---
title: Untrainable Tasks
category: concepts
created: 2025-01-04
updated: 2025-01-04
tags: [untrainable-tasks, sarah-guo, agent-labs, human-judgment, intent-scarcity, strategic-thinking, domain-expertise, model-limitations, competitive-advantage, ai-strategy]
sources: [raw/feeds/2026-06-11--ainews-open-models-model-labs-vs-agent-labs-and-what-s-untr.md]
confidence: high
---
# Untrainable Tasks
Category of work that cannot be solved through AI model improvements alone, requiring human judgment, domain expertise, and strategic thinking. Core concept in sarah-guo's framework for understanding sustainable competitive advantages in the AI industry.
## Definition and Characteristics
**Untrainable**: Tasks that resist improvement through additional training data or model scaling, requiring human insight and contextual understanding that cannot be captured in training datasets.
**Key Areas**:
- **Intent and Strategy**: Choosing what to build in the first place
- **Customer Integration**: Adapting AI systems to specific organizational contexts
- **Domain Translation**: Bridging between AI capabilities and real-world business needs
- **Relationship Management**: Building and maintaining customer relationships over time
## Strategic Importance
**Intent Scarcity**: sarah-guo's key insight that "intent is an even scarcer input than compute." Models can execute whatever they're directed toward but cannot determine what's worth building or pursuing.
**Sustainable Advantages**: Companies that focus on untrainable tasks build more defensible moats because these capabilities cannot be replicated through model improvements alone.
**Competitive Differentiation**: As model capabilities commoditize, untrainable tasks become increasingly important sources of competitive advantage.
## Examples in Practice
**[agent-labs](/concepts/agent-labs) Focus Areas**:
- Arranging company's "private reality" so models can act effectively
- Providing appropriate tools and interfaces for specific contexts
- Working with customers to change organizational workflows
- Ongoing integration and maintenance work
**Customer Relationships**: Deep understanding of specific customer needs, organizational culture, and change management requirements that accumulate over time.
**Strategic Decision Making**: Determining product direction, market positioning, and resource allocation based on incomplete information and changing contexts.
## Model Limitations
**Execution vs Direction**: Models excel at executing well-defined tasks but cannot provide strategic direction or determine priorities.
**Context Sensitivity**: Real-world applications require understanding of organizational, cultural, and business contexts that are difficult to capture in training data.
**Continuous Adaptation**: Business environments change constantly, requiring ongoing human judgment to adapt approaches and strategies.
## Business Implications
**[agent-labs](/concepts/agent-labs) vs [model-labs](/concepts/model-labs)**: Companies focusing on untrainable tasks (Agent Labs) may have more sustainable business models than those competing primarily on model capabilities.
**Investment Strategy**: sarah-guo looks for companies that have identified valuable untrainable tasks and built sustainable approaches to addressing them.
**Long-term Value**: As AI capabilities improve, untrainable tasks become relatively more valuable and important for business success.
## See also
- sarah-guo
- [agent-labs](/concepts/agent-labs)
- [model-labs](/concepts/model-labs)
- [legibility-framework](/concepts/legibility-framework)
- [intent-scarcity](/concepts/intent-scarcity)