The Work the Product Has to Learn to Carry

AuxoAI

Amaresh Tripathy described a client that could generate campaign emails faster while its approval process kept the campaign waiting. Making more content would have enlarged the queue. The expensive part of that AI decision was choosing the wrong thing to accelerate.

That is a useful introduction to AuxoAI, which announced a $20.7M Series A led by NTTVC on October 20. Co-founders Amaresh Tripathy, Arvinder Pal Singh and Kabir Singh are building a business around learning how an enterprise actually works before deciding what its AI should do.

The round also includes individual investors George Still Jr., Amit Zavery and Giacomo Sonnino. Their experience spans enterprise software and operating improvement. Their participation should be read as those individuals backing AuxoAI, rather than investments announced by their employers.

AuxoAI puts strategists and engineers alongside customers, keeping the people who understand the problem close to the people building the system. The company reports work with more than 40 clients across manufacturing, consumer goods, healthcare, telecom and other industries.

Damia, its proprietary workbench, is where that delivery work can accumulate. A Skill Catalog captures reusable methods. LinK/C organizes enterprise knowledge and context into a governed graph. The next team can inherit useful engineering patterns while still doing the discovery that belongs to its particular customer.

For an enterprise buyer, that combination changes the purchase. Somebody has to understand why an exception exists, which team can approve a change and how the new process will be used. A model can arrive with astonishing capabilities and still need an introduction to the office.

AuxoAI says the capital will expand its global forward-deployed talent, proprietary technology and delivery capabilities. Google Cloud's separate $750M partner initiative gives this category a larger route into enterprise budgets; AuxoAI has announced a Gemini Enterprise transformation partnership.

The financial question lives inside the delivery model. Reusable methods can save work on the next engagement, while customer-specific discovery consumes experienced people's time. AuxoAI has room to invest in both, with the economics becoming clearer as those teams take on more deployments.

Back in that marketing workflow, the person approving the campaign still has a job to do. AuxoAI's opportunity starts with understanding why the campaign is waiting and stays with the people changing how it moves.

Town

Menlo Ventures put its firm on Town in June. By October, the investor had become a co-lead in the company helping with its inboxes. Getting inside the workday turned into a place on the cap table.

Town announced a $108M Series B on October 19, co-led by Forerunner and Menlo Ventures at a $1B valuation. Andreessen Horowitz and First Round Capital returned, alongside Better Tomorrow Ventures, Precedent.vc, Scribble Ventures, Serena Ventures, Sparked Ventures and angels. The round follows June's $55M Series A.

Co-founders Jean-Denis Greze, CEO, and Tony Vincent, CPO, have worked together since Dropbox. Greze went on to become Plaid's CTO; Vincent led Applied AI Product at Google. They are now building an assistant that can know how somebody works without requiring that person to become an expert in configuring assistants.

Each person gets a Townie, connected to their email, calendar, documents and other tools. Menlo describes assistants preparing briefings, surfacing commitments and chasing follow-ups. That is an investor describing its own use, with all the enthusiasm that relationship brings. It also gives the investment a wonderfully practical starting point: somebody had already let the product near their email.

Now Town is extending that relationship across teams. A Townie can ask a colleague's Townie for information, help find a warm introduction, or run a draft past the assistants of the people who will review it. Each assistant still represents its person. Knowing the answer and having permission to share it become part of the same job.

Town says paying customers grew 4x and paying teams 9x since July, after reaching seven-figure ARR within weeks of launch. Those figures come without absolute customer counts or retention data. The team growth is particularly relevant to a product whose usefulness increasingly depends on the people working around you.

Shared workflows add another layer. Update the team's competitive talking points once, and Town says every rep's assistant can carry that change into call preparation. Adoption becomes part of the workflow itself.

There is a substantial buying decision tucked inside that convenience. Town says companies own the context their assistants build and can take it with them. Its trust policy keeps model training off by default and requires approval before sending emails or calendar invites. A company adopting Town is choosing how its working knowledge gets carried between people, and who gets to let it travel.

