separate $14.5M private financing last week for regulatory progress and commercialization. The NCI award funds a bounded proof package: manufacture the material correctly, characterize it to regulatory standards, and test a meaningful endpoint in harder wounds.

Tempo has promising human evidence, an FDA De Novo submission, private capital, and now federal support for the next trial. The work remains inside the wound, where enough healthy tissue must arrive before the next operation can begin.

Deeper Analysis 👇 https://devcuration.com/articles/tempo-therapeutics-2-25m-nih-sbir-wound-study

Basecamp Research

A biological sample collected under a benefit-sharing agreement can travel far without moving again. It becomes sequence data, model context, a therapeutic hypothesis, and, if the science holds, a design intended to change a patient's cells.

Basecamp Research has raised a $140M Series C to finance that handoff. S32 led the oversubscribed round. NVIDIA, Anthropic's Anthology Fund, NATO Innovation Fund, The Rockefeller Foundation, Singular, True Ventures, and others joined. Reuters reports an $800M valuation.

Co-founders Glen Gowers and Oliver Vince started the London company in 2019 around a biological AI weakness: most models learn from a narrow slice of life. Basecamp built partnerships across more than 30 countries, assembled proprietary evolutionary data, and trained EDEN to read and design DNA across a much wider biological record.

The new money is aimed at the less cinematic part of AI drug discovery. Basecamp plans to train a new EDEN generation and advance six AI-designed therapeutic programs toward clinical development, beginning with in vivo cell therapy. The approach pairs long DNA sequences designed by EDEN with enzymes intended to insert them precisely into the genome.

That matters because today's cell therapies can require cells to be removed, engineered, manufactured, and returned to a patient. Basecamp wants to reprogram cells inside the body, potentially making treatment more sophisticated and simpler to administer. The company says current manufacturing can cost hundreds of thousands of dollars per patient.

The evidence boundary matters just as much. Basecamp reports strong preclinical results across several modalities and disease areas. It has not provided a clinical-trial start date, and the six programs do not yet carry public patient data. The model may design a candidate. Biology still gets the final vote.

Basecamp is building for that next vote. Phil Lorenz leads technology as CTO. John Finn brings gene- and cell-therapy experience as CSO. Former Biogen executive Richard Pearce is joining as CBO to expand pharmaceutical partnerships. S32 General Partner and former Verily CEO Andy Conrad is joining the board.

The Series C turns Basecamp's data advantage into a development obligation. Investors are backing the full stack from biodiversity partnerships to models to therapies. Now each layer must preserve enough truth for the next one to work.

Clinical development will decide whether EDEN learned biology deeply enough to write something the body can use.

Deeper Analysis 👇 https://devcuration.com/articles/basecamp-research-raises-140m-ai-designed-therapeutics

Aqua Medical

The procedure begins with equipment a gastroenterologist already understands: a standard endoscope, a working channel, and a catheter carrying heated vapor.

What Aqua Medical is asking that setup to do is much less familiar. Its investigational PIMA procedure targets proximal intestinal mucosa in an effort to change metabolic signaling in people with type 2 diabetes.

Relevance Ventures led an undisclosed multimillion-dollar Series B to help Aqua Medical complete the U.S. RESTORE-1 pilot study and prepare for a planned pivotal trial, subject to regulatory authorization. Dean Newton, Relevance chairman and general partner, also joined Aqua Medical's board.

The amount matters less than the handoff it is financing. Aqua Medical's RF vapor ablation system has FDA clearance for its indicated gastrointestinal use. That clearance does not cover treatment of type 2 diabetes. PIMA remains investigational, and its safety and effectiveness for diabetes have not been established.

The engineering can place a catheter through a scope without an incision or fluoroscopy. The clinical program must determine whether treating proximal intestinal mucosa can produce a meaningful, durable metabolic effect.

Early evidence supplies a reason to continue, not a reason to declare victory. A peer-reviewed first-in-human pilot enrolled 27 patients, with 25 receiving the treatment dose. Investigators reported no serious adverse events, technical success in all procedures, and preliminary HbA1c reductions. They also described the study as small, open-label, single-center, uncontrolled, and insufficiently powered to establish efficacy.

