The Evidence Before the Title

Intrepid Growth Partners

When prediction gets cheaper, the expensive part of AI moves to everything around it. Judgment, proprietary data, distribution, and workflow redesign become the scarce assets.

Intrepid Growth Partners just put $525M behind that observation.

The firm formally launched with the final close of Fund I, an inaugural growth vehicle backed by more than 80 global LPs. Temasek, Abu Dhabi Investment Council, British Business Bank, Export Development Canada, and Business Development Bank of Canada are among the institutions named in the announcement.

That roster is backing a specific view of where value lands after a model makes prediction cheaper: inside companies that can reorganize an industry around better decisions.

Mark Machin brings the institutional lens after leading CPPIB and spending more than 2 decades at Goldman Sachs. Mark Shulgan built growth-equity platforms at CPPIB and OMERS. Ajay Agrawal studies why cheaper prediction makes its complements more valuable.

Together, they are building Intrepid around growth-stage businesses with commercial traction. Fund I had already invested in 9 companies before launch, including PhysicsX and StackAdapt, across healthcare, engineering, cybersecurity, tax, advertising, and enterprise software.

That 9-company starting line matters because it exposes the thesis to different operating systems. Intrepid now has to help founders carry AI into customer budgets, regulated environments, factories, clinics, and enterprise workflows.

Intrepid is treating the Canada-UK corridor as a source of research, engineering talent, and globally ambitious founders, while investing across the US and Europe. Toronto and London anchor a strategy that assumes the next important AI company does not need a Silicon Valley birth certificate.

The interesting test now sits below the model layer. Can these companies own the data, distribution, trust, and workflow change required to move from a sharp technical capability into the system an industry actually runs on?

A $525M final close does not answer that question. Intrepid has not disclosed fund performance, individual LP checks, or capital deployed. It gives the firm a durable mandate across operating systems where being wrong costs much more than a bad demo.

Fund I is capital with a thesis attached. Prediction may keep getting cheaper, but the institutions, founders, and teams responsible for turning it into better decisions are where the expensive work is moving.

Deeper Analysis 👇

https://devcuration.com/articles/intrepid-growth-partners-closes-525m-ai-growth-fund

Temporal

Samar Abbas and Maxim Fateev spent nearly 20 years teaching software how to remember what it was doing after something failed. AI agents just made that old problem worth $12.55B.

Temporal has raised a $550M Series E as its Durable Execution platform moves closer to the center of enterprise AI. The round includes Lightspeed, Wellington Management, Goldman Sachs Alternatives, Tiger Global, and others, plus returning backers a16z, Sequoia, Index, GIC, Sapphire Ventures, and Amplify.

An AI demo can look brilliant in an afternoon. Production is where the calendar, the API, the approval chain, and the network start taking turns breaking the spell. Agents may work for days, weeks, or months, so they must preserve state, recover from failure, and continue without asking a developer to rebuild the trail by hand.

That is the job Temporal prepared for before “agentic” became a funding adjective. Abbas and Fateev worked on workflow infrastructure across Amazon, Microsoft, and Uber, then founded Temporal in 2019 around a simple engineering promise: ordinary code should survive extraordinary amounts of real-world mess.

Temporal persists each step of a workflow and handles the retries, timeouts, task queues, and recovery underneath it. That matters when software is moving money, waiting for a person, coordinating services, or letting an AI agent act across systems it does not control.

The market is rewarding that distinction quickly. Temporal's $300M Series D in February valued the company at $5B. Seven months later, the Series E more than doubles that valuation. Temporal says annualized revenue run rate grew more than 200% year over year, net dollar retention stayed above 200% since February, and August volume reached 1.9T billable actions.

The $550M will support global operations, core platform primitives, reliability, and security. Temporal says its team doubled to 570 people over the past year. Capital at this scale gives the company room to become a standard, while a $12.55B valuation makes every enterprise deployment and product promise more visible.

