Agent Capability Spectrum
Six levels of AI agents interact with your interface — from HTTP retrievers to protocol-native systems. BiModal Design ensures every level is served.
Learn moreBiModal Design
An open-source framework for designing interfaces that serve humans and AI agents
Most sites are invisible to the AI agents that visit them. BiModal Design makes yours work across the full capability spectrum — from HTTP retrievers to vision agents to protocol-native systems.
Created by Joel Goldfoot
…or skip to the 5-minute Quick Start.
$
curl -s https://bimodal.design | grep "<main"✓
<main> … full content in the HTML — every agent can read it$
curl -s https://typical-spa.example | grep "<main"✗
<div id="root"></div> — empty shell; invisible to agents that don't run JSKey Framework Principles
Defense in Depth
Five architectural layers — from server-rendered content through semantic structure, structured data, APIs, to agent protocols — ensure graceful degradation across the entire spectrum.
Learn moreStandards-Based Approach
Built on schema.org, WAI-ARIA, OpenAPI, and emerging protocols like MCP — no custom attributes needed. Use the standards the ecosystem already understands.
Learn moreWhat agent benchmarks show
Approximate task-completion ranges from public benchmarks (2023-2026)
| Agent capability level | Task-completion range on the general web |
|---|---|
| HTTP retrievers (L0-1) | ~12-20% |
| Browser automation (L2) | ~35-50% |
| Vision agents (L3) | ~40-55% |
| API / protocol (L4-5) | N/A (protocol channel, not general-web browsing) |
| Human success (reference) | ~72-89% |
Read this table as directional, not measured. These figures come from public agent benchmarks (WebArena, VisualWebArena, ST-WebAgentBench, and related work) that evaluate agents on the current web — not from a controlled A/B test of BiModal Design against a conventional interface. BiModal Design maps the design patterns those benchmarks reward (semantic HTML, structured data, ARIA, agent protocols) to the capability levels above, but the framework itself has not been independently benchmarked head-to-head.
Sources: WebArena (Zhou et al., 2023, arXiv:2307.13854); VisualWebArena (Koh et al., 2024, arXiv:2401.13649); ST-WebAgentBench (Levy et al., IBM Research, 2024, arXiv:2410.06703).