The Work That Makes It Feel Normal

ASI

The FAA is trying to settle more traffic conflicts while the aircraft are still on the ground. ASI's funding lands inside that shift in when a decision gets made.

Weather, runway capacity and airline schedules already exist as data. The expensive part is getting people managing the same sky to work from a useful picture of what those conditions will do together. A route can look sensible until everyone else needs it too.

Boston-based ASI announced a $325M Series C at a $3B valuation on October 20. Spark Capital led, with affiliates of Citadel Securities, Meritech Capital, Craft Ventures, Energy Impact Partners and Andreessen Horowitz participating. The company previously operated as Air Space Intelligence.

Co-founders Phillip Buckendorf, CEO, Kris Dorosz, Chief Research Officer, and Lucas Kukielka have been building the business since 2018. CTO Jon Varsanik previously led engineering teams at Google, including YouTube infrastructure. That combination now has to carry predictive software into institutions whose operations keep moving during the upgrade.

The FAA selected ASI for Flow Management Data and Services and SMART, which coordinates schedules and trajectories before departure. The agency describes a common view of demand, weather and available capacity, giving traffic managers and airlines more opportunity to adjust together.

ASI puts the FAA award at $875M over 12 years. Its recently announced Army agreement is a 5-year, $99M IDIQ contract, a vehicle for orders rather than a statement that $99M has already become revenue. Those commitments sit alongside the financing, with different clocks and different obligations.

The company is also working with National Grid on planning and operational decision support. Outage mapping, resilience and distributed energy resources bring different constraints from aircraft routing, even when the underlying habit is familiar: model the consequences before committing the assets.

The new money will support deployments in energy, logistics and tactical edge environments, plus research and engineering hiring in Boston. Spark and a16z also participated in ASI's $34M Series B in 2023, so this expansion includes investors who have watched the work develop across several years.

For an airline dispatcher and an airspace manager, an earlier warning becomes valuable when there is still room to change the plan together. ASI is taking more responsibility for the view each side brings to that conversation before the next departure.

Deeper Analysis 👇 https://devcuration.com/articles/asi-325m-series-c-critical-infrastructure-ai

Pigment

Salary planning for 95,000+ people gives an AI planning company a very human assignment. At L’Oréal, the numbers in the model belong to employees, managers and decisions that eventually reach somebody’s paycheck.

Pigment says L’Oréal selected its platform to replace its global salary-planning tool. That puts the company’s ambition in a more useful setting than another immaculate AI demo: an organization where people need to agree on what the numbers mean and who gets to change them.

Co-founders and co-CEOs Eléonore Crespo and Romain Niccoli have announced a $250M Series E at a $2.3B valuation. Scaleup Europe Fund, managed by EQT, led, with WIL, Greenoaks, ICONIQ, IVP and Meritech participating. NVIDIA’s investment is planned, according to Pigment’s announcement.

Founded in Paris in 2019, Pigment connects finance, sales, workforce and supply-chain planning. A hiring assumption can travel into costs and capacity without each department rebuilding its own version of the business. For the person responsible for explaining the forecast, that shared context is part of the purchase.

Its $145M Series D in 2024 helped finance product development and North American expansion. The next assignment brings AI further into the planning cycle already sold to enterprises.

Pigment reports passing $100M in ARR this year and doubling ARR year over year. It also says agents create 85% of new metrics and 84% of accounts use AI weekly. Those figures describe adoption; a finance team still has to decide which assumptions deserve to become a plan.

Underneath the agents sits Graphite, Pigment’s proprietary, patent-pending compute engine. The company describes a governed model where people and agents work concurrently, using shared definitions and access controls. Automating more of the work makes that foundation increasingly consequential.

The Modeler Agent builds models and interfaces from plain-language instructions. The upcoming Planner Agent is intended to run planning cycles end to end. Pigment is putting the new capital into R&D and engineering as it extends both the engine and the work its agents can carry.

EQT is backing a Paris-built supplier with global enterprise customers and a route into its own network of businesses. Inside those businesses, somebody will still own the compensation assumptions, the hiring limits and the decision to revise them. Pigment’s expanding assignment is to keep that authority legible as more of the planning happens through agents.

