Trust Is the First Product

Members Mobile CUSO

Members Mobile CUSO just secured a strategic investment from four credit union investors, led by The Reseda Group, the wholly owned CUSO of MSU Federal Credit Union. The amount remains undisclosed, but the message could not be much louder. Credit unions are putting capital behind a model they believe belongs in members' pockets every single day.

Executive Chairman Gary Brandt, CEO Luis Jimenez-Tuñon, CTO/COO Peter Nussbaumer, and the Members Mobile team reached this milestone with backing led by The Reseda Group. Building a company is hard enough. Building one that convinces institutions known for measured decisions to become owners instead of spectators is a different level of validation.

The name Members Mobile carries more weight than it first appears. This is not just about mobile phones. It is about keeping members connected to the organizations they already trust. That tiny device in everyone's hand is no longer just a communications tool. It has quietly become the front door to financial relationships. Ignore that reality long enough and someone else gets invited inside.

That is what makes this investment interesting. Members Mobile combines nationwide 4G/5G mobile service with credit union financial products through its VIMplicity platform, giving institutions another way to engage members beyond the occasional app login or branch visit. Sometimes innovation is not about inventing a brand-new category. Sometimes it is about connecting categories that have been sitting next to each other for years, pretending they never met.

The Reseda Group deserves recognition here as well. Great investors are not simply writing checks. They are placing informed bets on where industries are heading before everyone else decides it was obvious all along. The three additional participating credit union investors, while undisclosed, reinforce that this is more than one institution making an educated guess.

Founders should pay attention to the pattern. Capital rarely falls in love with PowerPoint slides. It gravitates toward founders who solve real problems, understand their customers with almost uncomfortable precision, and create alignment where incentives finally make sense for everyone involved.

Members Mobile is operating at the intersection of telecommunications and financial services, but the bigger lesson reaches far beyond either industry. The companies creating the next wave of value are finding practical ways to make trusted relationships even more useful. That is usually where durable businesses begin.

Reo.Dev

Reo.Dev just raised an $11.3M Series A to pursue one of enterprise software's most difficult challenges: identifying buyer intent before traditional sales signals ever appear. Not the kind buried in another dashboard that tells you what already happened. The kind hiding inside engineering teams quietly evaluating software long before procurement committees and polished slide decks enter the conversation. That is the bet.

Achintya Gupta, Gaurav Jain, Piyush Agarwal, and the Reo.Dev team closed the round led by Elevation Capital, with participation from Heavybit, India Quotient, Foster Ventures, new investor Uncorrelated Ventures, and a group of angel investors and AI leaders. With this financing, Reo.Dev has now raised $15.3M in total funding.

Selling to developers has always been its own universe. Engineers rarely wake up hoping to fill out a lead form. They explore. They test. They clone repositories. They read documentation. They install packages. They solve problems.

By the time a traditional CRM notices movement, the interesting part of the story has usually already started somewhere else.

Reo.Dev built its AI-native GTM platform around that reality. Instead of asking sales teams to guess where demand might exist, it helps companies selling to engineering and technical teams understand where developer interest is already taking shape. That is a meaningful shift because timing has always beaten volume. More noise has never been a growth strategy. Better signals usually are.

Elevation Capital is backing a category that refuses to fit neatly inside yesterday's sales software. Heavybit, India Quotient, Foster Ventures, and Uncorrelated Ventures see the same opportunity. Markets have a funny habit of rewarding founders who solve problems everyone complains about but few actually fix.

The lesson for founders is worth stealing. The best companies rarely begin by asking, "How do we build another feature?" They ask, "Where is everyone looking in the wrong direction?" Reo.Dev looked past vanity metrics and toward developer behavior, where buying decisions often begin long before anyone schedules a demo.

The headline is the capital. The real story is conviction: founders staying obsessed with a difficult problem, customers validating that approach, and investors backing the next stage of growth. Watching AI reshape GTM is fascinating. Watching companies figure out where the real signals live before everyone else notices is even better.

Final Boss Sour

Final Boss Sour just raised $4M in strategic funding, bringing its total funding to $12M. The round signals that investors see more than another snack brand. They see a company that built a community first, then developed products to match it. Attention may be rented, but community has to be earned.

