Sam Schillace
Product visionary and AI thought leader exploring compounding systems, autonomous tool-building, and the future of software engineering with AI agents
Background
Sam Schillace is a prominent voice in the AI and software engineering community, known for his deep thinking on how AI fundamentally changes development practices and organizational structure.
Key Roles & Experience
- Product leadership – Experience building products at scale
- AI advocate – Early adopter and practitioner of agentic development
- Thought leader – Regular contributor to AI/engineering discourse
- Writer – Publishes insights on evolving AI engineering practices
Major Contributions
”I Have Seen the Compounding Teams”
Sam’s most influential essay articulates Compounding Teams—the observation that small groups of engineers leveraging AI agents can produce output of 10-15+ traditional engineers through recursive tool-building.
Core insights:
- Best tool for building tools is a tool (recursion)
- Productivity compounds exponentially, not linearly
- Infrastructure tools (git, markdown, filesystem) natural for AI
- AI context refinement becomes operational game
- Applies to all knowledge work, not just coding
Impact: Defined vocabulary and framework for understanding how AI amplifies team productivity
Compounding Correctness Observation
Sam recognized the inflection point when Claude 3.5 Sonnet v2 (October 2024) shifted from compounding error to compounding correctness:
Before: AI iterations accumulated errors, leading to system decay
After: AI iterations improved code quality, compounding improvements
This insight became foundational to understanding Software Factory and Agentic Development.
Knowledge Work Universalization
Sam argues the compounding teams pattern extends beyond software to all knowledge work:
- Research teams building tools to accelerate discovery
- Product teams using AI for feature planning
- Marketing teams orchestrating campaigns
- Bug fixing becoming automated and systematic
Implication: The 2-3 person “compounding team” model will eventually apply across all knowledge domains.
Key Concepts Developed
1. Compounding Systems
- Versus traditional linear scaling
- Knowledge accumulates, accelerating future work
- Applies to engineering, research, product, marketing
2. Context Refinement (“Context Engineering”)
- Structuring AI system’s knowledge base
- Iteratively improving through feedback
- Non-experts becoming capable through refined context
- Persisted context as competitive moat
3. Recursive Tool-Building
- AI builds tools that enable building better tools
- Self-improving infrastructure
- Exponential capability growth
- Eventually reaches autonomy
4. The Coordinate Problem
- Compounding systems require specialization and isolation
- Coordination costs negate benefits
- Small focused teams outperform large groups
- Future: competing model providers as checks/balances
Influence & Advocacy
On Industry Thinking
- Shaped understanding of AI’s impact on team dynamics
- Challenged assumptions about required team size
- Articulated path from “AI assistance” to “AI orchestration”
- Influenced how companies structure AI-driven development
On Product & Practice
- Provided framework for organizations adopting AI agents
- Validated practices at Every and other early adopters
- Created vocabulary: “compounding teams,” “context refinement”
- Guided thinking on scaling small teams with AI
Writing & Publications
Primary Platform
- Sunday Letters from Sam – Substack newsletter
- Regular essays on AI, engineering, team dynamics
- Accessible writing on complex concepts
- Thought-provoking questions about future of work
Notable Essays
- “I Have Seen the Compounding Teams”
- Various pieces on:
- AI’s impact on traditional roles
- How context determines capability
- Coordination challenges in AI systems
- Knowledge work transformation
Perspective on AI & Engineering
Core Beliefs
- AI changes human role, not eliminates it – Shifts from implementation to orchestration
- Context is everything – System capability determined by knowledge base
- Compounding beats speed – Accumulated knowledge beats raw iteration speed
- Knowledge work will transform – Not just software, all domains
- Small teams can scale – With proper AI orchestration
Vision for Future
- 2-3 person teams producing output of hundreds
- Knowledge becomes strategic moat
- Context engineering as critical skill
- Competing AI systems for quality assurance
- Ambient intelligence replacing traditional “software”
Relationship to Other Thinkers
Aligned With
- Dan Shapiro – “Five Levels from Spicy Autocomplete to Software Factory”
- Luke PM – “The Software Factory” concept
- Andrej Karpathy – “Software 2.0” thinking (AI as code)
Complements
- Anthropic – Building Claude models enabling compounding
- Cursor – Providing IDE infrastructure for compounding workflows
- Every – Practical demonstration of concepts
Impact & Legacy
Short Term (2026)
- Vocabulary for AI-enhanced teams
- Framework for adoption
- Validation of emerging practices
- Guidance for organizations experimenting
Medium Term (2027-2029)
- Industry-wide shift toward compounding systems
- New organizational structures emerging
- Team size optimization flipped
- Knowledge management becomes discipline
Long Term (2029+)
- “Compounding teams” become standard model
- Context engineering as recognized profession
- Traditional hiring models obsolete
- Sam’s thinking foundational to how work operates
Key Quotes & Ideas
On compounding:
“The best tool for building a tool is a tool”
On coordination:
“Compounding systems require specialization and isolation. Larger teams fragment productivity.”
On knowledge work:
“What applies to software engineering applies to all knowledge work—research, product, marketing, everything.”
On future:
“Small teams amplified by AI will outcompete large traditional organizations”
Current Work
- Writing regularly on AI and software evolution
- Advising organizations adopting agentic development
- Observing and documenting compounding teams in practice
- Thinking about next-generation coordination problems
Resources
- Substack: Sunday Letters from Sam
- Essay: “I Have Seen the Compounding Teams”
- References: Cited in Compounding Teams, Software Factory, Compound Engineering
Related Concepts
- Compounding Teams – His major contribution
- Compound Engineering – The methodology he describes
- Software Factory – Related enterprise approach
- Agentic Development – Technical implementation
- Context Graphs – Knowledge persistence mechanism
Vision Summary
Sam Schillace articulates a fundamental shift in how software and knowledge work will function: small teams of humans orchestrating AI agents, with productivity compounding through knowledge accumulation and recursive tool-building, eventually resulting in systems where 2-3 people produce what 50+ traditional employees would.
His thinking provides both framework and vocabulary for understanding the transformation happening in real-time across leading organizations.