Deeper Analysis 👇

https://devcuration.com/articles/pigment-raises-250m-series-e-autonomous-planning

Otter.ai

The sales budget is getting its own financier at Otter.ai. General Catalyst's Customer Value Fund is backing the work of winning customers without buying another piece of the company.

Otter announced $320M in growth financing on October 20. CEO and co-founder Sam Liang told Reuters the money is repayable with interest as sales and marketing generate revenue. Existing shareholders keep their stakes. The price of growth moves into the revenue the business has to earn.

Sam Liang and co-founder and CTO Yun Fu have a substantial audience to work with. Otter reports more than 40M users and teams at over 86% of the Fortune 500. That is a distribution advantage, with an important commercial distance still to travel: somebody using the tool inside a company is different from that company buying it for a department.

An employee can appreciate a clean transcript immediately, from a single meeting. A sales leader buying for a team needs the conversation to keep working after the call, carrying the prospect's commitments into the CRM and bringing useful context into the next meeting.

Otter is building toward that larger purchase. Its product now includes pre-call briefs, live coaching, searchable conversation history and automatic follow-ups. The April launch of its Conversational Knowledge Engine connected that history to other enterprise tools through AI Chat and MCP integrations.

The financing has a similarly specific job. General Catalyst describes CVF as pre-funding customer acquisition and taking a capped return from the customer value created. That is the fund's general model; Otter's exact repayment terms have not been disclosed. The structure gives the founders another way to fund selling while preserving ownership for the business they are still building.

There is history behind the choice. Otter announced a $50M Series B led by Spectrum Equity in 2021, including a $10M convertible note. It reported passing $100M ARR in March 2025. Neither figure supplies an updated revenue number or a clean lifetime equity total for this deal.

The new money is intended for enterprise growth, international expansion and product work. Selling into an organization means finding the buyer behind the people already taking notes, then giving that buyer enough continuing value to keep the software in the budget. Otter's sales team now has more financing behind those conversations, and a much larger product to put on the table.

Deeper Analysis 👇 https://devcuration.com/articles/otter-ai-320m-growth-financing-general-catalyst

Arda Industries

A design change has a second price: somebody has to teach the factory how to build it. Arda Industries is putting AI into the manufacturing engineer's side of that bill.

That engineer is already keeping today's line moving. Tomorrow's revision brings another assembly sequence, another set of work instructions, another decision about where people and tools belong. Faster product development keeps handing the factory more homework. The revision still has to reach the people building it.

Bob McGrew, co-founder and CEO, has taken that problem from OpenAI research and Palantir product engineering into a company built around production. Arda emerged from stealth on October 20 with $232M in total funding, including a $160M Series A at a $1.5B valuation.

Altimeter, Accel and Khosla Ventures co-led the Series A, joined by XYZ Venture Capital, 8VC, SVAngel and Felicis. The earlier $72M Seed was led by Accel and Founders Fund, with Khosla Ventures and XYZ participating. Accel is backing the team through both stages of the work.

Anduril was Arda's first design partner and customer. Arda now runs within ArsenalOS, bringing AI-assisted manufacturing planning into an operating defense business. That relationship gives the software a demanding place to learn what a useful production plan actually requires.

Arda reports that basic work instructions at Anduril went from 10 hours to 15 minutes; more complex engineering workflows went from up to 30 hours to roughly 2.5 hours. Those are reported task times. Turning that saved effort into a dependable production schedule still involves the people running the line.

Co-founder and Chief AI Officer Augustus Odena brings Adept and Google Brain experience. Co-founder and CTO Alex Mark worked on Palantir's defense engineering and IRS Direct File. Jakob Frick is a co-founder and advisor. The mix puts AI research beside experience delivering software inside institutions with actual operating obligations.

Manufacturing engineers review and refine Arda's plans. The company starts with design files and carries the work into assembly steps, stations and instructions. Keeping that chain current when a product changes gives an engineer a better place to spend judgment than rebuilding yesterday's documentation.

The funding expands research, engineering and deployments, with robotic execution in development. Arda's next customer will bring a different assembly, a different revision history and another team that has to keep shipping while the design moves forward.