Mural

Sinclair Toffa remembers money sent home to Togo losing as much as 20% to the journey. His family had supplied the reason to fix payments long before Palantir supplied the engineering experience.

At Mural, that history now sits behind a business selling financial infrastructure to other businesses. Co-founder and CEO Sinclair Toffa and co-founder and COO Chris Fernandes worked together at Palantir, connecting systems inside large institutions. They brought the integration problem into finance, where a customer waiting for money has little interest in which provider lost the thread.

Mural announced a $15M Series A on October 20, led by Tether, with North Island Ventures and existing investors participating. Total funding is now $28M. The New York company plans to expand its infrastructure, hiring, markets and licensing and compliance footprint.

The product gives fintechs and platforms a shared place to manage accounts, payments, identity and permissions across bank and blockchain networks. Dollars, local currencies and stablecoins can belong to the same customer experience. Behind that experience, someone still has to keep the balance, the person and the permission connected.

Mural's customer examples make the purchase easier to understand. Deel uses it for workforce payouts. Opera uses it for creator payments. The buyer gets engineers who deploy alongside its team, including people who understand the market where the product has to operate. An API contract only gets so far when the bank, the software and the customer all need a different explanation.

Mural reports $2.1B in payment volume since launching its API in 2025, serving end users in 128 countries. Those figures describe money movement through the platform. The company has not broken out how much activity comes from agents.

Agents give this account infrastructure another assignment. Limits, approved counterparties and approvals on larger payments let a business define what software may do with someone's money. That accountability is part of the product being sold. The person funding the account still needs an intelligible record of the decision, even when they never pressed send.

Tether is backing a team that has already had to connect those systems for paying customers. As Mural enters more markets, the founders' original family problem travels with it: the recipient needs the money to arrive, and somebody reachable needs to own the explanation when it does not.

Baba

Alicia Tennenbaum has spent her career getting people home from the hospital. At Baba, her experience in care transitions now sits alongside AI research, inside a company preparing to take more responsibility for what happens between medical appointments.

Founder and CEO Connor Sweeney has raised a $40M Series A for Baba, led by FirstMark. CVS Health Ventures and Valtruis joined returning investors General Catalyst, Genius Ventures and Ground Up Ventures. The October 20 announcement puts total funding at $47M, according to the company.

Tennenbaum previously led care transitions at Mount Sinai and later served as VP of Case Management at Med-Metrix. Baba appointed her to lead clinical operations in April. That is a telling hire for an applied AI company: somebody who knows how a discharge plan survives contact with an insurer, a family and the next provider.

Baba gives patients a dedicated care team, including advocates and licensed therapists, supported by AI. The work reaches into specialist visits, insurance paperwork, hospital discharges, behavioral health and help arranging transportation. A referral can be perfectly documented while the person who needs it is still trying to get there.

That gap gives this business its shape. Families need someone who can keep calling, secure the right information, coordinate the parties and stay with the problem. Health plans need the service to reach people whose care is difficult to organize. Delivering that attention takes labor, judgment and institutional relationships.

Baba says it has supported more than 10,000 patients and families, with more than 100,000 families waiting. Those are company-reported demand figures. The funding will expand clinical teams, advance AI research and develop health-plan partnerships as Baba moves toward value-based arrangements.

Its BabaBench research examines whether AI agents can finish extended advocacy tasks while respecting consent and other safety boundaries. For an insurer call, getting an answer and having permission to share the patient's information are separate responsibilities.

The commercial ambition reaches beyond making an advocate's day more efficient. Baba wants lower coordination costs to make sustained care viable for people who have been expensive to serve. The operating work now runs from Sweeney's AI team through Tennenbaum's clinical organization to the family waiting for the next part of a care plan to happen.

Mach9

Mach9 gave up its own capture hardware after customers asked for help with the drawings. That is an expensive kind of listening.