CEO James Hicks, Tommy Riggs, London Lazerson, and the Final Boss Sour team closed the round with backing from Evolution VC Partners, The Angel Group, Mondelez International SnackFutures Ventures, Melitas Ventures, GFR Fund, and other strategic investors. It feels less like another funding headline and more like another checkpoint unlocked.

The snack aisle has never exactly been known for originality. Bright wrappers, louder claims, same old formulas. Final Boss Sour looked at that landscape and decided the real boss battle was making sour snacks from real dried fruit instead of artificial gummy candy, then wrapping the experience inside a gaming universe that feels like it belongs on an old-school console instead of another forgettable package on a shelf. Sometimes the biggest innovation is realizing consumers want something worth talking about before they want something worth chewing.

The numbers explain why investors showed up. More than 2B organic video views across TikTok, YouTube, and Instagram. The largest YouTube subscriber base of any sour candy brand. The largest sour candy brand on TikTok Shop. A six-week retail test across roughly 2,000 Walmart stores. More than $1M in TikTok Shop sales during its early retail phase. Those are not vanity metrics. They are evidence that storytelling, product, and distribution can move together instead of waiting for one another to catch up.

Another lesson is hiding beneath the sour coating. The founders did not build around celebrity. London Lazerson has an audience of more than 10M followers, yet the brand deliberately stands on its own instead of borrowing a creator's face as a shortcut. That takes confidence because brands built on borrowed attention usually expire when the algorithm gets bored.

This new capital will help expand into Walmart, Kroger, H-E-B, Wegmans, and Hy-Vee, with Target and 7-Eleven planned for Fall 2026, while fueling new flavors and collaborations. Investors are backing more than candy. They are backing a repeatable engine where content creates curiosity, curiosity drives trial, and trial earns loyalty. In a market full of companies chasing the next level, Final Boss Sour remembered the oldest gaming lesson of all: beat the level in front of you first. Then keep pressing start.

Emergent

Emergent just turned a $130M Series C into a very loud market signal: software creation is moving from code-first to intent-first. The San Francisco company closed the round at a $1.5B valuation, led by Creaegis alongside MNI Ventures, Claypond Capital, and Sentinel Global, with Khosla Ventures, SoftBank Vision Fund 2, Lightspeed, and Y Combinator participating in the round. CEO Mukund Jha, CTO Madhav Jha, and the Emergent team reached this milestone with backing from both new and existing investors. Smart capital does not usually return because a deck looked pretty. It returns when the numbers start doing the talking.

Emergent is building an AI software creation platform that lets people create production-ready full-stack web and mobile apps using natural language instead of traditional coding. That is the headline. The traction is the punchline: more than 12M applications built, more than 200K paying customers, a $120M annualized revenue run rate, users in more than 190 countries, and roughly 70% of users who have never written code before.

That last number is the one to watch because it turns Emergent from another AI funding story into a distribution story. The platform is not only selling speed to engineers; it is selling agency to people who have been sitting on software ideas without a realistic path to ship them. For decades, software development rewarded people who could speak the language of machines. Now machines are learning to speak the language of people, and the market is starting to price what happens when millions of operators, small-business owners, and non-technical founders can build real software without waiting for a traditional engineering queue.

Mukund Jha and Madhav Jha did not stumble into this moment. Their backgrounds across Google, Amazon SageMaker, Dropbox, Sigma Computing, Mixpanel, Zenefits, Sandia National Laboratories, Columbia Engineering, and Penn State gave them the technical foundation to build something more substantial than a prompt-to-prototype tool. Emergent's multi-agent architecture is designed to generate, review, test, debug, deploy, and monitor applications, which is the difference between a demo that gets applause and infrastructure that customers pay for.

The founder lesson here is not really about AI. Investors chase momentum, but momentum is usually the receipt, not the purchase. The purchase is years of technical depth, customer obsession, rapid iteration, and solving a problem people are willing to pay to make disappear. Emergent's rise has been impressive. The bigger story is that the definition of builder may be expanding faster than the software industry is ready to admit.