RESTORE-1 brings that question into a U.S. pilot. Aqua Medical reported its first patient treated at UNC Health in July after prior international development. The Series B gives the team room to finish the study and build the regulatory work behind a potential pivotal trial.

CEO Robert "Bob" Haggerty and founder Virender K. Sharma, MD now carry that transition. Relevance is putting prevention-focused capital behind a future that depends on evidence surviving clinical scrutiny.

Aqua Medical completed a $15.5M Series A in 2020. The undisclosed Series B cannot produce a new total. It clarifies the obligation: turn a familiar route into a credible diabetes treatment without letting existing GI clearance outrun the indication being tested.

The catheter can fit through the scope today. The company now has to build the evidence that determines where the procedure belongs tomorrow.

Deeper Analysis 👇 https://devcuration.com/articles/aqua-medical-series-b-pima-diabetes-trial

Numeral

A company can add a state, product line, or billing system before lunch. The tax obligation may surface months later, after ownership has bounced between finance, operations, accounting, and whoever still opens the government mail.

Numeral raised a $100M Series C to make that gap less expensive. Insight Partners led, with Salesforce Ventures, Geodesic, Benchmark, Mayfield, FCVC, Y Combinator, and Uncork participating. Numeral now reports $157M in total funding.

Sam Ross, co-founder and CEO, did not arrive at this problem through a whiteboard exercise. Ross had operated ecommerce businesses and dealt with multistate sales tax himself. Matt DuVall, co-founder and CTO, brought product-engineering experience from Stripe and Notion.

The work is gloriously unsexy: register with the right agency, classify the product, calculate the rate, file the return, move the money, track the exemption certificate, and read the letter from a state nobody remembered adding to the workflow.

Numeral connects those jobs through a deterministic tax engine, AI, automation, in-house specialists, and service. It reports more than 3,500 customers, 90-country coverage, and over 40 billing, finance, and ERP integrations.

California's SB 122 will extend sales-and-use-tax treatment to certain prewritten software delivered electronically or accessed remotely in 2027. Product rules, customer location, billing data, and filing obligations are moving closer together, whether the org chart is ready or not.

Numeral reports 327% year-over-year transaction-volume growth and expects to process more than 80M transactions. Those are company-reported signals, not audited results, but they show the responsibility customers are handing over.

The capital will fund product development, expansion across software, manufacturing, distribution, and wholesale, and hiring. Numeral is also bringing accounting firms deeper into the model so advisers can refer, resell, implement, or work alongside the platform.

That partner channel may matter as much as another automation layer. Tax rules can be deterministic while the facts remain messy: a warehouse moves, a product bundle changes, or a certificate expires. Software can catch more of that motion, but trust stays with the people answering when a filing goes wrong.

The $100M gives Numeral room to make those handoffs less visible to the customer. Its next stage will live inside thousands of ordinary deadlines, each arriving while the business keeps selling somewhere new.

Deeper Analysis 👇 https://devcuration.com/articles/numeral-raises-100m-series-c-for-ai-tax-compliance

Soteris

A profitable insurance book can still carry millions of dollars in value destruction one policy at a time.

That is the gap Soteris has spent 5 years trying to make visible. Founder and CEO Sunit Shah built the company around a stubborn P&C reality: carriers sell a policy before its claims cost is known, then manage uncertainty through averages that can hide profitable risks beside policies draining margin.

Soteris is coming out of stealth with a product that estimates policy-level profit contribution across the policy lifecycle. The company says the API responds in under 250 milliseconds and insurers can act without changing rates, forms, filings, or headcount across the whole book.

The capital story needs accurate timing. Soteris disclosed more than $8M in Seed funding on September 22, 2026, led by Spider Capital with Intact Private Capital, Amplify Partners, DCVC, Webb Investment Network, and Overlook Ventures participating. An SEC Form D shows $8.05M sold from a $9.4M equity offering, including converting SAFEs, with the first sale in June 2025. This is a public disclosure tied to the launch, not a second $8M round raised this month.

