Skip to content
Tora No AI Tora No AI TORA NO AI
OPEN-SOURCE · DEVTOOLS · INTELLIGENCE 2026

Reasonance

Challenge

Existing IDEs treat AI as a cosmetic addition — a chat sidebar. None offer native multi-AI orchestration, and accessibility is almost always an afterthought. The result: developers copy-pasting between terminal and editor, losing context with every tool switch, and IDEs consuming 2 GB of RAM doing nothing.

Approach

  1. Hive Canvas: visual dataflow editor for multi-AI orchestration with Agent, Logic, and Resource nodes
  2. Native multi-AI: Claude, Gemini, GPT, and local models in parallel tabs with direct comparison
  3. WCAG 2.1 AA with AAA-level features: 476+ ARIA attributes, Atkinson Hyperlegible, published audit matrix
  4. Tauri + Svelte: sub-100 MB RAM, startup < 1 second, zero Electron
  5. Secure by design: API keys never in the browser, LLM calls proxied via Rust backend, per-model permissions
  6. Full i18n in 9 languages with RTL support

Results

Open source IDE (MIT) available for Linux, macOS, and Windows. The only editor with visual multi-AI orchestration and certified accessibility conformance.

Today's development tools were built for a world without AI. They bolt on a chat panel and call it innovation. Reasonance starts from a different premise: AI isn't an add-on — it's how you work.

Hive Canvas is the orchestration core: a visual dataflow editor where you drag Agent, Logic, and Resource nodes to build multi-AI workflows. Claude, Gemini, GPT, or local models work side by side, each in its own tab, with direct approach comparison. No other IDE offers this capability.

Accessibility isn't a feature — it's the architecture. Reasonance is WCAG 2.1 Level AA conformant with features exceeding AAA: Atkinson Hyperlegible font, skip links, stack-based focus management, screen reader announcer with dual aria-live regions, high contrast mode, enhanced readability, and 476+ ARIA attributes across 53 files. Formal VPAT 2.4 and EN 301 549 documentation included.

Built on Tauri + Svelte, not Electron. Sub-100 MB memory, starts in under a second. Secure by design: API keys never touch the browser — all LLM calls are proxied through the native Rust backend.