Deeper Analysis 👇 https://devcuration.com/articles/arda-industries-160m-series-a-manufacturing-ai

Eka Robotics

Teaching a robot by hand puts a person inside the cost of its education. Eka Robotics is trying to change who supplies the practice.

Co-founders Pulkit Agrawal and Tuomas Haarnoja have chosen reinforcement learning in simulation as the route to more capable robotic manipulation. A virtual robot can encounter variations in contact and learn from failed attempts before the hardware has to make the move. That gives the team a way to expand experience without asking somebody to perform every example.

Eka announced a $150M Series A on October 20, led by Andreessen Horowitz and F-Prime. Obvious Ventures and E14 Fund returned after leading the seed, alongside MFV Partners, Bison Ventures, Glasswing Ventures, Samsung Next and unnamed Fortune 500 strategics. The company plans to broaden its model's capabilities and accelerate customer deployments.

Agrawal brings robot-learning research from MIT. Haarnoja co-created Soft Actor-Critic at Berkeley and worked at Google DeepMind. Their commercial choice puts those backgrounds to work on a stubbornly physical problem: getting a machine to adjust while it is touching something.

Eka's Vision-Force-Action model pairs sight with force feedback. Its investor describes the model, sensing, controller and hardware as a joint design. The contact becomes useful information: a slipping object or resistance during insertion can call for a different movement while the job is under way.

For an automation buyer, that could change the scope of a deployment. A different part, material or fit can mean more engineering before a robot earns its place on the line. A skill that survives those variations gives the customer more freedom to change the work around the machine. A production team changing its product could then have one fewer piece of machinery to teach all over again.

Eka reports laboratory picking speeds of 1,500 operations per hour. Keeping productive behavior intact across real working conditions remains part of the deployment effort. The funding announcement offers no named customer roster or full-shift performance record, so those laboratory results need to stay in their laboratory context.

The interesting responsibility now runs back through the simulator. Each customer setting can expose another way an object moves, slips or resists. Agrawal and Haarnoja are building a business around how much of that physical education the next robot can arrive already carrying.

Deeper Analysis 👇 https://devcuration.com/articles/eka-robotics-raises-150m-series-a-for-force-aware-robots

Firefly

The person pressing a panic button needs an answer back. Firefly builds that acknowledgment into the device itself, using a vibration to confirm that the alert has been received.

That small exchange puts a human being inside the physical AI story. Someone needs help, someone else needs to know where to go, and the technology between them has a job bigger than collecting another signal.

James Batt Buckley, co-founder and CEO, and Alex Penn, co-founder and CTO, founded Firefly in 2024. The Manhattan company emerged from stealth on October 20 with a $20M Series A led by Definition Capital, with significant participation from BoxGroup and continued backing from Susa Ventures.

The investors have some history here. This is BoxGroup's third investment in Firefly; Susa led its seed round. Earlier round amounts and a valuation were not disclosed, so the $20M belongs to this financing.

James Batt Buckley helped grow Verkada's alarms business from schools into large enterprises. Alex Penn spent nearly a decade at Apple. Their shared assignment now reaches from the hardware somebody wears to the software a safety team has to understand while the situation is still developing.

Firefly pairs its panic buttons, beacons and gateways with Ember, a software platform that brings safety information and response workflows together. It also connects with existing infrastructure, including cameras and radios. A campus gets to bring equipment it already owns into the conversation.

The details make the proposition tangible. Beacons help locate alerts. Ember offers site mapping, notifications, roll call and configurable workflows. The gateway carries device signals into the software, with backup power and multiple connection paths built into the design.

Firefly says Front Range Community College already uses the system across 3 campuses. That is a named deployment, with safety outcomes still unquantified in the announcement. It gives the company a place to develop the product around real institutional routines.

The capital will support new physical AI products, engineering and machine-learning hires, go-to-market hiring, and expansion across education, healthcare and enterprise organizations.

Selling into those environments means becoming part of how people ask for help and how colleagues respond. As Firefly reaches more buildings, James Batt Buckley and Alex Penn are taking responsibility for keeping that exchange understandable to the person wearing the button.

Deeper Analysis 👇 https://devcuration.com/articles/firefly-raises-20m-series-a-physical-ai-safety

Daytona

A sandbox sounds like a small thing to become responsible for. Then the agents start asking for thousands of them.

