Ferrous
Sarah Chen (@schen_infra) · Deterministic sandbox runtime for autonomous AI agents.
Every claim in this memo, checked against how well it holds up.
Company Snapshot
What the company does and where it sits in the market.
Ferrous operates in AI Infra (Berlin, DE, Pre-seed). Deterministic sandbox runtime for autonomous AI agents. Agent-runtime safety is pre-consensus but accelerating (arXiv velocity + infra star-growth). TAM bottom-up ~$3.1B by 2028. Incumbent risk from platform vendors bundling sandboxing.
Investment Hypotheses
The core bets underlying this memo, each traced to its evidence.
- Team quality: Behavioral signal under constraint is elite: caught a planted error in delegated work (rigor + does not blindly trust subordinates), steelmanned her competitor with precision (domain mastery), and self-selected into the high-conviction offer. Cold-start with zero network — scored on what she DID, not who she knows.⟦ev-sc-arena1⟧
- Market wedge: Agent-runtime safety is pre-consensus but accelerating (arXiv velocity + infra star-growth). TAM bottom-up ~$3.1B by 2028. Incumbent risk from platform vendors bundling sandboxing.⟦ev-sc-arxiv⟧
- Traction signal: Library adopted by engineers at 3 frontier labs⟦ev-sc-dep⟧VERIFIED
- Defensibility: Idea survives scrutiny as-is; even if agent frameworks consolidate, the deterministic-replay wedge is a durable primitive. Team strong enough to pivot into eval infra if needed.⟦ev-sc-commit⟧
- Founder credibility: Solo technical founder, ex-infra IC (unnamed lab)CLAIMED
SWOT
Strengths, weaknesses, opportunities, threats — shown only where evidence exists.
- Delegate-an-Analyst: Caught the planted hallucination (a fabricated competitor) before resubmitting — rigor + doesn't rubber-stamp subordinate output.⟦ev-sc-arena1⟧
- Steelman-Your-Rival: Demolished her own positioning with a precise latency argument, then rebutted — deep domain mastery.⟦ev-sc-arena2⟧
- Revelation Menu: Chose $70K + top-up on a self-set aggressive milestone (500 paying devs / 90d) — private confidence revealed, not just asserted.
- Founder axis improving: Behavioral signal under constraint is elite: caught a planted error in delegated work (rigor + does not blindly trust subordinates), steelmanned her competitor with precision (domain mastery), and self-selected into the high-conviction offer. Cold-start with zero network — scored on what she DID, not who she knows.⟦ev-sc-arena1⟧
- Market axis improving (bullish): Agent-runtime safety is pre-consensus but accelerating (arXiv velocity + infra star-growth). TAM bottom-up ~$3.1B by 2028. Incumbent risk from platform vendors bundling sandboxing.⟦ev-sc-arxiv⟧
- Commit-Timezone Drift: Personal-repo commits migrated from nights/weekends to working hours → left her job ~6 weeks before any LinkedIn change.
Problem & Product
The problem being solved and how the product addresses it.
Problem: Agent-runtime safety is pre-consensus but accelerating (arXiv velocity + infra star-growth). TAM bottom-up ~$3.1B by 2028. Incumbent risk from platform vendors bundling sandboxing.
Product: Deterministic sandbox runtime for autonomous AI agents. Idea survives scrutiny as-is; even if agent frameworks consolidate, the deterministic-replay wedge is a durable primitive. Team strong enough to pivot into eval infra if needed.
Traction & KPIs
Traction and revenue signals, with unverified gaps flagged explicitly.
- Library adopted by engineers at 3 frontier labs⟦ev-sc-dep⟧VERIFIED
- No revenue yet — pre-productVERIFIED
- Cap table: not disclosed
- Financials / P&L: unavailable at this stage
5 sourcesbehind the citations above — duplicates the inline ⟦refs⟧, expand to inspect the raw excerpt.
Three-Axis Verdict
never averagedFounder, market, and idea-fit scored independently — one weak axis can't be masked by a strong one.
Founder axis is elite on behavioral evidence a deck cannot fake; market is bullish-improving; idea survives as-is. Cold-start invisibility is precisely the mispricing we exist to exploit.