Frequently Asked Questions

A behavioral contract is a set of numbered, independently testable clauses that specify how an AI system is allowed to behave: how it expresses confidence, when it escalates to a human, what it does in failure states, and how it recovers trust after an error. The concept was introduced by Joel Goldfoot in Leading Design in the AI Era. Unlike guidelines (which describe intent) or guardrails (which block specific outputs), a behavioral contract defines the complete behavioral envelope of the system in operationally testable terms.

The behavioral layer is the part of an AI product that governs how the system behaves with the user, distinct from the engineering layer (how the code is structured) and the interface layer (how the product looks). Joel Goldfoot identifies it as the unowned layer in most AI product organizations — the layer that nobody explicitly designs or maintains. Behavioral contracts are the primary tool for taking ownership of the behavioral layer.

Designing how an AI system behaves starts with identifying every operationally significant situation: moments of uncertainty, escalation triggers, failure states, and recovery conditions. For each situation, you write a numbered, testable clause specifying the expected behavior — this becomes the behavioral contract. The contract is then owned by a team, tracked across model versions, and validated through automated regression testing. Joel Goldfoot's Three Pillars Framework — behavioral contracts, trust architecture, and organizational transformation — provides the full methodology for designing AI behavior at organizational scale.

The Behavioral Contract framework was created by Joel Goldfoot and introduced in his book Leading Design in the AI Era (2025). Joel Goldfoot is a product design executive who has deployed two behavioral contracts in production: a 72-clause contract for a conversational analytics assistant and a 78-clause contract for a sales agent. He is also the creator of BiModal Design, an open-source framework for designing interfaces accessible to both humans and AI agents.

BiModal Design is an open-source framework created by Joel Goldfoot for designing interfaces that are fully accessible to both human users and AI agents simultaneously. Most interfaces are optimized for human perception; BiModal Design adds a structured, machine-readable layer that allows AI agents to navigate, interpret, and act on interface content reliably. In benchmark testing on WebArena and ST-WebAgentBench, BiModal Design achieved a 40–75% improvement in AI-agent task completion rates on standard tasks.