The team had started with robotics. Customers already had ways to collect detailed scans of roads and utilities. What occupied their people afterward was turning those scans into drawings an engineer could use. More data meant more work waiting in the office.

Alexander Baikovitz, CEO, built Mach9 with Haowen Shi, Michael Mong and Zachary Sussman, now CTO, after meeting at Carnegie Mellon's Robotics Institute. Founded in 2021, the San Francisco company went all in on software in 2024. The founders followed the customer into the less photogenic part of the job.

On October 21, Mach9 announced a $34M Series A co-led by First Round Capital and Bessemer Venture Partners, with Prudence and returning investor Quiet Capital participating. Mach9 puts its total funding at $49M. Quiet led its previously announced $12M Seed in November 2024.

Mach9's AI CAD software takes scans from equipment customers already own and drafts features such as curbs, lane markings, signs and poles. Surveyors and engineers review, adjust and finish the geometry. The person responsible for the deliverable gets a different place to begin. Exports include DXF and DGN, so the drawing can keep moving into existing design tools.

That matters to an engineering firm deciding which contracts its existing team can accept. In a Mach9-hosted case study, Pennoni reports a 50% increase in project capacity without adding headcount. Its team also describes bringing fragmented drafting work into one shared workflow, reducing the coordination required to get a drawing out the door.

The business model follows the work. Bessemer describes pricing by road mile or acre surveyed, giving a project manager a unit that belongs in the estimate. The software can be evaluated beside the assignment it helps deliver, with professional review still part of the job.

Mach9 reports use across 2,500+ projects and 100,000+ miles in all 50 states, including work for WSP, HNTB and NV5. Those are company-reported figures. The financing will support software expansion; Bessemer also points to more feature types, data sources and a larger go-to-market effort.

For Baikovitz and the team, widening that product means following more of the work between a scan and a finished engineering file. The next contract an experienced survey team can fit into its calendar is where that decision keeps paying its way.

Blacksmith

Thousands of CPUs can become somebody else's problem with a change to a GitHub Actions file. Blacksmith is volunteering to be that somebody.

A customer's tests can ask for thousands of CPUs at once, then disappear. Somebody has to keep enough machines ready, get the work onto them and avoid charging the customer for an infrastructure department disguised as a developer tool.

That is the job behind Blacksmith's $66M Series C, announced October 21 at a $1B valuation. Sequoia Capital led, with Premji Invest and returning investors Peak XV Partners, GV and Y Combinator participating. Blacksmith puts its total funding at $125M.

Co-founder and CEO Aditya Jayaprakash built the company with CTO Aayush Shah and co-CTO Aditya Maru. The Waterloo graduates brought experience from Faire and Cockroach Labs into infrastructure for continuous integration, the builds and tests a code change needs before it can merge.

The product fits into GitHub Actions. Teams change the runner configuration while keeping their workflows, and Blacksmith supplies compute, caching and the tools to see where a job failed. A familiar workflow gives the buyer less organizational work to do before finding out whether the service helps.

Ashby's customer story makes the purchase concrete. Its engineers tried larger runners, then more parallel tests, and still ran into queues. Blacksmith's case study describes 31 test shards running together and capacity jumping from 0 to 3,500 vCPUs. A faster machine helps only after the job gets one. For the buyer, the operational question is how long a proposed change waits before useful feedback comes back.

Blacksmith reports more than 10,000 company users, over 90M CI jobs each week and 17x ARR growth since January. Those are company figures. The weekly job count measures work handled; it tells us neither the revenue per job nor the margin left after running it.

There is useful chronology behind the capital, too. The $45M Series B announced in August was raised in March, according to Jayaprakash, so the Series C arrives about seven months after the previous close rather than two months after it.

The Series C will expand the team and compute footprint. Blacksmith says it operates just under 10 MW across US and European data centers and plans to multiply that footprint fivefold in 2027. Its customers can keep changing the code. Jayaprakash, Shah and Maru are buying the capacity to be ready when all those changes arrive for their checks.

