HappyRobot’s $150 Million Round, and the Calls It Actually Takes

HappyRobot raised $150 million in a Series C announced August 4, 2026, at a $1.2 billion post-money valuation. Prysm Capital led the round and Eurazeo co-led it. Andreessen Horowitz, Base10 Partners, and Y Combinator participated again, joined by strategics including Koch Disruptive Technologies, Kfund, Orange, T.Capital, Bankinter, Endeavor Catalyst, and Wave-X, according to the company’s announcement and to Fortune and SiliconANGLE.

The company says the round brings total funding to around $200 million. That total is the company’s figure. The more useful question for a brokerage is narrower than the valuation: which calls, emails, and documents the software is actually taking, and which results have been checked by anyone other than the vendor.

HappyRobot, which Reuters identifies as Happyrobot Inc., is based in San Francisco and was founded in 2022. Pablo Palafox is the chief executive. Fortune identified him as such in its August 4 report. This desk has not interviewed him.

Three rounds, three different sources

The Series C is the latest number, not the first. In December 2024 the company raised a $15.6 million Series A led by Andreessen Horowitz. GlobeNewswire’s later Series B release and Fortune’s August 2026 story both place Y Combinator and Baobab Ventures in the company’s earlier financing. This desk is using those accounts for the Series A rather than reconstructing a cap table.

On September 3, 2025, HappyRobot raised $44 million in a Series B led by Base10 Partners. Reuters reported the round. Existing investors Andreessen Horowitz and Y Combinator participated, and the company release also names Array Ventures among prior backers and Samsara Ventures, Tokio Marine, WaVe-X, and World Innovation Lab among new ones. Reuters said the company did not disclose a valuation and that an unnamed source put the Series B valuation around $500 million. That valuation is a report, not a company disclosure. Treat it as “reportedly.”

Reuters described the product in one line that still holds: software that automates communications, including rate negotiation and appointment booking, for freight operators. The Series B release says the workers communicate over phone, email, and chat, parse documents, browse sites, and log data, and that an “AI Auditor” reviews what those workers did. Tasks named in that release include negotiating rates, booking appointments, collecting payments, and recruiting staff. Recruiting is a long way from a check call. The brokerage use is the part this story can actually describe.

What the company says it does for brokers

On its logistics page, HappyRobot lists workflows for brokers and 3PLs. Read them as the company’s product claims.

It says agents can take inbound load tenders, screen for fraud, and answer requests. It says they counter offers with market data and log offered and accepted rates by lane. It says they nurture a carrier network and try to wake up dormant leads. It says they schedule docks and appointments, confirm those appointments, chase missed pickups, and cover after-hours calls. It says they place check calls at milestones, flag loads at risk, and push ETAs into downstream systems. It says they answer email quotes and inbound calls on booking, load status, and lumper requests. It says they collect proofs of delivery, bills of lading, and invoices, audit an invoice against a confirmed rate, take in claims, and call on overdue payments.

The company also says it is SOC 2 and GDPR compliant. That is a company claim.

None of those sentences means HappyRobot runs a brokerage. A rate confirmation still has a broker of record. An exception still needs a person who can own it. The company’s own Series C post says it is expanding past logistics into insurance, energy, utilities, telecommunications, and airlines, and that it works with more than 150 enterprise customers, naming DHL, Kuehne + Nagel, Naturgy, Repsol, and Uber. Several of those names are not brokerages. The freight desk is where the product started, not the whole company it is becoming.

A named brokerage deployment, with the result labeled

MODE Global, a Dallas non-asset logistics company, published its own account on February 9, 2026. It said the work started with inbound carrier sales: vetting a carrier, pitching a lane, negotiating. Outbound carrier sales and track-and-trace were described as later phases. MODE’s release says first response times were cut to zero and that answer rates hit 100 percent. Those are the customer’s reported results. Freight Broker News has not audited them, and they should not be read as a benchmark for every brokerage that takes a demo.

Fortune reported that revenue grew more than fivefold after the Series B and that net dollar retention topped 150 percent. Both figures come from the founder. The company’s own Series C post says it grew fivefold “since raising our Series B late last year,” which does not match a September 2025 close. This story uses Fortune’s attribution and drops the “late last year” phrasing.

What a broker should ask anyway

The pattern across Augment and Envoy AI is the same argument in different clothes: the TMS records the load, and a person still does the work between the screens. HappyRobot’s bet is that a large share of that in-between work is a conversation that can be scripted, logged, and handed back when it stops matching the script.

That is a narrower claim than “AI runs the brokerage.” It is also the claim a buyer can test. Which calls are in scope on day one. What the agent is allowed to offer. Where the human has to approve a rate. What gets written back to the TMS the desk already runs, in the customer’s own words rather than a slide. MODE’s write-up is useful because it names a starting workflow instead of a platform. The answer-rate figure is useful only as MODE’s figure.

Our analysis of assistants versus workflow tools puts this round next to the other approaches, and the stack guide keeps the phone agent in its lane: one layer, not the whole desk.