For founders and operators, the lesson is sitting inside the failure path. The agent that wins may be the one that can wait 3 days for approval, survive an outage, remember every decision, and finish the work without turning recovery into a human scavenger hunt.

Temporal spent years making software resume where it left off. Now the company has $550M to make that execution history as valuable as the intelligence running above it.

Deeper Analysis 👇 https://devcuration.com/articles/temporal-raises-550m-series-e-durable-ai-infrastructure

Shennon Biotechnologies

Li Sun is moving from CEO to President and CTO at the company she founded. Shennon Biotechnologies is putting Cyril Konto in the CEO seat while $12M of new financing pushes its science into a more demanding job.

The San Francisco biotech built an integrated engine for precision immunotherapy. PINTRA identifies targets from human tumor data. TCELERATOR screens more than 10M single-cell interactions at once. AFFINIS designs and optimizes binders. SERIS looks for off-target safety risks earlier.

Solid tumors do not hand drug developers one clean problem. A target still needs a functional, selective binder, safety evidence, regulatory work, manufacturing discipline, and a development path that survives contact with the clinic.

ShennonBio is attaching that platform to 3 named programs. SBT-121 is a trispecific antibody T-cell engager for small cell lung cancer. SBT-201 targets ovarian cancer. SBT-425 is a TCR-based engager for hepatocellular carcinoma that can pursue intracellular targets presented through HLA.

Future Ventures, NextGen Venture Partners, Samos Investments, Atypical Ventures, and Saras Capital supplied the new financing, with DCVC, Foundation Capital, and AV8 Ventures continuing their support. ShennonBio reports $25M raised to date. No lead, valuation, or named round type was disclosed.

Cyril Konto brings more than 20 years in oncology development and company building. At Ichnos Glenmark Innovation, he led the organization when AbbVie licensed a trispecific T-cell engager for $700M upfront and up to $1.225B in potential milestones. That history explains the operating experience investors are putting beside Li Sun's platform.

Li Sun remains responsible for scientific strategy, technology development, and pipeline innovation. The handoff treats company building and scientific invention as parallel jobs just when both become harder.

Over the next 2 years, ShennonBio plans to move a development candidate toward an IND filing and Phase 1 development, pursue selected licensing, and explore platform partnerships. Those milestones will turn company-reported scale, including more than 10,000 tumor samples and 11M single cells, into evidence the market can judge one program at a time.

The $12M buys time, people, and experiments. The more revealing decision is how ShennonBio divided responsibility for carrying a rare-cell discovery engine toward the clinic without asking the founder to stop inventing it.

Deeper Analysis 👇 https://devcuration.com/articles/shennonbio-raises-12m-and-names-cyril-konto-ceo

Agilis Air

The Army can now buy Agilis Air's Remora through a digital storefront.

The difficult part begins after the checkout button.

Agilis Air raised $1.4 million in early funding to move its drone program from qualification into repeatable production. EagleNXT, North Coast Ventures, Vici Capital, and angels participated.

The company says the money will help ramp its Youngstown, Ohio facility, complete certification work, and deliver against its U.S. Army pipeline.

That sequence matters more than the size of the round.

Remora has been accepted to the Army's UAS Marketplace, a procurement system designed to let approved buyers compare and purchase qualified drones. The Army said in August that the broader marketplace had already supported more than $750 million in sales.

That is not Agilis revenue. It is evidence that the purchasing channel is real.

Now Agilis has to prove it can serve that channel.

Its Remora is a Group 1 first-person-view quadcopter built for reconnaissance, training, and operation in contested or GPS-denied environments. Its current product page lists the aircraft from $1,500 and a complete system at about $2,500. It says the supply chain is fully domestic and NDAA compliant.

Those details put Agilis inside a demanding manufacturing equation in defense technology:

Build in the United States.

Keep the aircraft inexpensive enough to be treated as replaceable.

Make it consistent enough for soldiers to trust.

Then do it again at volume.

