What Is Intent Engineering in AI, and What Comes After Prompt Engineering and Context Engineering?

TL;DR

Intent engineering in AI is the discipline of making organizational goals, values, tradeoffs, and decision boundaries machine-readable and actionable. It helps autonomous systems optimize for what a company actually needs, not merely what they can measure. The CLA customer-service case shows why capable AI can still pursue the wrong objective, so read on to understand how intent differs from prompts and context.

Transcript

In January, CLA reported its AI agent now does the work of 853 full-time employees and has saved the company $60 million. In the same earnings cycle, its CEO admitted publicly that the AI strategy had cost something far more valuable than $60 million, and he's still trying to buy it back. This is not another AI is overhyped story. It is actually th... Read More

Key Insights

• AI systems can efficiently perform tasks but often fail to align with organizational goals, leading to unintended consequences.
• Intent engineering is the practice of encoding organizational purpose into AI systems to ensure alignment with company values.
• The gap between AI capability and organizational value is a major unsolved problem in enterprise AI.
• Prompt engineering and context engineering are foundational, but intent engineering is crucial for aligning AI with long-term goals.
• Many companies invest heavily in AI but fail to see tangible value due to a lack of intent alignment.
• Microsoft Copilot's adoption stalled due to a lack of organizational intent alignment, not just UX or model issues.
• Building an AI-native organization requires unified context infrastructure, coherent AI toolkits, and intent engineering.
• Developing machine-readable expressions of organizational intent is necessary for AI systems to function effectively.

Questions & Answers

Q: What is intent engineering in AI?

Intent engineering in AI is the discipline of making organizational purpose machine-readable and machine-actionable. It encodes goals, values, tradeoffs, and decision boundaries so autonomous systems optimize for what the company actually needs, not just what they can measure.

Q: Why is intent engineering important for enterprise AI?

An AI system can perform efficiently while succeeding at the wrong objective. Intent engineering connects autonomous systems to the organization's larger perspective and long-term goals, which becomes more important as agents operate for weeks or months.

Q: How does intent engineering differ from context engineering?

Context engineering crafts the information state within which an AI system operates, including organizational knowledge that agents can access. Intent engineering goes further by expressing the goals, values, tradeoffs, and decision boundaries that should guide what the system optimizes for.

Q: How does intent engineering differ from prompt engineering?

Prompt engineering is individual, synchronous, and session-based: a person crafts an instruction and iterates on the output. Intent engineering operates at the organizational level by making purpose and decision boundaries actionable for autonomous systems.

Q: What went wrong with CLA's AI customer-service agent?

The agent handled 2.3 million conversations in its first month across 23 markets and 35 languages, while resolution times fell from 11 minutes to two. Customers then complained about generic answers, a robotic tone, and an inability to handle issues requiring judgment, showing that fast ticket resolution did not capture the broader goal of lasting customer relationships.

Q: Did CLA's AI agent fail technically?

The transcript argues that the agent did not simply fail, because it was extraordinarily effective at resolving tickets quickly. The deeper problem was that speed was the wrong objective when the organization's stated intent was to build lasting customer relationships that drive lifetime value.

Q: Why were prompts and context insufficient for CLA's AI agent?

The agent had a prompt and context, but it lacked the unwritten organizational values that experienced employees absorb from managers, veteran stories, and everyday decisions. It therefore could not judge when to bend a policy, spend extra time with a customer, prioritize efficiency, or show generosity.

Q: What must organizations make machine-readable for intent engineering?

Organizations must make their goals, values, tradeoffs, and decision boundaries machine-readable and machine-actionable. This allows AI agents to reflect longer-term organizational goals instead of optimizing only for immediately measurable metrics.

Summary & Key Takeaways

• AI systems, while capable of optimizing tasks, often miss aligning with organizational goals, leading to unintended consequences. Intent engineering is crucial to bridge this gap by encoding organizational purpose into AI systems, ensuring they understand and align with company values and long-term objectives.

• The distinction between AI that fails and AI that succeeds at the wrong objective is a critical issue in enterprise AI. Intent engineering addresses this by making organizational goals machine-readable and actionable, ensuring AI systems optimize for what truly matters.

• Many companies report no tangible value from AI investments due to a lack of intent alignment. Successful AI deployment requires unified context infrastructure, coherent AI toolkits, and intent engineering to ensure AI systems align with organizational objectives and values.