SONATA

SONATA just came out of stealth with a $7M seed round led by Lux Capital, BoxGroup, and Sunflower Capital, with strategic participation from operators and founders at Ramp and Linear. CEO Sagan Schultz, MD, MBA, David Deng, and the SONATA team are making a significant bet on a painful truth: people have more health data than ever, and much of it still sits around waiting for someone useful to connect the dots.

That is the market gap SONATA is walking into. Wearables track sleep. Bloodwork produces pages of biomarkers. Genetic testing can surface risks most people barely know how to interpret. The average person has access to more signals than their parents could have imagined, yet the experience can still feel like carrying a stack of disconnected printouts into a system built to react after something breaks.

SONATA is trying to make prevention feel like an operating model rather than a wellness slogan. The company combines whole-genome sequencing, deep biological testing, clinical AI built in-house, and ongoing physician-led care into a membership launching across New York, San Francisco, and Los Angeles. The premise is simple: understand the person before the problem becomes louder than the plan.

There is a lesson here for every founder, even outside healthcare. Markets do not reward companies for generating more information when customers already feel buried by inputs. They reward companies that turn complexity into decisions. That is true in health, security, finance, infrastructure, and almost every corner of enterprise software pretending another dashboard will save the day.

The investor mix makes the round more interesting. Lux Capital, BoxGroup, and Sunflower Capital are backing a model where AI supports clinical judgment instead of replacing it. That matters because healthcare buyers do not need software that sounds smart in a demo and disappears when judgment gets hard. Operators and founders from Ramp and Linear add another product-building signal, which matters because healthcare does not need another polished interface layered on top of the same confusion. It needs systems that make hard information usable.

That difference is the story worth tracking.

The hard part starts now. SONATA has to prove that a premium preventive membership can create sustained value over time, not just a sharper launch story. But the thesis is worth watching closely. The next wave of healthcare AI may not be about removing people from care. It may be about giving physicians and patients a better way to act before the damage shows up on the bill.

General Compute

General Compute just secured a debt facility of up to $400M from Upper90 Capital Management. Read that again because this is not another venture round where everyone celebrates a bigger valuation and pretends dilution is a personality trait. This is infrastructure finance meeting AI at full speed, with capital backing hardware that most lenders would have avoided a year ago.

CEO Finn Puklowski, CTO Jason Goodison, the General Compute team, and the Upper90 team led by CEO Billy Libby closed a financing deal that reflects conviction surviving the uncomfortable questions. The facility begins with an initial $100M and scales with customer demand toward the full $400M. That matters because disciplined capital wants utilization, customers, and assets that produce cash flow, not just applause.

General Compute is chasing one of AI's most important battlegrounds: inference. Training models grabs headlines. Running them efficiently is where businesses either print money or burn it. The company has built an ASIC-first inference neocloud using AMD MI300X chips for prefill and SambaNova SN40 and SN50 chips for decoding, all exposed through an OpenAI-compatible API so customers can migrate without rebuilding applications.

The company says its platform can deliver up to 16× faster inference than conventional GPU clouds while securing more than $300M in price-protected hardware commitments and options on roughly 15 MW of air-cooled rack capacity. Air-cooled infrastructure drawing around 20kW per rack instead of well north of 120kW is not just an engineering detail. It is the difference between waiting years for new facilities and deploying into existing capacity measured in weeks.

Another layer is worth watching. Upper90 is reportedly financing inference-specific chips as collateral rather than the NVIDIA GPUs that have dominated previous AI infrastructure lending. Markets do not whisper when they find a new asset class. They test it with real money.

General Compute is an appropriate name because compute is becoming general infrastructure. Electricity did it. Bandwidth did it. Inference is lining up for its turn.

Founders chasing capital should notice the financing path. General Compute previously raised a $15M seed round led by FUSE VC with participation from Carya Venture Partners, Village Global Ventures, and Evercrest Capital Partners, following an earlier pre-seed round before reaching this milestone. Equity opened the door. Creative debt may accelerate the race.

Self Inspection

Self Inspection just closed a $10M funding round led by Sandberg Bernthal Venture Partners, with participation from U.S. AutoForce, Westlake Financial, Costanoa Ventures, Rebellion Ventures, BrightCap Ventures, DVx Ventures, and Karim Bousta, former VP of Worldwide Service at Tesla. CEO Constantine Yaremtso, Co-founder Evgeny Shevelev, CPO, CFO, and Board Director, and the Self Inspection team are betting that one of the automotive industry's oldest workflows is ready for a fundamental redesign.