CEO Shawn Theiss brings decades of experience building custom UAVs for defense customers. Executive Chairman Yon Raz-Fridman brings the company-building side. The operational bet is that bespoke experience can become a standard product line without losing the speed that made a small manufacturer useful.

Strategic capital may help. EagleNXT has already described its investment as a way to connect Agilis's manufacturing with its own drone platforms and software. North Coast Ventures and Vici Capital add Ohio-based capital around a Youngstown production story.

But a marketplace listing is an opening, not an outcome.

The next proof points are less glamorous: certification completed, units delivered on time, field feedback incorporated, and follow-on orders earned.

For Agilis, $1.4 million is enough to cross the next bridge. It is not enough to hide weak execution.

That makes this a clean early-stage test. Can a small Ohio manufacturer turn easier federal purchasing into dependable American drone production?

Deeper Analysis 👇 https://devcuration.com/articles/agilis-air-raises-1-4m-for-low-cost-army-drones

Aron

Procurement teams are hired to negotiate. Then the week disappears into supplier follow-ups, dirty exports, renewal language, and invoices that somehow forgot the price everybody agreed to.

Dominic Toselli knows that gap from the buyer's side. Before founding Aron, formerly NegotiateAI, Dominic Toselli led worldwide logistics procurement at Apple. The company is now disclosing $8M in backing from Menlo Ventures and Storm Ventures across its early pre-seed and seed financing.

The funding amount is useful. The handoff it finances is the better story.

Aron calls itself procurement's chief of staff. It organizes spend and contracts across ERP environments, runs RFQs, chases suppliers, checks clauses against the company's playbook, watches renewals, and matches invoices to the agreements behind them.

People keep the approvals and judgment. Aron takes the tactical queue that keeps those people from using either one.

That distinction matters in procurement because savings can vanish after the meeting. A buyer negotiates the rate. The supplier sends the invoice. A line drifts, an exception gets buried, or the agreement lives in a folder nobody checks until the renewal has already renewed itself.

Aron says its first RFQs average 21% savings and customers can prove ROI inside 3 weeks. Those are company-reported figures, not independently audited benchmarks. Its no-integration start is a sharp promise: enter through files teams already trust, then perform well enough to earn deeper access.

Menlo Ventures lists the company under its former NegotiateAI name, records a 2025 Seed partnership, and identifies Dominic Toselli as Founder & CEO. Public sources do not disclose the split between pre-seed and seed, a lead investor, valuation, or security type, so the honest accounting is $8M in total backing.

The market has plenty of software that can show a procurement leader where work is stuck. Aron is building for the next sentence: take it from here.

That sentence gets expensive when the system has to understand supplier identity, contract language, approval thresholds, category context, invoice variance, and the politics of asking a vendor to try again. It also becomes valuable when a skilled team gets time back for the negotiations and supplier decisions that still require a person in the room.

Aron's next proof will arrive quietly, one RFQ, renewal, and invoice at a time. The product becomes credible when procurement stops managing the agent and starts receiving completed work it can trust.

Deeper Analysis 👇 https://devcuration.com/articles/aron-raises-8m-procurement-chief-of-staff

Nara Health

Buying health benefits is supposed to settle a problem. Too often, it assigns the problem to an employee with an insurance card, three phone numbers, and an afternoon nobody planned to lose.

Nara Health is raising against that administrative gap.

The Chicago company, formerly Avant Health, has raised $14M across pre-seed and seed financing. The latest piece is a $10M seed led by Khosla Ventures, according to Axios.

Nara says it supports 25,000 members and has processed more than $600M in claims. Its platform acts as an AI-enabled third-party administrator for employer health plans, covering plan design, member support, claims, and prior authorization.

That is a broader job than putting a chatbot beside a benefits portal. A member question can touch eligibility, plan language, a provider network, clinical documentation, pricing, and an employer's plan design. Every handoff creates another place for the answer to become somebody else's problem.

Sidhartha Sinha and John Zicker are building Nara around the opposite idea: the administrative layer should carry the issue through.

