
What happened
Paris-based Remanence raised a $30 million seed round co-led by EQT Ventures and FirstMark, with Zero Prime, Emblem, Drysdale, Global Founders Capital and Yellow also participating.
Why it matters
Remanence automates the sandbox and grading setup that EQT Ventures says currently takes months of manual work per customer, letting enterprises retrain open-weight models on their own execution logs.
WHO IT HITSEnterprise AI teams running agents in production — especially in legal, customer support and enterprise software — could get a faster route to models tuned on their own execution data, while keeping model weights in-house.
Summaries like this, every morning.
Remanence is targeting a gap that appears once companies move AI agents from pilots into production. Agents call tools, fail, and get corrected every day, generating a large record of execution data that has mostly been discarded — so the next run starts from the same general-purpose model and repeats the same mistakes.
EQT Ventures frames the bottleneck as practical: improving a model with reinforcement learning requires a sandbox that mirrors the company's real systems and a way to grade whether the agent did the job correctly, and those environments are currently built by hand over several months for each customer. Remanence's pitch is to automate that pipeline, from analyzing logs to generating environments, running long-horizon reinforcement learning and inference, either fully managed or inside the customer's own cloud.
The startup's team comes from H Company, where the founders led post-training of open-weight models, and it is already engaged with customers in legal, customer support and enterprise software in Europe and the US. The new funding is earmarked for a dedicated GPU cluster in Europe and hiring in Paris, London and San Francisco. A likely implication, though not stated, is that as stronger open-weight models appear, customers could swap models into the same environments and retrain.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Oxide Computer Co., which raised $445 million in October to expand rack production, built its business on the…
OpenAI now requires gateways and routers that provide third-party access to its AI models to be reviewed and a…

Anthropic shares traded as low as $83.15 on Friday, down more than 14% on their second day of trading, leaving…

Running the official harness, Claude Opus 5.5 wrote an ARC model with no gradient descent and no forward calls…

A Zenn article by Hiromi lays out Continuous Reasoning Acquisition (CRA), a design for capturing veteran judgm…

Major US tech companies are refocusing generative AI competition from answer quality toward agent capability…