Every industry has that process everyone complains about but somehow keeps doing the same way. In automotive, vehicle inspections have lived in that category for decades. Clipboards. Subjective opinions. Photos that tell half the story. Everyone accepts the friction because that is how it has always been done.

Self Inspection looked at that mess and asked a better question. What if every smartphone became a professional inspection tool? What if AI could create a trusted, standardized record that followed a vehicle through its entire lifecycle instead of starting from scratch every time it changed hands? That is a much bigger opportunity than speeding up inspections. It is about creating confidence where billions of dollars depend on accurate condition data.

The numbers tell their own story: more than 1M vehicle inspections completed, more than $80M in customer savings, and more than 300K operational hours returned to businesses that would rather move inventory than argue over scratches and dents. Those are not vanity metrics. That is operational leverage.

The investor lineup makes perfect sense. Sandberg Bernthal Venture Partners recognized a category with room for a true system of record. Strategic investors like U.S. AutoForce and Westlake Financial know this is not theory because they live these workflows every day. Costanoa Ventures, Rebellion Ventures, BrightCap Ventures, DVx Ventures, and Karim Bousta doubled down on a team executing where it matters.

One of the biggest lessons founders can take from this raise is that enterprise customers do not buy AI because it is fashionable. They buy outcomes: save time, reduce costs, improve consistency, and build trust. AI just happens to be the engine under the hood.

The fresh capital will help Self Inspection expand its Vehicle Condition Intelligence platform, deepen its AI capabilities, reach more enterprise customers, and continue its expansion into Europe.

Infrastructure businesses rarely make the loudest headlines, but they are often quietly collecting compound interest while everyone else chases the next shiny object. When you become the trusted source of truth inside an industry, you are no longer selling software. You become part of how business gets done.

Databricks

Databricks just reminded the market that the biggest companies in AI are not necessarily the ones making the loudest noise. More often, they are the ones quietly building the infrastructure everyone else eventually realizes they cannot operate without.

The company announced it has signed a term sheet for a new strategic funding round led by existing investor Coatue at a $188B valuation, with the transaction expected to close later this summer. Databricks did not disclose the size of the financing in its announcement, although The Wall Street Journal has reported the investment is approximately $3B. The valuation itself tells you exactly how investors view the company's role in the future of enterprise AI.

CEO Ali Ghodsi, CTO Matei Zaharia, Executive Chairman Ion Stoica, and the Databricks team did not build this position overnight. They spent years turning Apache Spark into the foundation for something much bigger. While everyone else was chasing the next feature or the next headline, Databricks kept expanding the platform underneath modern data and AI. Funny how disciplined execution suddenly looks like perfect timing once the market catches up.

Today, more than 20,000 organizations, including 70% of the Fortune 500, use Databricks. That is not the profile of a company riding a trend. That is the profile of infrastructure becoming indispensable.

The strategy is just as interesting as the valuation. Databricks says the new capital will accelerate its AI roadmap through products like Unity AI Gateway, Genie, and Lakebase while also supporting future AI acquisitions and deeper research. None of those investments exist in isolation. Together, they strengthen an ecosystem that makes enterprise AI more governable, more scalable, and considerably harder for competitors to displace.

There is a lesson here that founders should pay attention to. Markets love shiny demos until procurement, compliance, security, and operations enter the room. Then the conversation changes. Durable companies solve the problems that appear after the excitement fades. They become the systems customers quietly build their businesses around.

That is why infrastructure companies keep surprising people. They rarely dominate the daily news cycle, but they often end up defining the next decade of enterprise technology.

Databricks appears determined to keep building that foundation, and Coatue's continued conviction suggests sophisticated investors believe the biggest opportunities in AI will not belong only to the companies creating intelligence. They will also belong to the companies making that intelligence usable, trusted, and operational at enterprise scale.