The company reports average savings of 15%, member engagement of 68%, a five-second average call-answer time, and prior-authorization turnaround within four hours. Those are company-reported figures, not independent benchmarks. They still reveal what Nara expects to be judged on: cost, responsiveness, and whether members can reach an outcome without learning the org chart of American healthcare.

Nara's wager is that software and human support belong inside the same operating system. Automation moves routine work. People stay available when the routine breaks.

That puts the third-party administrator in a more interesting position than the acronym suggests. A point solution can improve one step and leave the employer managing the seams. An administrator that shapes the plan, supports members, processes claims, and handles authorization can work across the sequence. It also owns more of the consequences.

Khosla Ventures is backing a test of whether modern software, accountable operations, and real humans can make healthcare's middle feel less like a relay race where everyone drops the paperwork.

The next proof arrives as Nara grows. Can its savings and service levels hold? Can automation absorb more work without flattening the exceptions? Can employers buy fewer seams instead of one more dashboard?

That is where a seed round becomes an operating story.

Deeper Analysis 👇 https://devcuration.com/articles/nara-health-raises-14m-ai-native-health-plans

Cornelis Networks

The most expensive machines in an AI cluster can spend a shocking amount of time waiting for the network.

That is the opening Cornelis is financing with $205M. The company is expanding from scale-out interconnects into scale-up networking, where the fabric inside a rack influences how much useful work those accelerators deliver.

Cornelis introduced Active Compute Fabric alongside the raise. The architecture combines lossless transport, in-fabric acceleration, and programmable compute. The network can operate on data in transit, adapt to workloads, and offload collective operations that would otherwise leave accelerators waiting.

CEO Lisa Spelman is leading that move with CTO Charles Archer and the co-founders who still run critical parts of the business: Philip A. Murphy Jr., president and COO; Gunnar K. Gunnarsson, VP of Solutions Delivery and Support; and Vladimir K. Tamarkin, VP of Platform Engineering.

That team turned Intel's former Omni-Path business into an independent networking company. CN5000 is shipping. The 800G CN6000 is sampling with customers ahead of expanded availability expected in Q4 2026. The capital will support manufacturing, deployments, production, partnerships, and the next product roadmap.

Qualcomm Technologies is part of the product story. Tony Pialis, EVP and GM of Data Center, is joining Lisa Spelman at AI Infra Summit to discuss networking as a first-order rack-scale design choice.

The capital accounting deserves precision. Cornelis calls this $205M in funding, but it has not disclosed a round label, valuation, full investor roster, or exact mix of securities. A July SEC filing showed $112M sold inside a $245M offering that included equity, debt, warrants or options, and convertible notes. The public record does not fully reconcile those figures yet.

IAG Capital Partners led Cornelis' 2022 Series B. Joel Whitley, an IAG partner, supplied the investor perspective in today's announcement, framing open-standard scale-up and scale-out networking as a market worth more than $55B by 2030. The official release does not name a lead for the new financing.

Cornelis now has enough capital to make networking carry a larger share of the AI system's workload. Customers will decide whether openness and programmability translate into better utilization and a credible alternative inside the rack. That decision now moves into CN6000 qualification labs, manufacturing lines, and production clusters.

Deeper Analysis 👇 https://devcuration.com/articles/cornelis-raises-205m-active-compute-fabric

Open Cosmos

Satellite intelligence can arrive after the fire, flood, or communications failure has already changed shape. Open Cosmos is financing the minutes in between.

The European space company has raised a €300M Series C, approximately $347M at the ECB reference rate on September 14. ETF Partners led the round, with Lightrock, ICF - Institut Català de Finances, Entrepreneurs First, two international pension funds, and a broader syndicate participating.

Rafel Jorda Siquier founded Open Cosmos in 2015. Jordi Barrera Ars and Aleix Megias Homar remain beside him as CTO and COO. Their company has moved from making small satellites easier to launch into joining spacecraft, communications, data processing, and customer decisions into one operating system.