Kodama Systems

Merritt Jenkins is giving a logging contractor a new recruiting radius. The skidder stays in the woods; its operator can work from home.

That changes a stubborn staffing conversation. A contractor still needs someone who understands the machine and the crew, but getting that person to a remote jobsite no longer has to be part of every shift. Kodama Systems is building its forestry business around that separation.

On October 19, Kodama announced $7M in financing co-led by Konvoy Ventures and Breakwater Ventures, with Blue Forest Asset Management, NYA Ventures, Breakthrough Energy Ventures, Congruent Ventures, Dangerous Ventures and M1C participating. The company reports $14M in total funding, with this financing supporting deployments, equipment-manufacturer partnerships and autonomy development.

CEO and co-founder Merritt Jenkins started Kodama with co-founder Matt Verminski in 2021. Their company history describes nine months spent visiting loggers before settling on remote operation as a practical route into the industry. That is a useful order of operations for a robotics company: spend enough time around the work to learn which part of the day the customer would gladly buy back.

Kodama's RS1 combines remote control with supervised autonomy. Satellite connectivity carries video and machine information to the operator, while sensors and onboard computing support work in rugged conditions. Experienced people remain part of the operation, including the complicated moments that make a forest a poor place to assume everything will repeat neatly.

The purchase also reaches beyond software. Kodama's skidding service packages equipment, a remote operator, maintenance and support into a day rate. The local crew still has physical work to do, including routine fueling and greasing after the initial deployment period. Somebody has to keep the diesel moving, even when the controls have changed address.

Kodama says equipped machines moved more than 100,000 tons of timber across the southeastern US in the preceding six months. That gives the financing a commercial setting: equipment working alongside crews, generating the experience needed to expand the service.

For contractors, the interesting purchase is a more dependable shift. For Kodama, every additional machine brings another crew, support relationship and stretch of woods into the same operating promise. The work of making that remote shift feel routine now travels with the deployment team.

Reliable Robotics

Reliable Robotics' remote pilot still has somebody to talk to. Air traffic control remains part of the job while the aircraft handles taxi, takeoff and landing.

That human connection is worth keeping in view as the Mountain View company adds $115M in capital. Aviation autonomy has to make sense to the people running an airport, the operator moving cargo and the controller managing everybody else's flight. A demonstration has to fit into their working day.

Co-founders Robert Rose, CEO, and Juerg Frefel, CTO, came out of SpaceX's flight software and avionics work. Reliable applies that experience to existing aircraft, starting its FAA certification effort on the Cessna 208B Grand Caravan. An established airframe gives the engineers a concrete aircraft to automate and operators a familiar machine around which to plan. The sale carries the obligations of an aircraft installation: somebody must maintain the equipment, train the people and keep operations coordinated. Those obligations travel with every additional airport.

Nimble Partners returned to lead the financing, joined by new investors Gideon Capital, RockCreek, Geek Ventures, Gold Hill Capital and Sparkwave Capital, with Island Green Partners participating again. Allen & Company supported the fundraise.

The accounting deserves a clean landing: $25M of the new package is a Silicon Valley Bank term loan, leaving $90M in equity. Reliable calls the enlarged Series D $275M, including that debt, after April's $160M financing. The company reports a valuation above $1B and more than $380M in cumulative equity capital.

There is a useful public witness to the operating work. On October 19, the FAA reported a Reliable demonstration with Albuquerque's aviation department using a remotely piloted Caravan at the Sunport. The program gives the FAA data on integration while Reliable continues working toward commercial certification.

The new capital is headed toward aircraft, production facilities and engineers. Seattle gets a Radar Center of Excellence; the planned San Diego office brings another pool of aviation and defense expertise. Reliable says its team now exceeds 250 people.

For a cargo operator, the proposition reaches beyond what the flight controls can do. Navigation, detect-and-avoid, remote command and airport procedures have to arrive as a service the operator can actually use. Reliable is putting more people around that handoff, from the radar lab to the person talking with the tower.