Neo

Neo just stepped out of stealth with $100M in total Seed and Series A funding, and the timing says almost as much as the term sheet. AI agents are moving from sidekick demos into the enterprise bloodstream. They can invoke tools, call APIs, move data, trigger workflows, and make software feel less like a static interface and more like a digital workforce. That sounds great right up until security asks the only question that matters: who, or what, just did that? That is the gap Neo is chasing.

The Boston-based company is building what it calls Agentic Software Control, a platform designed to give SecOps teams inventory, risk intelligence, attribution, policy control, and real-time enforcement across AI agents, AI-enabled applications, browsers, identities, plugins, MCP servers, and software that is rapidly gaining autonomous behavior.

Andreessen Horowitz led Neo's Seed round. Bessemer Venture Partners led the Series A. Craft Ventures and Merlin Ventures also participated. The capital will support engineering, product development, go-to-market initiatives, and operational growth as enterprises figure out that AI autonomy does not politely wait for governance committees to catch up.

The founding team is not learning cybersecurity from a pitch deck. CEO Nicholas "Nick" Warner helped scale SentinelOne through major commercial growth and its IPO. CTO Shlomi Salem spent more than a decade building advanced threat research capabilities. Chief Product Officer Eran Shirazi brings deep product and engineering experience from EasySend and Unit 8200. That matters because category shifts usually reward operators who have already seen the last platform war up close.

The market signal is sharp. Gartner projections cited in Neo's launch materials say enterprise applications with agentic capabilities could rise from 5% in 2025 to 40% by the end of 2026. CyberArk research, cited by Bessemer, shows that 68% of enterprises deploying AI agents currently have no identity controls governing them. So the enterprise is giving AI agents more agency while still arguing about the control plane. Classic.

The bigger lesson is simple. Investors do not just fund products. They fund timing, technical judgment, founder credibility, and the uncomfortable moment when old assumptions stop working. Neo is betting that endpoint security, identity, and AI governance are about to collide into a new control category. AI agents are multiplying. Enterprise adoption is accelerating. Security does not get a vote on whether this happens. It only gets to decide whether it is ready when the software starts acting on its own.

Apaluma

Apaluma just raised $5.55M in Seed funding, and this one says a lot about where serious enterprise AI is finding traction. Not in another chatbot chasing attention. Not in another dashboard collecting digital dust. This is about taking decades of fragmented government regulatory records and turning them into something people can actually use without needing a treasure map and three cups of coffee.

Founder & CEO Alicia J. Keyes and the Apaluma team closed the round led by Crosslink Capital and Kickstart, with Haven Ventures and Overwater Ventures joining as co-leads. Orange DAO and NM Vintage Fund also participated as pre-seed and seed investors. Capital follows conviction, and conviction usually follows founders who have lived the problem instead of simply studying it from across the street.

Alicia J. Keyes spent years working at the intersection of economic development and government before building Apaluma. That experience shows up in the product. The company is transforming fragmented regulatory records into a unified ecosystem by connecting permits, environmental filings, violations, enforcement records, and legacy documents through data ontology rather than simple keyword searches. That distinction matters because context beats chaos every time.

The name Apaluma almost feels fitting. Pulling scattered pieces of information into the light is exactly what the platform is designed to do. Government agencies have spent years surrounded by valuable data that too often behaves like a library where someone removed every aisle marker. The information exists. Finding it quickly is the real challenge.

The company's flagship deployment, Project Velocity, with the New Mexico Environment Department illustrates the opportunity. According to company-reported figures, the platform organized more than 18M pages across more than 715,000 regulatory documents while integrating 15 environmental data APIs into a searchable experience powered by its AI assistant, Luma. The company also emphasizes traceable responses backed by cited sources, an important distinction for regulatory environments where accuracy is not optional.

There is a business lesson buried inside this funding announcement that founders should pay attention to. Venture firms rarely invest because technology sounds impressive. They invest when founders combine domain expertise with a problem that refuses to disappear. Software can be copied. Lived experience usually cannot.

For organizations navigating environmental compliance, permitting, public records, or regulatory operations, the value proposition is straightforward. Faster access to trusted information creates faster decisions. Faster decisions create better outcomes.

