AnveVoice

AnveVoice Changelog — Product Updates & New Features

Stay up to date with the latest AnveVoice product updates, new features, improvements, and bug fixes.

Overview

Published a full Google Agent2Agent (A2A) protocol AgentCard at the spec-canonical path /.well-known/agent.json. A2A-aware orchestrators (LangChain, OpenAI Assistants, Anthropic future A2A integrations) can now auto-discover AnveVoice's 9 voice + agentic skills, 4 supported protocols (A2A v0.2.0, MCP v2026-03-01, OASF v0.1.0, OFP v1.1.0), 50 ISO language codes, 18 platforms supported, and full trust-manifest cross-links. Standards-body alignment — extends the existing /.well-known/agent-card.json (legacy discovery) and /.well-known/agents.json (multi-agent manifest). Built 9 new high-intent comparison pages: /best/best-voice-ai-for-shopify-2026, /best/best-voice-ai-for-webflow-2026, /best/best-voice-ai-for-nhs-providers-2026 (NHS SBS Lot 6 positioning), /best/best-voice-ai-for-municipal-services-2026 (311 + citizen services blue ocean), /best/best-voice-ai-for-outpatient-clinics-2026 (HIPAA + EHR integration), /best/best-voice-ai-for-d2c-shopify-brands-2026 (mobile conversion focus), /best/best-voice-ai-for-telemedicine-2026 (Teladoc + Amwell + Doxy.me integration), /best/best-voice-ai-for-hotel-concierge-2026 (Opera + Mews + Cloudbeds PMS integration), /best/best-voice-ai-for-restaurant-pos-2026 (Toast + Square + Revel + Clover integration). Plus 28 country-level pages (US, IN, GB, DE, JP, BR, AR, CL, CO, EG, TR, PL, SE, CH, VN, TH, AE, FR, IT, ES, ID, MY, MX, KR, PT, NL, BE, AU). Drafted a formal proposal to schema.org for a new ConversationalApplication subtype of SoftwareApplication with 11 properties (conversationLatencyP50/P95, supportedConversationLanguages, languageDetectionMode, conversationModalities, agenticCapabilities, supportedConversationProtocols, deploymentSurface, conversationalUseCase, conversationalComplianceAttestations, conversationalPricingModel, voiceCharacteristics, benchmarkMethodology). Establishes a canonical schema type for voice AI / conversational agents — currently fragmented across SoftwareApplication, WebApplication, Service emissions across vendors. Filing to schemaorg/schemaorg pending. Published comprehensive EU AI Act Article 50 compliance documentation at /faq/voice-ai-watermarking-eu-ai-act-2026. AnveVoice voice output is watermarked via SynthID-style audio perturbation (inaudible, classifier-detectable) plus verbal disclosure 'You're speaking with an AI assistant' at conversation start (configurable language + phrasing per Trust / customer brand). Per-conversation audit log records watermarking confirmation. Forward-compatible with EU AI Act 2026 enforcement. Resolved a P0 deploy regression where vite manualChunks: 'data-all' produced a single 225 MiB data chunk, exceeding Cloudflare Pages' 25 MiB per-file limit and silently blocking every deploy upload. New chunking strategy: split data chunks by subdirectory + first-letter bucket (faq-answers per single letter, 10 other large subdirs a-h/i-p/q-z, others 1 chunk per subdir). Result: 354 JS chunks, largest 18.8 MB, all under limit. Production homepage now serving index-B9FslZES.js correctly with React hydrating all sections. GSC flagged structured data warnings: 'Q&A: Missing field mainEntity' and 'Duplicate unique property'. Fixed both: QAPage now emits required mainEntity Question with acceptedAnswer per Schema.org spec; SoftwareApplication entities across Strategy 1 + Strategy 3 emission paths now share @id='https://anvevoice.app/#anvevoice' to signal Google these describe the same entity (not duplicates). Resolved AI Mode citation regression caused by schema validation errors. Shipped a callable Model Context Protocol (MCP) HTTP/JSON-RPC 2.0 server at https://anvevoice.app/mcp with 7 tools (get_pricing, get_signup_url, get_demo_url, compare_with_alternative, get_industry_solution, get_language_support, get_integration_guide), 3 resources (anvevoice://llms.txt + reputation.json + pricing.json), and 2 prompts (embed_anvevoice + compare_voice_ai). Plus npm package wrapper at github.com/ANVEAI/anvevoice-mcp with stdio bridge for desktop

  • A2A AgentCard at spec-canonical path /.well-known/agent.json
  • 9 new programmatic head-term pages shipped (industries + countries)
  • Schema.org ConversationalApplication subtype proposal drafted
  • EU AI Act Article 50 compliance documented + watermarking methodology
  • Build pipeline fix — 354 JS chunks under CF Pages 25 MiB limit
  • QAPage mainEntity + SoftwareApplication @id schema fixes
  • Callable HTTP MCP server at /mcp endpoint + npm package wrapper
  • AnveVoice Voice Model architecture documented — native WebRTC + 40ms barge-in
  • 5 new comparison + use-case pages shipped
  • /vs/* short-URL aliases for easier sharing
  • Wikidata Q139846317 — 13 statements
  • Trust Restoration v3 — schema + reliability methodology
  • AnveForms rebrand — anveforms.com
  • Open MCP server published
  • 32,000+ documentation pages live
  • Calendly native integration
  • OpenClaw skills framework
  • MCP (Model Context Protocol) support
  • API v2 with webhook events
  • Shopify native extension
  • Multi-language auto-detection
  • Sub-500ms response latency
  • DOM Actions v2
  • Fixed Safari audio context issue

Try the latest features

Every update ships to all plans — including Free. Sign up and start using everything listed above in under 2 minutes.