Trajectory, Founded by Ex-Google and Apple Researchers, Raises Funding From Sequoia in Back-to-Back Round

As closed-source models have gotten more expensive, businesses have found workarounds by customizing open-source AI for their specific needs. Some are also trying to improve the software they use that allows models to use tools and take actions, otherwise known as a harness.

Now, startups that help businesses with both tasks are seeing more investor interest. One such startup, San Francisco-based Trajectory, recently raised $40 million in new funding at a $300 million valuation, including the investment, just two months after announcing its last round at a $115 million post-investment valuation, according to a person with knowledge of the funding.

Sequoia Capital led the round, which also saw participation from Nvidia and Bessemer Venture Partners, the person said.

Trajectory is far from the only startup trying to help businesses customize open-source models. Other companies like Applied Compute and Thinking Machines Lab have raised billions of dollars to do the same thing.

Trajectory has said that it’s different from these other startups, though, because it’s trying to come up with a more automated and user-friendly way for businesses to customize open-source AI by themselves without requiring teams of consultants or lots of existing technical knowledge.

It’s doing so through new techniques like self-distillation policy optimization, which essentially helps models learn from feedback they get out in the wild. For instance, an AI-powered customer support agent might be helping a person get a refund from a pizza restaurant that delivered them the wrong item. If the agent gets stuck, it might pass along the support ticket to a human customer support agent, who might realize that they need to check a certain database to ensure that the reason for the refund is valid first.

That feedback would then get passed back to the AI agent, which then asks itself, “given what I now know, what should I have done differently?” Its updated decision, given that feedback, then gets fed back into the model to ensure it doesn’t make the same mistake in the future.

Trajectory not only wants to help businesses use open-source AI better, but closed-source AI too. It does so by helping businesses tweak the harnesses they’re using with those closed-source AI models to get them to run for cheaper or with more accuracy.

The 12-person company’s customers include AI sales startup Clay, AI customer support startup Decagon and AI legal firm Harvey, according to its website.

Trajectory was founded in May by former Google DeepMind researchers Ronak Malde and Michael Elabd and former Apple researcher Arjun Karanam. In the spring it raised $15 million in seed funding from investors like Conviction Partners, Bessemer and Radical Ventures.

Here’s what else is going on…

Big Number

Intel is raising $15 billion in a public offering of its stock, taking advantage of the enormous improvement in investor sentiment it has enjoyed in the past year to raise money. Intel stock is up 175% over the past 12 months. Intel shares fell 4% Monday. The iconic chip maker, which has lost its strong market position over the past few years to rivals such as TSMC and Nvidia, has shown signs of a nascent turnaround in recent months under a new CEO, Lip-Bu Tan.

People on the Move

Caitlin Kalinowski, who previously led OpenAI’s robotics effort, has joined Anthropic as a member of technical staff working on research, according to her LinkedIn. Kalinowski’s move is the latest sign of Anthropic’s interest in robotics. Anthropic last month released new research on using Claude to control robots, and The Information reported last month that Anthropic had previously held acquisition talks with robotics startup Physical Intelligence.

Earnings Report

Enterprise software firm Monday.com reported a 22% lift in second quarter revenue, but projected a slowdown in third quarter revenue growth to between 16% and 17%, partly because of the impact of a big layoff it undertook last month. Monday shares fell 9%.

Deals and Debuts

See The Information’s Generative AI Database for an exclusive list of private companies and their investors.

A group of private equity giants has signed a preliminary agreement with Nvidia to raise $500 billion to help finance AI infrastructure, Nvidia said on Monday. The firms are Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR. The partnerships haven’t yet been finalized, per an Nvidia press release. Nvidia also has the option to backstop up to 25% of the financing for any project that’s part of the new partnership, The Information reported on Monday.

OpenAI bought about $7 billion in shares from employees in a recently closed tender, according to a person familiar with the deal. The buyback happened at the $852 billion post-money valuation reached in a March funding round.

Sony Group and Taiwan Semiconductor Manufacturing Co. are planning to invest about 1 trillion yen ($6.3 billion) in a joint venture that will start mass producing next-generation image sensors in Japan as early as 2029, the Nikkei newspaper reported.

Volta Infrastructure, a company that develops and runs large-scale AI data centers and rents out computing power, raised $300 million at a $2.4 billion valuation in a combined seed and Series A funding round led by Azora, Andreessen Horowitz and Altimeter Capital.

Corma, a startup that offers an AI system trained specifically to defend companies against AI-powered cyberattacks, raised $60 million in a seed funding round led by Sequoia Capital.

VibeIQ, a startup whose AI software helps apparel, footwear and consumer-goods brands decide what products to make, raised $22.5 million in a funding round led by Volition Capital.

Atlas Motion, a company that designs high-performance electric motors and actuators for defense equipment, raised $11.5 million led by Greycroft.

Blaze.tech, a company that lets healthcare organizations build their own custom, privacy-compliant software without writing code, raised an additional $5 million in pre-seed funding from Friale.

Anthropic, Australia’s Macquarie Asset Management and the Singaporean sovereign wealth fund GIC said Monday they were launching a new entity to develop, operate and lease AI data centers for Anthropic to meet demand for its Claude models.

OpenAI on Monday announced an expansion of Daybreak, its cyber defense service which it launched earlier this year, and GPT-5.6-Cyber, a new model for advanced, authorized cybersecurity work.

Anthropic will watermark text generated by its models, including Claude, to comply with European regulations, the company announced.

Spotify on Tuesday said that it will begin labeling AI-generated artists with “AI Persona” profile tags and ban their music from its editorial and algorithmic recommendations.

LTX, a world model developer, released LTX-2.5, an open-weight world model, on Tuesday.

Thank you for reading the AI Agenda Newsletter! I’d love your feedback, ideas and tips: [email protected].

If you think someone else might enjoy this newsletter, please pass it forward or they can sign up here.