Trellis Bioscience

Trellis Bioscience just raised $9M in Series C funding, bringing its total equity financing to approximately $35M. The Redwood City clinical-stage biotechnology company is using the round to advance calpurbatug, also known as TRL1068, through a Phase 2 trial for chronic prosthetic joint infections. That is not the kind of market that wins because the deck looks expensive. It wins because the science keeps surviving contact with reality.

New Science Ventures and Easton Capital continued backing the company. AMR Action Fund, The Doctor Group of Dallas, Texas, and orthopedic surgeons with deep prosthetic joint infection expertise joined the round. Junjun Gao, Principal at the AMR Action Fund, also joined the Trellis board, giving the syndicate a clear antimicrobial-resistance signal rather than the usual biotech wallpaper.

The company is developing native human monoclonal antibodies through its CellSpot platform, which identifies naturally occurring antibodies from human donors. Calpurbatug targets bacterial biofilms, the protective structures that make prosthetic joint infections so difficult to treat with standard antibiotics.

That detail matters. PJI is not a clean software problem with a better dashboard waiting at the end. It is implanted hardware, infection, surgery, antibiotic resistance, patient pain, hospital cost, and a clinical system looking for something better than brutal tradeoffs.

CEO Dr. Lawrence M. Kauvar, President & COO Dr. Stefan Ryser, Chairman Stote Ellsworth, CFO Jeff Blake, Chief Business Officer Elisabeth Whitney, Dr. Estelles, and the broader Trellis Bioscience team are building in one of those categories where patience is not optional. Biology does not care how polished the narrative sounds. It cares whether the mechanism works.

The Phase 2 trial is evaluating TRL1068 with DAIR, or debridement, antibiotics, and implant retention. Trellis expects topline results in Q2 2027. That timeline will not satisfy anyone addicted to instant momentum, but serious biotech rarely moves on internet time.

The lesson for founders is simple. Capital does not always chase the loudest story. Sometimes it follows years of evidence, technical discipline, regulatory progress, and a team willing to keep walking through the difficult part of the work. That is the part worth watching. If the Q2 2027 data strengthens the case, Trellis Bioscience will have a sharper argument for antibody science in hard-to-treat infections.

A trellis does not grow the plant. It gives growth something to climb. That is what this round looks like for Trellis Bioscience: structure around the science, fuel for the next clinical milestone, and another reminder that the best companies often compound before the market notices.

Ulisses

Ulisses just raised a €1,088,050 seed round, approximately $1.3M, to accelerate its next stage of growth. More importantly, the funding gives the company something every infrastructure startup needs: time to build, recruit, and turn an engineering advantage into a market advantage.

CEO Mattia Tartaglia, CTO Alessandro Spadaro, and the Ulisses team secured backing from ARGO Accelerator, the MOST National Center for Sustainable Mobility, and more than 30 angel investors coordinated by Stefano Ceci. The round also includes operators such as former Telepass CEO Gabriele Benedetto, FederlegnoArredo Eventi CEO Luca Palermo, Fic CEO Maurizio Tursini, and former NTT Data Italia VP Teresa Trivisonno. It looks less like passive capital and more like a tactical industrial network.

Ulisses is not selling another shiny IoT demo. The company is working on one of the least glamorous and most expensive problems in the physical economy: knowing where critical assets actually are with enough reliability to make decisions. Bad location data creates delayed shipments, idle equipment, unnecessary labor, and dashboards that look expensive while quietly lying to everyone in the room.

That is why the company's patented mesh network and intelligence layer matter. Ulisses is trying to turn messy field signals into trustworthy operational data for companies that cannot afford to guess. Warehouses, rail depots, smart ports, and logistics networks do not need more noise. They need infrastructure that tells the truth when the environment gets ugly.

The seed capital will support commercial growth, team expansion, and continued technology development. That is the right sequence for infrastructure companies. Technical credibility opens the door, but commercial execution decides whether the product becomes a successful pilot or a system customers start depending on.

The lesson for founders is simple. Serious investors still fund hard problems when the pain is real, the buyer is obvious, and the technology has a reason to exist beyond sounding good in a pitch deck. Every digital transformation eventually runs into the physical world. Companies like Ulisses are making sure that world can finally talk back with accurate data. That is not hype. That is infrastructure doing its unglamorous, useful job.