That sequence says something about the build. Soteris says its first loss-ratio product has been live since 2020 and has scored more than 100M submissions representing over $180B in premium. The company reports 5 to 15 points of loss-ratio improvement for customers. Several proofs of concept for the new profit layer identified potential book EBITDA increases of 70% to 125%.

Those outcomes are company-reported, not independently audited. Their credibility depends on the signal holding across books, lines, contracts, market cycles, and governance reviews.

Spider Capital Partner Minsoo Chi, also listed as a Soteris director, is backing a narrow form of applied AI: make the existing book's economics visible before an insurer reaches for broad rate action or blunt portfolio cuts. The syndicate brings insurance, technical, and early-stage context.

U.S. P&C insurers posted $60.9B in net underwriting income in 2025, according to AM Best. A strong year can still hide the policies creating or consuming the margin. Soteris is betting profitable growth lives below the average, where carriers can protect good business while the bad is visible.

The financing carried Soteris through stealth. The public work now moves into the carrier workflow, where every score eventually has to change a real decision about what to write, fix, renew, or walk away from.

Deeper Analysis 👇 https://devcuration.com/articles/soteris-discloses-8m-seed-with-policy-profit-ai-launch

Commonweal Ventures

Washington can open a market and scramble the investment case in the same election cycle.

Commonweal Ventures built Fund II around the problems that stay expensive after the campaign banners come down.

The New York firm has closed $54M in new capital: a $51M Fund II plus a separate $3M co-investment vehicle from the State of New York. The plan is roughly 20 pre-seed and seed investments, generally between $500K and $2.5M, in companies working across healthcare, energy, public safety, government services, defense, manufacturing, and other markets where public institutions shape demand.

That last part is easy to flatten into “government tech.” Nate Loewentheil and Ron Bloom are underwriting something more specific: businesses that can use government as a customer, investor, regulator, R&D partner, or market maker without becoming captive to one administration's favorite program.

There is real money in Washington's priorities. There is also real danger in confusing a speech with a durable budget, a campaign promise with procurement, or today's signature initiative with tomorrow's bipartisan need.

Commonweal's answer is a network built to read both sides of that ledger. Its advisers include people who have run federal agencies, worked across presidential administrations, led major cities, and operated inside the industries the portfolio serves. That network cannot make procurement simple. It can help founders understand which public-sector relationships create scale and which ones create a dependency with an election date attached.

The first fund offers an early record, with the right attribution. The Wall Street Journal reports that Commonweal's $23.2M Fund I backed 21 startups and, according to the firm, produced a 17% net IRR and 1.5x net TVPI as of June 30. The portfolio includes companies such as Concourse, Crux, Starbridge, and Advocate, each working where private software meets public systems or nationally important infrastructure.

Fund II now has to turn that thesis into another portfolio while political capital is getting louder and more crowded. The attractive companies will not be the ones that learned the current administration's vocabulary fastest. They will be the ones solving a public problem sturdy enough to keep its budget, buyer, or regulatory importance when the vocabulary changes.

That is a narrower lane than “sell to government,” and a more demanding one. Commonweal just raised $54M to find about 20 founders capable of staying in it.

Deeper Analysis 👇 https://devcuration.com/articles/commonweal-ventures-closes-54m-fund-ii

Enveda

Evolution has been testing chemistry for billions of years. Pharma still spends fortunes asking whether one molecule deserves the next trial.

Enveda built PRISM to search the chemical work that plants, microbes, and the human body have already done. It reads mass-spectrometry data, guides experiments, and helps scientists decide which signals may become medicines.

That thesis just attracted a $311M Series E led by Catalio Capital Management. Durable Capital Partners, ICONIQ, Lightspeed, Surveyor Capital, accounts advised by T. Rowe Price Investment Management, Digitalis Ventures, an unnamed sovereign wealth fund, and Alderline Group joined as new investors. Baillie Gifford, Premji Invest, FPV Ventures, True Ventures, Kinnevik, Dimension, Lifeforce Capital, and Lux Capital returned.

It brings Enveda's reported total raised to more than $845M. Catalio co-founder George Petrocheilos joined the board.