Daytona is building a business around the engineer who inherits that responsibility: starting computers, preserving their state, keeping workloads apart and cleaning up when the work ends. The model gets the interesting assignment. Somebody still has to run the place.

On October 20, Daytona announced a $60M Series B led by FirstMark Growth at a $1B valuation, taking total funding to $91M. FirstMark also led its $24M Series A in February. That is a second investment in a company expanding the amount of work it takes off an application team's hands.

CEO Ivan Burazin, CTO Vedran Jukić and chief architect Goran Draganić founded Daytona in 2023, drawing on their experience in development environments. Earlier rounds included a $2M pre-seed and $5M financing. The company moved toward AI runtimes as the same infrastructure problem acquired a customer that could multiply its own workload.

An agent can start a computer, pause it, preserve progress and fork into parallel attempts. Daytona's platform spans Linux, Windows, macOS and GPU environments. For a team building an agent product, those capabilities become decisions about what it will maintain itself and what it will trust a supplier to keep working.

WRITER supplies a concrete example. In Daytona's customer case study, co-founder and CTO Waseem Alshikh describes the maintenance burden of running a homegrown sandbox platform beyond 20,000 sandboxes a day. The case reports 300,000 sandboxes provisioned in a month on Daytona and engineering capacity redirected toward WRITER's own product. Those are customer-attributed results published by the supplier, with different periods behind the daily and monthly figures.

Qualcomm Ventures, Samsung Next and Okta Ventures joined the Series B, alongside investors including Patrick Collison, Tobi Lütke, Aravind Srinivas, Eric Yuan, Olivier Pomel, Adam D'Angelo and Brendan Foody. Returning firms include E2VC, Upfront Ventures and Pace Capital. The attraction is a place inside the customer's daily execution, where another agent task becomes another reason to use the infrastructure.

Daytona plans to broaden its Agent Cloud and is hiring in engineering, go-to-market and operations in San Francisco. Its customers will keep deciding which parts of their computing estate to hand over, one workflow and one support conversation at a time.

Deeper Analysis 👇 https://devcuration.com/articles/daytona-raises-60m-series-b-agent-cloud

CapitalOS

Nir Dremer is selling software companies a way to stay out of the finance staffing business. A platform can put its name on a card while CapitalOS takes on the machinery required to keep that card useful after launch.

That offer now has a $21.9M Series A behind it. LGVP led the October 21 round, joined by WEX Venture Capital, Group11, Great North Ventures and Night Capital, with continued support from existing investors. Nir Dremer, cofounder and CEO, and Matan Goldschmiedt, cofounder and CTO, are building financial infrastructure for B2B platforms that already know how their customers work.

A property management platform knows which building needs the repair. A field service platform knows the job. That context has value when a business needs to pay somebody, allocate an expense or collect an invoice. CapitalOS supplies the financial execution underneath the workflow, with cards, bills, wallets and receivables as the building blocks.

The buyer also gets an answer to an awkward staffing question: who handles underwriting, capital, compliance, fraud and payment support once customers start using the feature? CapitalOS describes specialized agents across those functions, supervised by experts. Its own customer-operations job listing includes human escalation when the existing rules cannot resolve a case. Somebody still owns the difficult call.

Guesty makes the proposition concrete. In CapitalOS's customer account, card purchases flow into property-level ledgers, with receipts and memos captured around the transaction. The property manager gets a financial workflow shaped around the business already on the screen. The software provider gets a reason to remain in the working day after the booking has been managed.

CapitalOS also connects card and bill-payment data with Layer's embedded accounting. That extends the purchase into the books, where a transaction has to remain intelligible to the person closing the month. Better financial software has to survive that handoff.

The company reports more than 20x volume growth over 18 months and says the financing will scale its infrastructure. Volume is the disclosed measure; revenue, valuation and the economics of that growth remain undisclosed.

For a software founder considering embedded finance, the supplier conversation now reaches past the API. CapitalOS is offering to take the financial operations job while the platform keeps the industry relationship, and every new customer arrives with both.