Mandrake Bio

Mandrake Bio just raised approximately $1.9M in a pre-seed round, and the company is pursuing a vision that stretches well beyond another AI application. It is using AI to design the building blocks of life. That is what makes Mandrake Bio interesting. The Bengaluru-based startup raised the round co-led by Activate and Antler, with participation from Spectrum Impact, DeVC, and strategic angel investors including Dr. Vijay Chandru, Paras Chopra, Sanjiv Rangrass, and Vatsal Dusad.

Founder and Chief Editing Officer Tanay Lohia and Chief Scientific Advisor Dr. Kutubuddin Molla are building at the intersection of generative AI, structural biology, computational biology, and wet-lab validation. Mandrake Bio is not trying to squeeze another percentage point out of existing gene-editing systems. It is taking a harder path: designing programmable gene-editing enzymes from scratch instead of modifying naturally occurring tools like CRISPR-Cas9. That is a fundamentally different engineering philosophy. When biology becomes something you design instead of something you merely discover, the design space gets much bigger.

The first commercial focus is agriculture, where more precise and compact gene editors could help accelerate crop improvement, improve resilience, and reduce agricultural inputs. That is where the platform thesis starts to matter. Longer term, the same infrastructure has implications for therapeutic applications as gene-editing technologies mature.

The funding will help expand Mandrake Bio's AI protein-design platform, grow research capacity across AI and biophysics, strengthen compute infrastructure, and accelerate wet-lab validation of engineered enzymes.

There is a lesson here for founders. Deep-tech investors rarely fund PowerPoint ambition. They fund conviction backed by technical execution. Mandrake Bio is operating in a category where the problem is hard, the validation cycle is real, and the upside compounds if the infrastructure works.

Some of the most valuable AI companies of the next decade may not look like software companies at all. They may look like biology companies, materials companies, chemistry companies, and infrastructure businesses quietly using AI as the engine underneath.

Activate and Antler are backing an ambitious technical vision early, alongside Spectrum Impact, DeVC, and strategic angel investors who understand both biotechnology and company building. Engineering biology is a long game. The founders building that future deserve attention before everyone else realizes what they are creating.

Extellis

Extellis just added Shell Ventures to its extended seed round, and while the amount remains undisclosed, the signal could not be much louder. When one of the world's largest energy companies decides to back a satellite imaging startup, it is not collecting startup trading cards. It is buying a front-row seat to what comes next.

CEO Dr. Michael Boyarsky, Dr. David R. Smith, COO Jeff Ludwig, and the Extellis team welcomed Shell Ventures as the newest investor. Oval Park Capital, Duke Capital Partners, First Star Ventures, New Industry VC, Front Porch Venture Partners, EGB Capital, and Blue Lake VC continue backing the mission they recognized from the beginning.

The company name is Extellis, and that's fitting because it is extending what satellites can actually see. Traditional optical satellites are incredible until clouds show up, the sun clocks out, or Mother Nature decides she's in charge. Synthetic aperture radar does not ask for perfect weather. It simply gets to work. Extellis is taking that capability and scaling it with patented metasurface antenna technology developed through Duke University research, allowing satellites to capture dramatically higher volumes of SAR imagery while keeping the economics practical. Sometimes innovation is not about making something louder. It is about making something impossible to ignore.

This company grew from Dr. Michael Boyarsky's doctoral research alongside Dr. David R. Smith's pioneering work in metamaterials. That matters because the best startups rarely appear out of thin air. They spend years hiding inside laboratories, whiteboards, rejected experiments, government research grants, and countless conversations that nobody outside the room ever hears. By the time everyone calls it an overnight success, the founders have already lived several careers.

The funding itself also tells founders something worth remembering. Before venture dollars arrived, Extellis secured $3.3M in non-dilutive funding, including $2.3M in DARPA support. They built real technology, protected meaningful intellectual property, assembled credible investors, and focused on solving a problem that industrial customers actually care about. Capital usually follows conviction that has already been demonstrated. Investors may write the checks, but founders earn the opportunity long before the paperwork starts.

