v1.0.0 is Live

Product Updates

New features, improvements, and fixes for the Wirevox platform.

June 27, 2026
v2.0

Wirevox Platform Update (v2.0)

Massive feature rollout including a complete billing overhaul, agent versioning, real-time sync, and agency enhancements.

💳 Billing & Subscriptions Overhaul

Feature Billing

A complete rebuild of our payment and ledger system to ensure accuracy, atomicity, and robust subscription handling.

  • Ledger-Based Billing: Transitioned to a highly accurate usage ledger, replacing the legacy estimated cost metrics.
  • Automated Dunning: Introduced multi-stage automated dunning and suspension workflows for past-due accounts.
  • Accurate Call Costs: Resolved pricing staleness for test calls. Exact costs and duration (minutes/seconds) are now finalized and displayed in call logs at call completion.
  • Stripe Webhook Enhancements: Rebuilt our webhook parsing to flawlessly process add-ons (concurrent calls, extra numbers) and handle invoice changes dynamically.
  • Closed Credit Loopholes: Deployed a database migration to secure atomic credit deductions and patch tracking exploits.

🤖 Agent Versioning & Management

Feature UX

Safely iterate on your AI agents without impacting live traffic.

  • Draft & Publish Flow: Introduced a new Version History panel. Live calls and webhooks run on published snapshots, while you make edits securely in a draft state.
  • 1-Click Duplication: Added a quick duplication tool for existing AI agents.
  • Playground Upgrades: Standardized UI dropdowns, added reliable Start/Stop controls for testing, and fixed configuration state bugs during active test calls.

🏢 Agency & Whitelabel Layer

Feature Architecture

Strengthened our multi-tenant infrastructure for agency use-cases.

  • Unified Workspace Members: Refactored workspace member models and restored associated automated tests.
  • Agency Tiers & Subaccounts: Strictly gated agency features to the highest plan tier. Improved the generation of agency slugs and database provisioning for subaccounts.
  • Team Roles Refinement: Updated permission roles within team settings and eliminated redirect loop bugs for agency accounts.

🎨 UI/UX & Quality of Life

UX Feature

A darker, cleaner, and instantly responsive dashboard experience.

  • Native Dark Mode: Purged hardcoded colors to deliver a flawless dark mode across all interfaces, alongside a redesigned Voice Selector sidebar.
  • Real-time WebSocket Sync: The dashboard now updates instantly via WebSockets, eliminating the need for manual refreshes.
  • Global Notifications: Rolled out a centralized toast notification system for better user feedback.
  • Visual Polish: Added user profile avatars, revamped core modals, and introduced safety guardrails for phone number assignment.

🛠️ Architecture & Maintenance

Fix Architecture

Under-the-hood stability improvements and technical debt cleanup.

  • Voice Pipeline Standardization: Replaced outdated prototype scripts with a tracked, formal VoicePipeline architecture.
  • Codebase Organization: Cleaned up repository structure, fixed relative imports, and removed legacy prototyping files.
  • Crash Resolution: Fixed a webhook router syntax issue that previously caused intermittent server crashes.
June 22, 2026
v1.1.0

Wirevox Platform Update (v1.1.0)

RAG context buffer, latency circuit breakers, and Gladia V2 STT Engine integration.

🧠 RAG & Core Voice Architecture

Feature Architecture

The most recent upgrades to the Knowledge Base and LLM retrieval pipeline.

  • Semantic Phrase Protection: Added a specialized regex layer to protect time-sensitive phrases (e.g., "right now", "no refund") from being incorrectly stripped as conversational filler during vector DB queries.
  • Sliding RAG Context Buffer: Implemented a FIFO memory buffer that retains the last 3 Knowledge Base retrievals. This allows the AI to successfully answer vague follow-up questions without failing the DB search.
  • Latency Circuit Breaker: Wrapped vector database queries in a strict 1500ms timeout. If the database spikes or hangs, the pipeline auto-aborts the search and allows the AI to answer using existing baseline memory, preventing dead silence.

🎙️ Speech-to-Text & Telephony Core

Improvement Telephony

Upgrades to how the pipeline handles human speech and interruptions.

  • Gladia V2 STT Engine: Successfully integrated and shipped the Gladia V2 Live API as an alternative to Deepgram.
  • Semantic Barge-In Filtering: Implemented dynamic confidence scaling, echo cancellation, and filler handling in the speech pipeline. The AI now ignores background noise and conversational fillers ("mhm", "yeah") but instantly stops speaking when the caller intentionally interrupts.
  • Bilingual TTS Voice Switching: Hardened the language protocol to enforce strict 2-letter language tags (e.g., [EN], [FR]) when switching between dual primary/secondary languages on the fly. Prevented language selection from overriding the STT engine rules.

Ready to build your
AI workforce?

Stop missing calls and start scaling your conversations today.

Start Your Free Trial Talk to Sales