Pebblebed

Pebblebed Ventures Research

Investment Thesis

Pebblebed Ventures is a frontier deep-tech venture capital firm founded in 2022 by experienced operators from OpenAI, Facebook, and Stripe. The firm explicitly backs "the foundational layers of progress"—companies building developer platforms, AI systems, robot operating systems, simulation engines, and formal verification tools. Their thesis centers on investing in technically elite founders solving genuinely hard problems where the solution compounds quietly over time.

The firm was co-founded by Pamela Vagata (creator of FBLearner Flow at Facebook, founding member of OpenAI, former AI lead at Stripe) and Keith Adams (founder of Facebook AI Research, former Chief Architect at Slack). This deep expertise in systems thinking, distributed computing, and AI infrastructure positions them uniquely to evaluate frontier tech opportunities.

Sector and Technology Focus

Pebblebed invests across multiple deep-tech verticals:

AI Infrastructure & Systems:

• AI model optimization and efficiency (Lemurian Labs - universal AI compiler)
• AI security and vulnerability detection (Vidoc Security - AI security engineer)
• Formal verification and theorem proving (Logical Intelligence)
• AI for scientific and mathematical discovery (Principia Labs, Aneta)
• Generative AI for creative tools (KREA - "Adobe for the AI era")

Developer Platforms & Tools:

• WebAssembly infrastructure (Dylibso - plugin system developer tools)
• DevOps and deployment platforms (Northflank - self-service PaaS)
• ML dataset versioning (Oxen - Git for ML datasets)
• Enterprise data intelligence (Datalab, Select Star - AI-ready data catalogs)

Robotics & Autonomous Systems:

• General-purpose AI for robotics (Tau Robotics - AI for robotic arms)
• Simulation platforms (Zeromatter - simulation for autonomous systems, founded by Tesla Autopilot engineer)
• Brain-computer interfaces (Orbit Engineering - non-invasive BCI)
• Autonomous vehicles and underwater robotics (Ulysses - autonomous underwater vehicles)

Biotech & Life Sciences:

• Genomic screening (Orchid - whole genome screening for IVF)
• AI-assisted healthcare (Headquarters Health - AI-assisted telehealth for concussion care)
• Biotech innovation (Anomaly Bio - stealth mode biotech)

Applied AI Solutions:

• Enterprise AI coding (Augment - on-premise AI coding assistant)
• Document intelligence (Datalab)
• Voice synthesis (Storyteller - AI-powered voice synthesis for creators)
• Business automation (Lynq, Adaptive Computer - natural language app building)
• Commercial real estate analytics (Build.inc)

Research Residency Program: Pebblebed operates an intentional learning community called "Research Residency" where researchers explore "ideas too weird for a roadmap." Current residents include Mathew Vanherreweghe (studying alternatives to transformers) and Jenny Guanni Qu (building AI security tools to find zero-day vulnerabilities).

Stage Focus and Check Size

Pebblebed is explicitly a seed-stage focused venture capital firm with check sizes in the $250K-$500K range for pre-seed and $1M-$3M for seed rounds. Recent data shows:

• Minimum check size: $10,000 (per F4-OS database)
• Maximum check size: $500,000 (per F4-OS database)
• Most frequent range: $500K-$2M (inferred from portfolio and recent activity)

Recent co-led investment in Lemurian Labs ($28M Series A in December 2025) indicates they may participate in larger follow-on rounds but their primary focus remains seed and pre-seed stages.

Fund Structure and Capital

Fund I: Completed with 9 investments. Most recent deal with Northflank (November 2024).

Fund II: Announced August 2024, raised $125M. Currently in active deployment mode. Recent activity includes:

• Co-led $28M Series A in Lemurian Labs (December 2025) with Hexagon
• Participated in OpenMind Series A (August 2025)
• Pre-seed and seed investments throughout 2024-2025

Total AUM: Approximately $125M (Fund II, with Fund I already deployed)

Lead Tendency and Investment Approach

Pebblebed leads and co-leads seed rounds. Evidence:

• Led Lemurian Labs Series A (December 2025) alongside Hexagon
• Described as "leads on ideal too weird for a roadmap" in their Research Residency program
• Portfolio depth suggests active board/advisor seats across portfolio
• Strong operational support through research residency and community building

Recent Activity and Portfolio Growth

Latest Investment: Lemurian Labs - $28M Series A (December 2025)

2025 Activity:

• Multiple seed and pre-seed investments across AI infrastructure, developer tools, and robotics
• Active participation in later-stage rounds (Series A for strategic portfolio companies)
• Research residency producing published work and AI research contributions

Notable Portfolio Breadth: 30+ portfolio companies across deep tech sectors

Fund Status: Fund II is actively deploying with momentum through end of 2025.