Viswa Colluru founded Enveda around a useful provocation: drug discovery has become capable at developing a good molecule and painfully expensive at finding the right one. CTO August Allen and the team built a physical loop where AI narrows the search, automation tests candidates, and medicinal chemistry prepares the strongest for development.

Now comes the part that makes every platform slide answer to patients.

Enveda has 3 medicines in human trials. ENV-294 is in Phase 2 for atopic dermatitis and asthma after the company reported an average 85% improvement in eczema severity at day 42 in Phase 1b. ENV-308 showed favorable tolerability across 88 healthy volunteers in Phase 1 and is being developed for metabolic health, including weight maintenance after GLP-1 treatment. ENV-6946 is in Phase 1 for inflammatory bowel disease.

Those results deserve attention and restraint. ENV-308 has not yet proved weight-maintenance efficacy, the ENV-294 result remains early, and none of the candidates is approved. The $311M finances the handoff from promising discovery to the larger trials, development systems, regulatory work, and clinical decisions that can punish a weak molecule no matter how elegant the model that found it.

The investor mix is the tell. Healthcare specialists are underwriting clinical judgment. Technology investors are underwriting the platform loop. Enveda has to make both cases true inside one company.

PRISM made nature's chemistry searchable. This round pays for the slower, more consequential work of learning which discoveries can carry the weight of a medicine.

Deeper Analysis 👇

https://devcuration.com/articles/enveda-raises-311m-series-e-for-clinical-drug-pipeline

UltraSight

A hospital can own the ultrasound cart and still lack the trained hands to turn it into a usable cardiac study. The machine is present. The clinical answer is waiting on acquisition skill, quality control, interpretation, and a workflow that can move before the patient does.

UltraSight just raised a $24M Series B2 to work on that gap. ALIVE HealthTech Fund led, joined by Deep Insight, Star51 Capital, Connecticut Innovations, eHealth Ventures, Mayo Clinic Ventures, NYU, and Iron Nation.

CEO Davidi Vortman and co-founder and CTO Itay Kezurer are building around a practical constraint: sonography expertise takes years to develop, while the need for timely imaging keeps showing up beyond the echo lab.

UltraSight's Echosystem combines training, real-time AI guidance, and analytics so trained non-sonographer clinicians can acquire focused cardiac-ultrasound images under qualified physician oversight. Its FDA-cleared software guides probe position for 10 standard adult transthoracic echo views across compatible systems. It assists acquisition without turning interpretation into an unsupervised software task.

That distinction matters because the evidence is getting more specific. A JAMA Cardiology study evaluated 1,302 focused cardiac-ultrasound exams acquired by novice operators for a moderate-or-greater aortic-stenosis workflow. 96.6% were suitable for automated analysis, with 93% sensitivity and 96% specificity. Expert review of positive and uninterpretable exams lifted positive predictive value from 49.4% to 91.1% while reviewing roughly 10% of exams.

That is a credible workflow signal, not permission to generalize one study across every cardiac condition. Mayo Clinic is an investor and research collaborator, and the announcement disclosed Mayo's financial interest. Medical AI earns trust by carrying those limits into the story.

UltraSight says the Echosystem is deployed across more than 10 commercial sites. The new capital will support U.S. expansion, ICU and circulatory-support workflows, and strategic collaborations. Rick Geoffrion has now joined as board chair.

The commercial test is now larger than whether AI can guide a probe. UltraSight has to fit training, hardware compatibility, physician oversight, reporting, and patient management into a repeatable health-system purchase. The cart may already be in the building. This round pays for the clinical handoff around it.

Deeper Analysis 👇

https://devcuration.com/articles/ultrasight-raises-24m-series-b2-for-cardiac-ultrasound-ai

Ema

A request that starts in HR can end up touching IT, payroll, finance, policy, identity, and a manager who has no idea why the ticket is now wearing their name.

That handoff is where enterprise AI stops being a demo and becomes an operating decision.

Ema has raised a $77M Series B led by Creaegis, with Accel, S32, and Prosus increasing their investments. The round brings company-reported total funding to $140M and more than quadruples Ema's valuation from the prior round. The valuation remains undisclosed.

