Trajectory, founded by ex-DeepMind, Apple, OpenAI and Meta researchers to build continual learning models, raised a $15M seed at a $115M post-money valuation

Trajectory is betting the rapid iteration cycle that supercharged vibe-coding can help all kinds of companies build AI products that learn continuously.

Wired 2026-05-27 Maxwell Zeff

Context & Ripple Effects

Trajectory enters a funding cycle already marked by large bets on former DeepMind talent: Ineffable Intelligence raised a substantially larger seed to pursue “superlearners.” The shared emphasis is on AI systems that improve their capabilities, rather than only delivering a fixed model experience.

The company’s focus also connects to the rise of vibe-coding and AI workflow products, where rapid user feedback and iteration are becoming part of how products are built and refined.

First-order effects

The $15M seed gives Trajectory resources to recruit and develop its continual-learning models, with its $115M post-money valuation setting an early benchmark for the company.
Trajectory’s stated product direction makes user interactions central to its model-improvement approach, differentiating it from AI products positioned primarily as static assistants or one-off automation tools.

Second-order effects

Startups pursuing adaptive agents, workflow automation, and AI app-building will face sharper competition for research talent, early customers, and investor attention around the claim that products can learn from use.
Companies adopting such systems will need to judge whether rapid iteration produces reliably better behavior, rather than simply faster changes to an AI product.

Third-order effects

If continual learning proves dependable in deployed products, competitive advantage may shift from access to a base model toward the quality of feedback loops, product instrumentation, and the ability to safely incorporate user interactions.
The pattern would make evaluation and controls around production-time model changes more important, since the product behavior could evolve after initial deployment.

The trend: Trajectory is one data point in the move from general-purpose AI models toward products whose capabilities are designed to improve through ongoing real-world use.

Discussion

@natolambert Nathan Lambert on x
The most likely way continual learning manifests in the coming few years is through products used directly for knowledge work. Sort of how cursor can continually train their models with real-world data and RL, Claude, Copilot, and co will see if they can for knowledge work. I
@scobleizer Robert Scoble on x
New trend: continual learning. I first showed you this with Levangie Labs from @blevlabs. You “grow” an agent by teaching it. It gets better over time. It works because it has a better memory, so can try things, and learn, and remembers what it learned, so it gets better over
@baseten @baseten on x
Excited to support the @trajectorylabs team as their inference partner. Powering live, continuous post-training for frontier-scale models takes serious infrastructure, and we're thrilled to build the future of continual learning together.
@togethercompute @togethercompute on x
The best research labs are building what comes after static models. Congrats to @trajectorylabs on the launch! Excited to have them training on the AI Native Cloud as they push the frontier on Continual Learning!
@neilkale Neil Kale on x
Super excited to announce that I've joined @trajectorylabs! We're pushing the frontier of RL research to build the platform for continual learning. Our team is building paradigms to teach, align, and trust models that adapt in real time as users interact with them. We're hiring!
@rronak_ Ronak Malde on x
Today, @MichaelElabd, @QuantumArjun, and I are excited to announce Trajectory. We are a research lab and product company building the platform for Continual Learning. Our platform unlocks the signal already sitting in product usage, so companies can continuously post-train [video…

Chronicles