OpenOrbit builds and operates missions. OpenConstellation lets governments share Earth-observation infrastructure. ConnectedCosmos adds sovereign broadband and direct-to-device IoT, while DataCosmos turns imagery and sensor feeds into operational intelligence. A government responding to a wildfire, damaged infrastructure, or a communications outage buys a shorter route from something changing on Earth to a decision made by people who can still influence the outcome.

Open Cosmos says its latest constellation can cut delivery of Earth-observation intelligence from as long as 48 hours to as little as 30 minutes. The company also reports 5 consecutive years of profitable growth, $370M+ in contracts signed over 3.5 years, and capacity to manufacture 1 satellite a day across 4 factories in the UK, Spain, Portugal, and Greece.

These are company-reported signals, not audited factory output or fleet-wide service proof. They matter because the business already has customers, missions, factories, and a decade of technical decisions behind it.

The €300M will expand manufacturing and push ConnectedCosmos, OpenConstellation, and DataCosmos further into deployment while Open Cosmos adds engineering, manufacturing, and software talent.

ETF Partners and the syndicate are financing a demanding handoff: stated capacity into repeatable production, constellation growth into dependable coverage, and orbital data into intelligence that arrives before the situation on the ground becomes somebody else's version of events.

Europe's sovereignty argument now reaches the factory schedule, spectrum plan, software queue, and customer waiting for an answer. Open Cosmos has the capital to connect those pieces at a scale where the minutes saved can become the product.

Deeper Analysis 👇 https://devcuration.com/articles/open-cosmos-raises-300m-series-c-european-space

responsAbility Investments

Blended finance begins with an awkward question: who agrees to take the first loss when a clean-energy loan goes wrong?

For responsAbility Investments AG's Asia Climate Fund, part of the answer came from public and development-finance institutions. Germany's BMZ, acting through KfW, supplied EUR58M for a first-loss tranche. FMO entered as an anchor investor. IFC joined the structure.

The result is a $461M final close for responsAbility's largest closed-end climate investment fund. That number is the vehicle's total commitments, not a fresh $461M dropped into the account in August. The strategy had already reached $460M at its fifth close in January, against a $500M target.

responsAbility says the concessional layer mobilized more than 5 times its amount in commercial capital. Public money did not replace the private market here. It changed the order of risk enough for more of the private market to show up.

The fund will provide private credit to renewable-energy, electric-mobility, energy-efficiency, circular-economy, and related climate businesses across South and Southeast Asia. These are markets where the demand story can look enormous in a presentation while the debt still has to survive local currencies, contracts, policies, counterparties, and equipment that operates far away from an investment committee.

Nadia Nikolova leads responsAbility as CEO. Dr. Stephanie Bilo, Chief Client & Investment Solutions Officer, described the close as evidence that Asia's transition is becoming an institutional investment opportunity. Ewout Van der Molen, Head of Climate Finance, has spent the strategy's public record explaining why the first-loss layer matters.

The fund is already past the theory stage. responsAbility reports approximately $204M committed across 17 portfolio companies, about 44% of final commitments. The manager also reports 8.63M tonnes of lifetime CO2 emissions reductions through year-end 2025 and expects roughly 16M tonnes over the investments' lifetime. Those figures are company-reported and unaudited, which turns measurement into part of the work rather than decoration around it.

Now the capital stack becomes an underwriting record. Every borrower still has to build, operate, collect, repay, and prove that the climate outcome survived contact with the business model. The final close gives responsAbility more room to finance that work across Asia; the 17 companies already in the portfolio are where the structure begins earning its name.

Deeper Analysis 👇 https://devcuration.com/articles/responsability-closes-461m-asia-climate-fund

BoomerangHR

Companies spend years learning how someone works. They see the judgment, the recovery, and the quiet competence that never makes it into a résumé. Then the person leaves, payroll stops, and that knowledge becomes a memory with an old email address.

Boomerang has raised $12.7M around the value companies abandon at that moment.

