FUSE INTELLIGENCE
Changelog
Release notes for VocalFuse and VibeFuse. VibeFuse entries use v0.x semver; VocalFuse uses v1.x.
VibeFuse — canvas header now holds zoom and widget focus/hide/close; HyperFrames Renders lists the workspace renders folder; sidebar overflow for long widget/skill lists; marketplace styling publish
VibeFuse — Firecrawl works without Docker: paste a Firecrawl API key in Settings → Tools and the MCP plus built-in search/scrape tools use the cloud API; installer re-downloads when the installed exe version is stale
VibeFuse — voice Piper speaks only a start cue and a final answer (no mid-tool TTS); Linear widget GraphQL stays under the complexity cap; browser tab clicks work on a zoomed canvas
VibeFuse — GitHub repo listing via PAT (no terminal fallback), Discord MCP default guild + snowflake IDs, Linear MCP/widget, installer CLI skip checkboxes
VibeFuse — GitHub MCP / Firecrawl local MCP, HyperFrames widget with AI tooling, Claude Code–compatible skills, FreeClaudeCode CLI clone with any provider/model
Website SEO & technical improvements: fixed version inconsistencies, corrected sitemap.xml rewrite rule, added missing alt text attributes for accessibility
SEO improvements: updated meta descriptions and structured data for better search visibility; added export format roadmap documentation
SEO improvement: updated homepage title and description for better click-through rate and search visibility.
SEO improvements: updated homepage title/description/keywords targeting voice transcription keywords; updated VocalFuse product page SEO; updated Speech to Text Windows page SEO; updated Otter AI alternative page SEO with expanded keywords
VibeFuse — YouTube widget, voice dock polish (left equalizer, right transcript), maximize layout fix, edge resize handles, Ctrl+Tab sessions, per-session workspace path, browser save-as-widget, terminal restore
VibeFuse public preview — widget canvas, agent CLIs, voice dock, sessions, browser widget, marketplace folder
After a product key verifies, the desktop loads that account’s profile (name, email, Basic/Pro, store-notes-on-website). Basic keeps Note Taking hidden
Security hardening: removed all embedded database and FTP credentials from the app binaries — license verification and update checks now use HTTPS-only endpoints
Pill mode dropdown uses dark-mode row styling; app icon and in-app logo match the website VocalFuse emblem
Basic vs Pro gating: Note Taking and email settings are Pro-only; website profile email-to and cloud notes toggle sync via license verify
Note Taking mode also saves each session as JSON under C:\VocalFuse\Notes\
Note summaries use Featherless GLM-5.3-Flash; Account → Notes groups each session’s summary and transcripts in one card
Installer license verify handles Hostinger compression and parses API responses correctly
Installer license check uses HTTPS API (no embedded MySQL redistrib required)
Note-taking uploads now sync to fuseintelligence.org Account → Notes
Fixed settings update status text drawing over the header
Smaller settings dialog and pill UI; dark-mode text fields; fixed Recording Mode spacing
Removed NEUS note-proof uploads and NEUS SDK from the desktop app
Cloud note sync — note-taking mode uploads transcriptions to your account via license key; Account → Notes viewer
VocalFuse public beta — local on-device dictation, global hotkeys
Reduced memory footprint during long sessions
Note Taking and Continuous capture modes
Website launch — accounts, pricing, developer docs