Deeper Analysis 👇 https://devcuration.com/articles/capitalos-raises-21-9m-series-a-embedded-finance

Pascal

A familiar compressor is an unusually useful ally for a company trying to change refrigeration. Pascal has built its materials strategy around the pressure range equipment manufacturers already know how to handle.

That choice connects a Harvard chemistry lab to a beverage cooler. A buyer needs cold drinks, dependable equipment and a sensible electricity bill. The scientific novelty has to earn its place inside that purchase.

Pascal announced $8.5M in additional venture funding on October 21, co-led by Sazze Partners and LB Investment, with KDB Silicon Valley participating. Separate grants add $839K. The financing supports initial deployments with a multinational beverage manufacturer whose name remains undisclosed.

Adam Slavney, co-founder and CEO, Jinyoung Seo, co-founder and CTO, and Jarad Mason, co-founder and CSO, founded Pascal in 2023 from their work at Harvard. Their commercial assignment reaches beyond finding a material that responds to pressure: build a cooling system a manufacturer can actually produce.

Pascal's solid refrigerants work with CO2 as the pressure-applying gas. Two cartridges alternate between heating and cooling, moving heat out of the refrigerated space. The company is designing around familiar industrial components, which gives its material innovation a route into the existing refrigeration supply chain.

That matters because a new refrigerant can bring new demands for compressors, pipes and production equipment. Engine Ventures describes compatibility with off-the-shelf HVAC components as a threshold the founders set for viability. Manufacturing habits belong inside the chemistry decision.

Pascal reports efficiency parity with conventional vapor-compression equipment. The beverage deployment is expected to validate energy and cost improvements; those savings still need a customer operating record. Its broader work on air conditioning, industrial chillers and data centers remains a development program.

The first application gives the team somewhere specific to learn. Cooling a beverage cabinet puts the materials, controls and assembled equipment into the same commercial conversation, with a customer waiting for a useful machine.

Sazze Partners, LB Investment and KDB Silicon Valley are funding that passage into use. The component bill comes with the invention. Pascal's wider HVAC ambitions will travel through the specifications, components and service expectations of the people buying its first coolers.

Deeper Analysis 👇 https://devcuration.com/articles/pascal-raises-8-5m-solid-cooling-pilots

Slingshot Insurance

On both sides of Slingshot Insurance's early customer relationship sits Brock Noland. He co-founded the supplier and runs Vertical Insure, the embedded-insurance business helping design what that supplier builds.

That arrangement tells you where the pressure comes from. Vertical Insure can find the buyer, understand the transaction and design coverage. Getting a carrier's authority and technology to keep up has been another assignment entirely. The customer has helped build a company around the missing support.

Slingshot emerged from stealth on October 22 with $27M in seed financing co-led by Rally Ventures and Skyward Group. Founder and CEO Ethan Genteman previously served as Palomar Holdings' Chief Actuarial Officer. Dan Wick is CTO. Rally also co-founded Slingshot; Justin Kaufenberg and Jeff Hinck join Brock Noland as co-founders and board members. Investor Jim Hays is set to join the board.

The economics get personal at checkout. Slingshot's announcement says Vertical Insure serves 750,000+ customers at a $25 average premium. Those are the partner's figures, and they explain the assignment: an inexpensive policy has little room to pay for administrative friction.

Somebody still has to price the risk, connect the capital, maintain the records and explain how the numbers arrived. A small premium doesn't make those obligations disappear. It makes the cost of meeting them harder to absorb.

Slingshot says its AI-native infrastructure brings those workflows together through APIs, connecting software platforms, managing general agents, distributors and capital providers. The company describes versioned records and named approvals, so a changed figure keeps its history. Its website offers programs with no minimum fees.

For a vertical software founder, that changes the supplier conversation. The application may already be where a family registers for sports or a traveler makes a booking. The insurance partner's operating model helps determine which coverage ideas can afford to reach those customers.

Rally has experience on both sides through Vertical Insure. Skyward brings a specialty-insurance perspective to the financing. Their investment backs an attempt to make smaller programs practical, with a design partner already carrying the problem into the build.

Brock Noland has a seat where that infrastructure gets designed and a business waiting to use it. The next coverage idea will arrive with customers attached, and somebody at Slingshot will have to make the program work at their price.