VECKTA

Asking five suppliers for a solar quote can leave a finance team with five different projects to evaluate. VECKTA wants the buyer to settle what belongs on the property before anyone starts competing for the contract.

The San Diego company announced an $11M Series A on October 20, led by Origin Ventures. The investment backs a planning and procurement business built around an awkward purchasing problem: a competitive price means little when the proposals disagree about what you are buying.

CEO Gareth Evans co-founded VECKTA with Calvin Fine, Head of Market Ops; Dan Roberts, Head of Sales; and Felipe Sarubbi, Head of Product. The team brings site analysis and supplier selection into the same process, giving the customer a business case before the sales pitches arrive.

Start with a property address. VECKTA evaluates local conditions and recommends a system configuration, then helps issue a standardized request for proposals to vetted suppliers. Finance can compare offers against a common scope instead of trying to reverse-engineer the assumptions hiding inside each bid.

Across a portfolio, that also means deciding which buildings should wait. An attractive rooftop is useful only when the economics, equipment and project terms make sense together. Keeping marginal sites under review gives the buyer somewhere to put a sensible idea that is not ready for a purchase order.

VECKTA says customers contracted for systems at 128 sites this year, paying 20%-45% less than bids they had sourced themselves. Those are company-reported bid comparisons, rather than measurements of installed-system returns. Procurement itself can change the investment case.

Tech Square Ventures, VoLo Earth Ventures, Cove Fund, Catalyst by Wellstar, Engage, Emerald Bay Ventures, She's Independent Investments and Mobilize AI x Energy Ventures also participated. The capital will expand VECKTA's U.S. customer team, product, partnerships and supplier network, with broader asset lifecycle support planned.

Origin partner Scott Stern previously traded electricity products, according to his firm biography. That background fits a company selling a clearer purchasing decision in a market where the same technology meets different economics at each address.

For Gareth Evans and the team, expansion brings more people who need help turning an energy assessment into a contract their finance department can defend. The next supplier entering the process should already know which question the buyer is asking.

Peak Energy

A battery can spend part of its working life keeping itself comfortable. Peak Energy wants more of that electricity left for the customer.

The Burlingame company builds passively cooled sodium-ion storage for utilities, independent power producers and AI data centers. Removing active cooling takes equipment, maintenance and an electricity bill out of the design. For somebody buying an asset expected to work for decades, the supporting machinery belongs in the purchase decision.

Peak announced the closing of its oversubscribed $108M Series B on October 21. GM Ventures led, with Maverick Silicon, Infinite Capital, Eclipse, Xora Innovation and LG Technology Ventures participating. The money supports domestic manufacturing, customer commitments and expansion into data-center power infrastructure.

Co-founder and CEO Landon Mossburg works alongside co-founders Cameron Dales, President and CCO, and Liam O'Connor, COO. CTO Paul Durkee leads the system's development. Their assignment stretches from what happens inside a cell to what a utility has to service outside it.

GM brings more than a check. Under the partnership announced in June, it develops sodium-ion cells in Michigan and retains exclusive manufacturing rights. Peak incorporates those cells into its storage systems. That division gives an automaker's battery expertise another customer: the electrical grid.

Peak says it has more than $1B in commercial commitments. Jupiter Power's agreement helps explain the shape of that demand: approximately 720 MWh for 2027, plus an option for a further 4 GWh across 2028-2030. Those later volumes depend on the option. A reservation is useful commercial evidence without becoming delivered equipment or booked revenue.

Sacramento now sits inside that promise. Peak's planned 183,000-square-foot factory is designed for up to 4 GWh of systems annually, with production expected in Q1 2027. The company reports potential energy-cost savings of up to 20%; customers will experience the economics through operating sites, service requirements and power available when they need it.

For data-center buyers, storage arrives with a less glamorous question than the next chip launch: how much of the electricity purchased reaches useful work? That overhead belongs in an AI infrastructure budget. Peak's financing puts more resources behind answering it with manufactured systems, while Jupiter's 2027 delivery window draws closer to the Sacramento production calendar.