The applications stretch far beyond pretty pictures from orbit. Agriculture, energy infrastructure, disaster recovery, maritime operations, and environmental monitoring all depend on reliable Earth observation. Data loses value when clouds become gatekeepers. Reliable information becomes infrastructure, and infrastructure has a funny habit of creating entirely new markets once the economics finally make sense.

Quorum

Quorum received a strategic investment from Enlightenment Capital, and the interesting part is not the missing dollar figure. It is what the deal says about where enterprise AI is starting to matter. Quorum, co-founded by CEO Alex Wirth and Jonathan Marks, has spent more than a decade building software for one of the messiest knowledge jobs in business: government affairs. Federal, state, local, school board, EU, and international policy environments do not wait politely for teams to catch up. The work is dense, political, high stakes, and usually buried under more documents, stakeholders, hearings, and regulatory shifts than any human team can comfortably process.

That is where Quorum has built its position. The Washington, D.C.-based company provides an AI-powered government affairs and public affairs platform for legislative tracking, stakeholder management, grassroots advocacy, PAC management, and policy intelligence. More than 2,000 organizations use the platform, including over 50% of the Fortune 100.

Those numbers matter because this is not AI as a parlor trick. It is AI embedded in a workflow where better information can change timing, strategy, and outcomes. The investment will help Quorum accelerate its product roadmap, expand its agentic AI capabilities, and strengthen its role as a system of record for government affairs professionals.

There is a useful lesson here for founders. Build around a mission-critical workflow. Earn trust over years. Become infrastructure before everyone else realizes the category is infrastructure.

Government affairs has never had an information shortage. It has had a translation problem. Too much policy movement, too many stakeholders, too much noise, and not enough clarity about what deserves action right now.

That is why this investment is worth watching. Enterprise AI is moving away from generic assistants and toward systems that sit directly inside important workflows. The durable companies will not just summarize work. They will understand the data, the context, the sequence, and the consequences well enough to help teams decide what to do next.

Quorum is trying to close that gap by turning policy data into decisions. Enlightenment Capital is backing the next phase of that bet, and the broader market signal is clear: vertical AI with domain depth is starting to look far more durable than generic automation with a better demo. That is the quiet part of the AI market right now. The most valuable products are not always the flashiest ones; they are the ones serious teams trust when the work is too complex, too regulated, and too consequential to fake.

Runta

Runta just raised a $20M seed round at a post-money valuation exceeding $100M, led by Andreessen Horowitz. CEO Guanlan Dai and the Runta team reached a milestone that says something bigger than the size of the check: enterprise AI is moving from model envy to operational control.

For years, the AI conversation has obsessed over models. Bigger models. Faster models. Smarter models. Meanwhile, enterprises are asking the less glamorous question that eventually decides whether this technology gets trusted with real work. What happens after an AI agent starts making decisions with access to inboxes, credentials, production systems, and payment methods? That question is where Runta lives.

The company is building the runtime execution layer and sandboxing infrastructure for autonomous AI agents. In plain English, it focuses on what an agent can actually do once it is running. Limit access. Cap spending. Protect secrets. Govern network egress. Monitor every action. Create accountability. None of those capabilities earn the loudest applause at AI conferences, but every one becomes priceless the first time an autonomous agent touches production.

Andreessen Horowitz leading the round makes sense. Infrastructure becomes more valuable as complexity compounds, and every new agent introduced into an enterprise creates another moving part that needs boundaries. Software has guardrails because humans learned expensive lessons. AI agents are about to learn from the same history unless companies build the control layer before the traffic shows up.

The timing matters as much as the technology. Organizations are moving beyond experiments. They are giving agents real work, real permissions, and real responsibility. That changes the conversation from "Can AI do this?" to "Should AI be allowed to do this?" Those are two very different boardroom discussions.

The founder lesson is worth paying attention to. Runta did not raise capital by chasing the loudest trend. It identified the friction that appears after adoption begins. Markets reward companies that solve tomorrow's operating risk before customers feel it at scale.

The investor syndicate also reflects the category's importance. Alongside Andreessen Horowitz, investors including Jeff Dean, Fei-Fei Li, Ali Ghodsi, Ram Shriram, and Thomas Wolf are backing the company. AI is becoming less about building intelligence in isolation and more about building trust around what that intelligence is allowed to do.