01 · Intelligence & infrastructure
Capability at a viable cost.
Models, inference efficiency, compute, evaluation tooling and agent runtime.
Does the system materially improve capability, reliability or unit economics?
Fam Capital Research · Intelligence Layer
AI / Investment thesis
From intelligence to execution.
We look for AI systems that turn technical capability into reliable work, measurable customer value and durable business advantages.
Model → system → work01 / AI
Reliable
work.
work.
The commercial system around a model matters as much as the model itself.
01 / The shift
From generated content to completed work.
AI capability is advancing quickly. But a model’s ability to produce an answer, a draft or an API response is not the same as a system’s ability to finish a task safely inside a real workflow.
The opportunity widens when intelligence can use tools, work with context, operate within permissions, surface uncertainty and improve from evaluated feedback. The investing question is therefore not simply which model is strongest. It is whether the complete system produces an outcome customers can rely on and pay for repeatedly.
02 / Our investment lens
Four layers between capability and advantage.
We do not presume that value must concentrate at the model layer or the application layer. We examine the part of the stack where a system becomes more useful, more reliable or more difficult to reproduce.
02 · Agent systems
Execution inside boundaries.
Tool use, memory, planning, permissions, evaluation and execution verification.
Can it complete a clear task repeatedly and recognise or handle failure?
03 · Applications & workflows
Outcome over interface.
Vertical applications, enterprise workflows and delivery of a useful result.
Does it solve an important problem that a customer will keep paying to solve?
04 · Data & distribution
Advantage that compounds.
Legally usable proprietary data, evaluated feedback, channels and customer relationships.
Does growing use create an advantage that rivals cannot easily copy?
03 / What we back
Systems with a path from intelligence to durable work.
We seek products and infrastructure that take responsibility for the entire job: what must happen, what can fail, who must approve, what the full cost is and why the customer will return.
Selected portfolio context
Anthropic
Frontier models advancing reliable, high-performance intelligence.
Selected portfolio context
NVIDIA
Compute, simulation and deployment infrastructure for intelligent systems.
Selected portfolio context
Tesla
Autonomy at scale across vehicles, robotics, data and compute.
04 / What we study
The open questions around reliable execution.
Our research follows where task-level reliability becomes more important than a headline benchmark, and where operating systems—not prompts alone—create a defensible product.
Evaluation in production
How a system measures task success, detects uncertainty and makes failure visible before the cost compounds.
Human intervention
Which approvals improve trust and quality, which create an uneconomic labor layer, and when escalation is the product.
Workflow ownership
Where software becomes embedded in the operating sequence instead of adding another surface for users to supervise.
Model change risk
How faster, cheaper or more capable models can create new products while compressing a current product’s advantage.
05 / The execution standard
A system earns value when it can carry the work.
We underwrite the complete delivery system—not only a model result or a product demo. Reliable execution, economic delivery and durable advantage must move together.
01
Task success
Does the system complete the relevant task at a quality level the workflow can accept?
02
Failure cost
How are errors detected, contained, corrected and priced into the true delivery model?
03
Total economics
Do inference, review, implementation, support and failure handling still leave a compelling unit model?
04
Compounding edge
Do workflow depth, distribution, data rights and switching costs become stronger with use?
Capability creates possibility. Reliable execution creates value.
A Fam Capital investment perspective on intelligence, execution and durable advantage.