Kalos Ventures and Album Ventures co-led the seed round, with Monta Vista Capital, Capital Eleven, and Strada Education Foundation participating. BoomerangHR emerged from stealth with an AI-powered platform for offboarding and corporate alumni management.

Founder and CEO Jeff Wald co-founded WorkMarket, the contingent-workforce software company acquired by ADP in 2018. This workforce question begins after the employee has already left.

BoomerangHR connects an employer's HCM and ATS, keeps one profile for each former employee, and matches qualified alumni to open roles. Employers set guardrails around rehire eligibility, seniority, location, and time since departure before outreach begins.

The same record can surface referrals, show where alumni landed, preserve a warm path into an account, and keep exit-interview information useful after the badge stops working.

The outside hiring funnel is getting louder. Gartner predicts that 1 in 4 candidate profiles worldwide could be fake by 2028. ADP payroll data found that returning employees averaged 31% of new hires and reached 35% in March 2025.

The familiar candidate is no longer the fallback. In a noisy market, a known performance record becomes a form of recruiting infrastructure.

Kanish Raajkumar, co-founder and CTO, is building it with co-founder and VP of Engineering Andrei Tulai and co-founder and COO Mallory Meiser. The company says hundreds of employers use BoomerangHR across millions of alumni profiles. That figure is company-reported, and the announcement does not disclose valuation or a detailed use-of-funds plan.

A database cannot repair a bad exit, and automation cannot manufacture trust after a company treated someone's final week like a security incident. The software becomes valuable only when the relationship was worth preserving.

The sharper bet inside this round is that departure has future operating value. BoomerangHR gives HR teams a way to keep the record alive without building a McKinsey-sized alumni department.

The next recruiting advantage may already know the systems, customers, and people. The work begins while that person is still walking out the door.

Deeper Analysis 👇

https://devcuration.com/articles/boomeranghr-raises-12-7m-employee-alumni-ai

SENA Health

By the time a patient reaches a clinician, the healthcare system has already asked a front desk to solve half a dozen other problems. Scheduling, intake, referrals, insurance checks, refills, and prior authorization all compete for the same people and the same hour.

SENA Health has secured an undisclosed Series A led by Rittenhouse Ventures to expand its Clinical Command Center and global operations. The company will also add business-development and sales capacity while continuing to invest in AI and technology.

The amount and valuation were not disclosed. That leaves the useful part of the announcement in plain view: Rittenhouse is financing an operating model built around the work that happens before, between, and after clinical encounters.

Founder and CEO Dr. Anthony J. Wehbe built SENA Health as a physician-led company. Co-founder Joe Lyons and CTO Pieter Oosthuizen are helping turn that idea into a 24/7 access and care-coordination layer for medical groups, health systems, and employers.

SENA's AI agents handle administrative volume across scheduling, intake, referrals, medication refills, prior authorization, and patient communication. Trained coordinators supervise the work and step in when judgment, context, or empathy has to cross the line from software into care.

That combination matters because healthcare AI rarely fails in the demo. It gets tested in the handoff, when a scheduling rule collides with a patient's reality, a referral arrives without the right record, or a call needs escalation instead of a faster answer.

SENA says its system connects to existing EHR and practice-management workflows. The company reports 3M+ digital touchpoints and 500+ trained coordinators. Inc. ranked it No. 12 in its 2026 Northeast Regionals and No. 400 on the 2026 Inc. 5000.

Rittenhouse Ventures invests in capital-efficient B2B software, AI, and tech-enabled services, with deep healthcare-IT experience. Partner Sushma Rajagopalan, a former technology CEO, will join SENA's board as the company moves toward wider deployment.

Series A capital gives SENA room to sell and build while making the human-oversight promise more demanding. More markets mean more integrations, more edge cases, and more moments when a patient needs the system to distinguish a completed task from the right outcome.

The next phase belongs in those moments. SENA Health and Rittenhouse Ventures are betting that administrative scale can return time to care without letting accountability disappear into the software.

Deeper Analysis 👇 https://devcuration.com/articles/sena-health-series-a-rittenhouse-ventures