{
  "id": 20430817,
  "title": "Remanence Raises $30 Million To Help Companies Train and Own Their AI Models",
  "url": "https://urgent.news/2026/11/17/remanence-raises-30-million-to-help-companies-train-and-own-their-ai",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-17T11:57:43.000Z",
  "source": {
    "name": "Ventureburn",
    "slug": "ventureburn",
    "url": "https://ventureburn.com/remanence-raises-30-million-companies-train-own-ai-models/"
  },
  "original_language": "en",
  "account": "Remanence, a Paris-based startup, has secured $30 million in seed funding to assist companies in training and owning their AI models. The round was led by EQT Ventures and FirstMark, with additional participation from Zero Prime Ventures, Emblem, Drysdale Ventures, Global Founders Capital, and Yellow. Notable investors, such as Olivier Pomel (Datadog), Philippe Corrot (Mirakl), Romain Niccoli (Pigment), and Mehdi Ghissassi (AI71), also contributed to the round. The founders of Remanence, who previously worked at H Company and Google DeepMind, aim to make building and owning AI agents more affordable and accessible. Their platform integrates with existing AI systems, capturing execution traces and converting them into customized training settings. By employing reinforcement learning, Remanence boosts open-weight models, making them less expensive to run while maintaining data relevance. The company is set to expand its operations with new offices in London and San Francisco, but has not yet announced any customers or pilots.",
  "summary": "Each time that a business operates an AI agent, they pay in credits. The more they use an agent, the higher their bill gets. Remanence raises $30 million in seed The post Remanence Raises $30 Million To Help Companies Train and Own Their AI Models appeared first on Ventureburn .",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}