IMIO

IMIO ships its microbes asleep. The farmer decides when they wake up.

That is an unusually practical way to sell biology. Growers already have planting windows, equipment and a fertilizer bill. A living product has to survive the trip into that operation before it can justify taking up any of the budget.

Co-founder and CEO Charles Smith built IMIO with co-founder Victoria Holden around putting microbes to work. Its Re-Gen products use fungi, bacteria and yeast to help break down crop residue and make nutrients available for the next crop. Neal Connors serves as fractional CTO, bringing decades of bioprocess experience to the manufacturing work.

The company announced $12M in funding on October 20, a seed round led by Sofinnova Partners. Chipotle's Cultivate Next venture fund, Starshot Capital, Grow Together Salinas, Earth Foundry and Serra Ventures also participated. The money supports a new headquarters, lab and biomanufacturing operations in Chicago after IMIO's move from Vermont.

The process gives that investment somewhere specific to go. Each microbial strain is grown and tested individually, combined in precise ratios and freeze-dried. Growers add water before application, then put the biology directly onto the residue it is meant to break down.

Corn stalks can be both a field-preparation problem and a store of nutrients. IMIO is selling the work of making that second property useful sooner, so a grower can get more from material already sitting on the farm.

IMIO reports an average corn yield increase of 11.9 bushels per acre across university and farm trials covering more than 10,000 Midwest acres in 2024 and 2025. Those are company-reported results, with outcomes varying by field conditions and management. A buyer still needs to understand the response on their own acres and the costs of getting it.

That makes handling part of the commercial argument. A concentrated powder that stays shelf-stable until activation has to fit a farm's storage and application routines. IMIO's own guidance also asks growers to confirm tank-mix compatibility and follow the product label, which is where a promising input meets an actual day's work.

Chicago puts the expanding production operation nearer the growers IMIO wants to serve. As more product leaves that operation, the relationship carries into the next buying decision: what the grower saw in the residue, what changed in the field and how much fertilizer the next crop will need.

Rosie

Mitra Raman and Karthika Shankar took the first shifts themselves. Before Rosie could send somebody into a family's home, its founders went in to do the childcare and household work.

That is a useful place to learn what a customer means by help. The calendar can say babysitting while the parent would rather play with the child and have somebody else handle the kitchen. A household rarely organizes its needs around the categories on an app.

Mitra Raman, co-founder and CEO, and Karthika Shankar, co-founder and COO, built Rosie around that flexibility. The Seattle company announced $7M across two oversubscribed financing rounds led by Bling Capital and Reach Capital. The money supports expansion into Austin and San Francisco.

Graham & Walker, Marketplace Capital, Ingeborg Investments, Precursor Ventures, Mother Ventures and Yonder Ventures participated, alongside angels including DoorDash co-founder Evan Moore. The $7M covers both rounds; the announcement leaves their individual amounts and formal round labels undisclosed.

Rosie launched its current service in January 2026. The company reports more than 17,000 hours of support for over 250 member families and seven-figure ARR within eight months. Those are company-reported figures, with enough activity behind them to make the operating choices worth examining.

Families buy monthly hours they can use for childcare, laundry, meal prep, organizing or getting the house back in order. A regular Pro learns the household, with a Backup Bench available when plans change. Familiarity becomes part of what the membership is selling.

AI handles matching, scheduling, preferences and payments behind the scenes. The human arriving still has to know how to do the work. Rosie's Pro onboarding includes background and reference checks, a skills assessment and paid shadow sessions. Saving the parent coordination time creates a reason to keep the evaluation deliberate.

Karthika Shankar brings Amazon recruiting experience to that side of the business. Mitra Raman brings an engineering background to the software around it. Their early shifts supplied a shared view of what both sides have to get right when somebody opens the front door.

Rosie is recruiting an Austin general manager as it prepares to take the service beyond Seattle. That person will have a very local assignment: build a group of Pros whom families want to invite back, then make the next booking easier because the relationship already exists.