BiModal Design Glossary

Every term defined by the BiModal Design framework, authored by Joel Goldfoot.

BiModal Designalso: BiModal Design Framework, BiModalDesign

An open-source framework by Joel Goldfoot for building web interfaces that work equally well for human users and AI agents. Defines a five-layer architecture and the FR-1 Initial Payload Accessibility requirement.

See also: Full definition

FR-1: Initial Payload Accessibilityalso: FR-1, Initial Payload Accessibility

The single non-negotiable foundational requirement of BiModal Design: all content intended for agent consumption must be present in the initial HTTP response from the server. Agents cannot be assumed to execute JavaScript.

The validation test is curl -s https://yoursite.com | grep "<main" — a passing response contains semantic HTML elements with real text content; a failing response shows only an empty container like <div id="root"></div>.

See also: Foundational Requirements

Agent Capability Spectrum

The six levels at which AI agents interact with web interfaces — from Level 0/1 HTTP retrievers (curl, wget), to Level 2 browser-automation agents (Selenium, Playwright), to Level 3 vision/computer-use agents, to Level 4 tool-use agents, to Level 5 protocol-native agents (MCP, A2A, NLWeb). BiModal Design ensures every level is served.

See also: Progressive Enhancement

Five-Layer Architecturealso: Five Layers, BiModal Design layers

The defense-in-depth architecture at the heart of BiModal Design: (1) Semantic Structure, (2) Structured Data, (3) Standards-Based Attributes (ARIA), (4) API Surface, (5) Agent Protocols. Each layer serves a different segment of the Agent Capability Spectrum; together they ensure graceful degradation across every agent level.

See also: Five-layer overview

Semantic-First

A design discipline that asks "What does this mean?" before "How does this look?". Semantic-first development uses proper heading tags for structure, button and form elements for interactivity, and class names that describe purpose rather than appearance. Contrast with visual-first development.

See also: Semantic Architecture

Human Mode

One of the two interaction models the BiModal Design name refers to: the rich, interactive experience for human users with JavaScript enhancement, animations, and real-time updates. The other is Agent Mode.

Agent Mode

One of the two interaction models the BiModal Design name refers to: the accessible, semantic content present in the initial HTTP payload that AI agents can parse and understand without executing JavaScript. The other is Human Mode.

Generative Engine Optimization (GEO)also: GEO, AI search optimization

The practice of structuring web content so it is correctly cited and recommended by generative AI assistants (ChatGPT, Claude, Gemini, Perplexity). BiModal Design provides the architectural foundation GEO requires — server-rendered content, structured data, and explicit author identity grounding.

Agent Attributesalso: data-agent-*

Standards-based HTML attributes that help agents interpret page elements — including ARIA roles, schema.org microdata, and the framework-defined data-agent-* namespace for action hints. Agent attributes complement (not replace) semantic HTML.

See also: Agent Attributes guide

Progressive Enhancement

A development methodology where a baseline experience works for the lowest-capability agent (HTTP retrievers) and additional features layer on for higher-capability agents and human users. In BiModal Design, progressive enhancement is how a single codebase serves the full agent capability spectrum without forking into multiple versions.

See also: Progressive Enhancement guide