Portfolio Highlights

Active Companies (Sample):

• Lemurian Labs - AI compiler making AI hardware-agnostic ($28M Series A, Dec 2025)
• KREA - Generative design platform ("Adobe for the AI era")
• Zeromatter - Simulation platform for autonomous systems (founded by Tesla Autopilot engineer)
• Tau Robotics - General-purpose AI for robotic manipulation
• Logical Intelligence - Formal verification and theorem proving
• Augment - Enterprise AI coding assistant with on-premise deployment
• Orchid - Genomic screening for IVF embryos
• OpenMind - Decentralized embodied intelligence
• Vidoc Security - AI security engineer for vulnerability detection
• Northflank - Self-service PaaS built on Kubernetes
• Oxen - Git for ML datasets
• Select Star - Automated data catalog for AI-ready data
• Spectral Labs - Generative model for CAD
• Dylibso - WebAssembly infrastructure for plugin systems
• Aneta - AI automation for scientific workflows

Research Portfolio (Stealth):

• Wired Labs
• Anomaly Bio
• Keru

Decision Process and Engagement Style

Decision Structure: Partnership model with collaborative decision-making among principals

Decision Timeline: Typical 2-4 weeks for seed decisions with strong conviction

Involvement Level: Board seats and active advisory involvement; strong operational support through:

• Research residency program access
• Pebbleverse community platform for portfolio companies
• Direct expertise from principals with deep systems and AI backgrounds
• Technical mentorship from residents and advisory network

Warm Introductions: Preferred but not required given strong reputation and active sourcing

Founder Preferences

Pebblebed explicitly seeks technical founders who:

1. Are solving genuinely hard technical problems
2. Have deep domain expertise (systems, AI, robotics, formal verification, etc.)
3. Bring operational experience from scale-ups or hyperscalers (Facebook, OpenAI, Stripe, Tesla, Google, etc.)
4. Demonstrate "elegance in complexity and possibility in constraints"
5. Are building foundational layers that "quietly compound"

They invest across multiple founder archetypes:

• Scientists and researchers building new tools (Logical Intelligence, Principia Labs)
• Serial operators from big tech (many founders from Meta, OpenAI, Stripe)
• Category creators in new AI application areas (KREA, Augment)
• Hardware/robotics engineers building AI systems (Tau, Zeromatter)

Geographic Focus

Primarily San Francisco Bay Area with emerging presence in:

• East Coast: Some portfolio companies based in NYC and Boston
• International: Select companies in Europe and Asia (e.g., Vidoc is Poland-founded)

Thesis and Anti-Thesis

Thesis Summary: "We back the foundational layers of progress. Developer platforms that accelerate creation. Robot operating systems that make autonomy real. Simulation engines training the next generation of AI. Formal verification securing billions in value. Foundational systems that quietly compound—built by founders who see elegance in complexity and possibility in constraints."

Anti-Thesis (Explicit Exclusions):

• Consumer social apps without defensible moats
• Fintech payment processing
• Healthcare services (though support healthcare infrastructure and tools)
• Pure SaaS without technical defensibility
• Me-too solutions in crowded markets

Investment Philosophy and Culture

Pebblebed is distinctive in venture for:

1. Technical depth: Multiple partners with PhD-equivalent expertise and systems thinking
2. Research orientation: Active research residency program producing published papers and code
3. Long-term thinking: Willingness to back ideas that may take 5-10 years to mature
4. Community building: "Pebbleverse" community for portfolio companies and ecosystem partners
5. Founder-first: Strong emphasis on founder autonomy and elegance in problem-solving
6. Multi-disciplinary: Simultaneously investing across AI, robotics, biotech, and developer tools

Knowledge and Expertise Access

The fund provides access to:

• Technical mentorship from partners with backgrounds in:

  • Facebook AI Research and large-scale ML systems (Keith Adams)
  • OpenAI and frontier AI (Pamela Vagata)
  • Slack architecture (Keith Adams)
  • Stripe infrastructure (Pamela Vagata)
  • Research & AI scaling (Michael Thomas from Meta/Cruise/CZI)
  • Operational excellence (Tammie Siew from Sequoia)

• Research residency program with access to elite researchers

• Founder network across deep tech sectors

• Technical advisory network (published researchers, systems experts)

Competitive Positioning

Pebblebed competes with and differs from:

• a16z: Broader scope, later stages; Pebblebed is more specialized in deep tech seeds
• Menlo Ventures: Similar stage focus; Pebblebed more thesis-driven on frontier tech
• Conviction: Direct competitor on deep tech seeds; Pebblebed has more AI/robotics focus
• Y Combinator: Similar seed stage; Pebblebed is institution-building while YC is accelerator

Summary

Pebblebed Ventures is a uniquely positioned frontier deep-tech venture capital firm with exceptional technical expertise, clear investment thesis, and active deployment through Fund II. The team's backgrounds in foundational AI, systems architecture, and large-scale infrastructure give them distinctive conviction and ability to support companies building the next generation of computing platforms. The combination of ambitious portfolio (30+ companies across AI, robotics, biotech, and developer tools), active research program, and founder-first culture makes them a highly attractive investor for technical founders solving hard problems.