Surojit Chatterjee, founder and CEO, and Souvik Sen, co-founder and CTO, built Ema around AI Employees that execute work across enterprise systems instead of answering a prompt and leaving the human to finish the process.

Ema says its platform supports more than 240,000 Wipro associates across 65 countries, handling more than 2.9M queries a year and enabling 100+ actions. The company reports that some resolutions moved from days to seconds, employee satisfaction increased 20%, and HR operations cost fell 50%.

Those case-study figures are not independently audited, although Wipro's reporting confirms the scale. The deployment shows agentic software touching policies, systems, and requests.

Creaegis is underwriting the economics behind that claim. Managing Partner and CIO Prakash Parthasarathy described Ema as a product platform rather than a services model dependent on bespoke implementations.

Enterprise AI can quietly become consulting with a model attached. The software looks repeatable until every customer needs another integration team, exception map, and group of people keeping the agents from wandering into production furniture.

Ema reports 50x revenue growth over 24 months. TechCrunch says it has 50+ active enterprise deals, 1M+ active users, and $150M+ in bookings, which represent multiyear contract value, not ARR. Ema also reports roughly 180% net dollar retention and gross margin near 80%.

The $77M will fund product development, go-to-market expansion, and growth across APAC, EMEA, South America, and the Middle East. Every new market adds systems, regulation, language, and operating history.

If Ema can carry one governed layer across those differences while customers expand from one workflow into dozens, AI Employees start competing with software seats, implementation budgets, service desks, and manual operations at once.

The next proof will arrive when a request crosses departments and still reaches an accountable outcome without summoning another project team.

Deeper Analysis 👇 https://devcuration.com/articles/ema-raises-77m-series-b-enterprise-ai-employees

Hubble Network

The package that disappears in transit is rarely valuable enough to carry a satellite modem. Neither is the tool left at a jobsite, the pallet in the wrong yard, or the label moving through 3 logistics networks.

Physical visibility has an economic border. Companies track what can justify the hardware, power, service plan, and integration. Everything below that line travels on trust and somebody's spreadsheet.

Hubble Network just raised a $200M Series C led by Smith Point Capital at a $1.6B valuation, bringing total funding to $300M. Seraphim Space, Carthona Capital, Earthshot Ventures, Y Combinator, and RPM Ventures participated.

The financing arrived with an operating milestone: Hubble opened its satellite network to Bluetooth Low Energy devices.

Co-founder and CEO Alex Haro and co-founder and CTO Ben Wild built the company around a stubborn idea. Instead of adding a cellular modem, GPS receiver, SIM, or dedicated satellite radio, Hubble uses the Bluetooth chip already inside the product. A firmware update sends supported location and sensor data through terrestrial gateways or satellites.

That changes the purchasing decision. Hubble says connectivity can add less than $0.50 to a new design's bill of materials. Tracking can move past containers and vehicles into packages, tools, and routine inventory that used to be cheaper to lose than connect.

The company reports 100M+ terrestrial gateways across 170 countries and 3M+ square miles, plus 500,000+ active devices, more than 10 times the prior-year level. It currently operates 6 satellites in low Earth orbit and plans to reach 60 by 2030.

Those numbers are Hubble's, and the next test is operational. A signal from orbit proves the physics. Customers still need predictable coverage, useful update intervals, secure data, supported hardware, and economics that hold beyond pilots.

The partner list shows how Hubble intends to close that gap. Texas Instruments supports select BLE chips. InPlay is working on sub-$1 tracking hardware. Reelables brings printable labels to cargo. Samsara extends equipment visibility. Muon Space is building the next satellites.

Each partner removes a reason for the customer to wait.

The $200M funds a constellation, but Hubble is really moving the price line that decides which parts of the physical world deserve to be visible. If that line keeps falling as the network gets denser, the largest market may be the ordinary objects businesses stopped trying to track years ago.

Deeper Analysis 👇

https://devcuration.com/articles/hubble-network-raises-200m-for-bluetooth-satellite-scale