Deeper Analysis 👇 https://devcuration.com/articles/slingshot-insurance-raises-27m-embedded-insurance

Watchtower Labs

Akhil Iyer went looking for a company to fund and ended up running the missing one. Watchtower Labs grew out of a military planning problem that followed an investor back to his own side of the table.

Upstarts reports that Iyer and Shield Capital founder Raj Shah heard the need from Admiral Samuel Paparo: better logistics planning. When their search for an investment came up short, they built a company. That is a fairly demanding way to answer a customer request.

Watchtower emerged from stealth on October 21 with $35M in Seed financing, co-led by Shield Capital and Conviction, with Founders Fund and IQT participating. Shield incubated the business. Iyer, a former Marine special operations officer and Shield investor, became co-founder and CEO.

Alongside Iyer are co-founder and CTO Suren Aghajanyan and co-founder and Head of Engineering John Morman, both former Microsoft engineering leaders. Their task starts with the arithmetic behind a military operation: fuel, transport, spare parts, munitions, routes and time.

A commander can have several plausible courses of action and resources for only some of them. Then a delivery slips or a route changes, and the staff has to work through the consequences again. The useful answer includes the assumptions that made a recommendation possible.

Watchtower combines AI, simulation and optimization to help planners explore those choices. Conviction describes language models interpreting the request while deterministic engines perform the calculations. Watchtower says its system exposes the constraints and tradeoffs behind options, with people retaining decision authority. That division of labor matters to the buyer: an officer needs to challenge a plan's fuel or transport assumptions without having to reverse-engineer a fluent paragraph from an AI assistant.

There is already a government procurement path underneath the launch. The Defense Innovation Unit confirms Watchtower received a January 2026 prototype contract through its Joint Sustainment Decision Tool program. The program also includes other suppliers, giving military buyers alternatives as the tools develop.

The financing will support hiring, more applications, testing and deployment. As Watchtower reaches into broader operational planning, its engineers will keep meeting the officer whose recommendation depends on the next resupply arriving. Every new application adds another set of assumptions that person needs to see.

Deeper Analysis 👇 https://devcuration.com/articles/watchtower-labs-35m-seed-military-planning

Dayforce / Ribbon AI

The person applying for a night shift may be available after the recruiting desk has gone home. Ribbon built an interview business around giving that person somebody to talk to.

Dayforce has signed an agreement to acquire the Toronto company, bringing conversational recruiting technology toward a platform that already handles HR, pay, time and talent. The October 23 announcement leaves the purchase price and closing date undisclosed. Native integration into Dayforce Agentic Recruiting is planned after the transaction closes.

Arsham Ghahramani, Ribbon's co-founder and CEO, and co-founder and COO Dave Vu have built software that conducts two-way interviews, follows up on answers and turns the conversation into structured information for recruiters. Candidates can participate outside office hours. The company says the final hiring decision stays with the recruiter.

That changes the invitation an employer can make. An applicant who is working, caring for someone or simply unavailable during the recruiting team's day has another way to be heard. The convenience belongs to the candidate, while the employer gets material its people can review when they return.

David Ossip, Dayforce's founder and CEO, is buying into that connection. Dayforce reports that Ribbon has hundreds of customers and more than 1M completed interviews. Those are company-reported activity figures; they give the acquisition an operating history without settling whether a particular employer will make better hires.

There is venture history here, too. Ribbon announced $8.2M in financing in March 2025 from Radical Ventures, with Social Leverage, Cadenza Ventures and GD1 participating, according to transaction counsel Torys. The acquisition now offers the founders a route into Dayforce's customer base. Financial terms give us no basis for calculating anybody's return.

Enterprise recruiting buyers have seen this direction before. Workday completed its Paradox acquisition in October 2025, putting a conversational candidate-experience agent beside its recruiting software. The commercial question is becoming how much of the applicant relationship a company's existing HR supplier can support.

For Dayforce, the work continues into the recruiter's morning. A conversation completed overnight needs to arrive with useful context, a clear assessment and a person ready to act on it. Ribbon's place in the hiring process will extend from the applicant who made time to speak to the employer deciding how carefully to listen.

Deeper Analysis 👇 https://devcuration.com/articles/dayforce-to-acquire-ribbon-ai-agentic-recruiting