Covers Monday, Nov 2 (after the morning Daily Brief) through early Tuesday, Nov 3 (Eastern). Several of Monday's rounds went to companies that charge for work done rather than for software seats. Andreessen Horowitz led a $75M Series B for Tutor Intelligence, which rents out factory and warehouse robots by the hour. Sequoia valued Trajectory, a post-training platform about a year old, at $300M. Churchill Capital Corp XI shareholders approved the Agility Robotics merger, but the more telling number in the filing is the 61.5% of public shares redeemed. Nine financings cleared our checks, including one Oct 30 round the last Daily Brief missed (marked "catch-up"). Each is confirmed by the company, its lead investor or both. Figures companies report about themselves are labeled company-reported.

Tutor Intelligence, a Watertown, Mass., robotics company, announced a $75 million Series B led by Andreessen Horowitz. Union Square Ventures and Neo, which led its Series A and seed rounds, also invested. Total funding is now $117 million (Tutor release via Robotics & Automation News · a16z · Tutor · The Robot Report). No valuation was disclosed. Tutor builds its own robots: Cassie, a single-arm mobile robot for case picking and palletizing, and Sonny, a two-armed robot for picking individual items. It puts them to work before they are fully autonomous. Remote human operators, which the company calls "tutors," provide input as the robots work, and the data from that work trains Tutor's robot foundation models. Customers pay by the hour or by the pick, starting at $14 an hour, with no upfront purchase and no long-term contract. Tutor handles deployment and maintenance. The company says its robots now work in almost four times as many facilities as a year ago, and that it is on track for more than 1,000 robots under contract by the end of 2027 (company-reported). a16z adds that nearly a third of customers that received Tutor robots in the past 12 months were existing customers adding more. Why it matters for investors: Hourly pricing turns a robot sale into a utility bill, which removes the large capital budget that has kept most small and midsize plants from automating. The cost moves to Tutor. Because customers make no upfront purchase, Tutor carries the hardware, so each new customer consumes capital before it pays back, and margins depend on how much human help each robot-hour still needs. That makes this round as much a financing of a fleet as of a model. Diligence questions: What share of robot-hours still needs a remote tutor, and how fast is it falling? What is the payback period on a deployed robot at $14 an hour? With no long-term contracts, what does churn look like after the first year, and how will Tutor fund 1,000 robots without heavy dilution?

Historical lens. Tutor's model combines two older ideas. The first is pricing. Xerox unveiled its 914 copier in the fall of 1959 and shipped the first one to a customer in March 1960. Early forecasts called for perhaps 5,000 placements in three years, but by the end of 1962 Xerox had shipped 10,000, according to its own history (Xerox). In April 1963, TIME reported that a 914 cost Xerox about $2,500 to make and rented for an average of $5,000 a year: $95 a month, plus 3.5 cents per copy beyond 2,000 (TIME, 1963). The second is the person inside the machine. Amazon's crowdwork marketplace, Mechanical Turk, launched in 2005 and took its name from an 18th-century chess "automaton" that concealed a human player (Pew Research Center). Wolfgang von Kempelen built that device in Vienna in 1770 (University of Houston, Engines of Our Ingenuity). Our interpretation, not established fact: Tutor bills the way Xerox did and, unlike the Turk, is open about its humans. The figure to watch is how quickly the human share of each robot-hour shrinks.

Trajectory, a startup that lets companies keep retraining AI models on their own production data, announced a $45 million Series A led by Sequoia at a $300 million valuation. NVIDIA, Bessemer and Bain Capital Ventures also invested (Trajectory · TechCrunch). On the same day it opened a self-serve version of its platform. Customers supply tasks and a way to score them, then train, evaluate and deploy their own model in one place. Two features are in research preview. Continual Grading learns a team's standards from a few examples and applies them across production data. Group Size 1 trains on real production interactions, so a model can keep improving after launch. Trajectory says teams at Harvey, Rogo, Mercor and NVIDIA already train on it. It also says it took an NVIDIA Nemotron Lightning model from 0% to an 8.3% all-pass rate on Harvey's LAB legal benchmark, matching Anthropic's Opus 5.5 at what it puts at 100 times lower cost (company-reported). TechCrunch reports that the company was founded about a year ago and has 18 employees. Why it matters for investors: A $300 million price for a company this young rests on the team and on a thesis: as token bills rise, companies will want smaller models they own and tune, not rented frontier models. That is a bet against the big labs bundling post-training into their own APIs, and the risk is that they do. NVIDIA is an investor, a customer and the supplier of the models in Trajectory's showcase result, which helps distribution but concentrates exposure. Diligence questions: How much of today's usage is paid rather than design-partner work? Do results like the LAB score hold up on customers' own tasks? What stops a lab or a cloud provider from shipping the same loop as a feature?

Churchill Capital Corp XI shareholders approved the merger with Agility Robotics and every related proposal on Monday (Agility release · Churchill XI 8-K · The Robot Report). The company describes it as more than 84% of votes cast in favor. The 8-K shows 21,069,042 shares for, 1,334,087 against and 2,623,822 abstaining, with about 44.93% of the voting power represented. The filing also gives the final redemption count. Holders of 25,451,804 Class A shares, about 61.5% of those outstanding at the deadline, redeemed, and about $261.8 million (roughly $10.29 a share) will leave the trust. Expected gross proceeds stay at about $465.1 million before expenses: $164.1 million from the trust, $201.0 million from the Foxconn-led PIPE and $100 million Agility has already received. Churchill XI also moved its incorporation from the Cayman Islands to Delaware on Monday and renamed itself Agility Robotics, Inc. Closing is expected Nov. 4, with trading as AGLT (warrants AGLTW) on Nov. 5. Why it matters for investors: Read the 84% with care. The sponsor holds all 13.8 million Class B founder shares, for which it paid $25,000 in total (S-4/A), and those shares alone made up about 55% of the shares represented at the meeting (our arithmetic). Only about 27% of Class A shares outstanding at the record date were represented at all (our arithmetic). The approval rate therefore largely reflects the sponsor's own shares. The 61.5% redemption rate is the public holders' answer on price. Diligence questions: Where does AGLT trade against the $10 PIPE price after the 180-day lockup on existing Agility holders, which we noted yesterday? How much of the remaining public float is held by redemption-arbitrage funds likely to sell early? Does $465 million gross, before expenses, fund the path to the early-access Digit 5 deliveries planned for the first half of 2027?

Agility Robotics' roughly $465M is not listed. It is a de-SPAC financing we covered on Nov 2, and the vote did not change the figure.

Round notes

Not investment advice. The AI Cap Table is published for information only and is not investment, legal, or tax advice. Nothing in this newsletter is an offer or solicitation to buy or sell any security. Funding data comes from public sources we cite and verify, but errors are possible; do your own due diligence and consult a qualified adviser before making any investment decision.