tech.eu via Reddit

Inherent raises $50M for self-improving science AI

funding agents open source ai-funding

Key insights

  • Inherent raised a $50M seed round co-led by Index Ventures and Radical Ventures, with NVIDIA Ventures among additional backers.
  • The founding team draws from DeepMind, Microsoft, and Reka AI and includes a former Biden White House AI policy advisor.
  • Faraday is designed for iterative human-AI collaboration on hard open scientific problems using self-improving agents.

Why this matters

Self-improving AI for scientific discovery is one of the most consequential and hardest-to-validate bets in frontier AI, and a $50M seed signals investors believe the underlying technical architecture is within reach. The team's direct pipeline from DeepMind, combined with policy-connected advisors like Matt Clifford, gives Inherent unusual leverage to shape both the research agenda and the regulatory environment around autonomous scientific agents simultaneously. If Faraday works as described, it creates a direct competitive challenge to DeepMind's own research division and forces larger players to accelerate timelines or move to acquire in the self-improving scientific AI segment.

Summary

London AI lab Inherent emerged from stealth with a $50M seed round to build self-improving AI for scientific discovery, staffed by ex-DeepMind, Microsoft, and Reka AI researchers. Index Ventures and Radical Ventures co-led the raise; NVIDIA Ventures participated. Their product, Faraday, pairs humans with AI agents capable of improving themselves iteratively on hard scientific problems. A former Biden White House AI policy advisor is among the founders, and ex-UK government AI tsar Matt Clifford joined as adviser. Essentially: (Inherent, Index Ventures, Radical Ventures) are betting self-improving AI becomes a genuine scientific collaborator, not just a faster search tool. - Faraday targets long-horizon unsolved scientific problems, not enterprise automation. - $50M seed is among Europe's largest AI stealth-to-launch rounds in 2026. - The founding team's policy credentials signal early positioning for government research contracts. The team's combined research and policy pedigree positions Inherent to shape both the science and the regulatory framework around autonomous scientific agents before most comparable labs have shipped.

Potential risks and opportunities

Risks

  • DeepMind, which has pursued autonomous research agent work through AlphaFold and related programs, could pre-empt Faraday with a competing product before Inherent hits a credible peer-reviewed scientific benchmark.
  • The 'self-improving' framing will draw immediate regulatory attention in the UK, where AI rules are actively being drafted; Matt Clifford's advisory profile may increase scrutiny as much as it deflects it.
  • Index Ventures and Radical Ventures face reputational exposure if Faraday's early scientific claims fail independent replication, given the high-profile $50M seed and the lab's explicit commitment to measurable breakthroughs.

Opportunities

  • Scientific compute providers (CoreWeave, Lambda Labs) and cloud platforms (AWS, Google Cloud) are natural infrastructure partners for Faraday's self-improving agent workloads as usage scales.
  • UK-based pharmaceutical and biotech firms (AstraZeneca, GSK, Recursion Pharmaceuticals) are positioned as early design partners, aligned with the UK government's ongoing AI in life sciences initiative.
  • Competing labs building research automation tools (Sakana AI, FutureHouse) may face investor revaluation as Inherent's $50M seed resets market expectations for what a credible self-improving scientific AI lab requires at entry stage.

What we don't know yet

  • What 'self-improving' means technically for Faraday: no architecture details, benchmarks, or evaluation methodology have been disclosed, leaving the core claim unverifiable at launch.
  • Whether Inherent has secured scientific institution or government research partnerships, which the founders' policy background and Matt Clifford's advisory role suggest may already be in negotiation.
  • How Faraday handles safety constraints on self-modification: the announcement named no alignment framework, external oversight mechanism, or red-teaming process.

Originally reported by tech.eu

Read the original article →

Original headline: Inherent Emerges From Stealth With $50M Seed to Build Self-Improving Scientific AI — Ex-DeepMind Team, Index Ventures, NVIDIA Ventures