Deploy AI voice agents on Asterisk phone systems

v7.5.6 is an in-place feature and reliability release. It does not migrate
databases, reassign Agents, or change Audio Profiles.

• Outbound lead context is delivered or the call fails closed - ARI
  origination now uses its documented variables object, restoring routing,
  identity, AudioSocket, AMD/consent, and campaign metadata. Nonempty lead
  custom_vars is bounded, confirmed before provider startup, recovered after
  an engine restart, and redacted from diagnostics
  (#613).
• Hangup intent markers can be scoped per Agent - each Agent can inherit,
  extend, or replace the global end-of-call phrases. New calls capture an
  immutable policy, and Full Local negotiates call-scoped support so older
  servers and malformed overrides fail closed
  (#619).
• Post-call summaries can use configured modular LLMs - each webhook can
  select an enabled LLM provider and configure its model readiness, timeout,
  word limit, and prompt. Explicit selections never fall back to another
  provider; summary failures leave {summary} empty while webhook delivery
  continues. Existing webhooks retain the legacy OpenAI behavior until
  configured (#618).
• Summary prompts stay isolated from the live Agent persona - provider
  adapters receive the webhook's summary instructions as authoritative job
  context, and the default Groq LLM moves to openai/gpt-oss-120b.

See the v7.5.6 changelog,
migration notes, and
validation matrix.

v7.5.5 is an in-place feature release. It does not migrate databases, reassign
Agents, or change Audio Profiles.

• Collapsible Admin UI sidebar - the left navigation can collapse to an
  icon-only rail, with hover tooltips and a persisted preference, reclaiming
  space on smaller displays (#596).
• Pre-call variables now flow into post-call webhooks - each pre-call
  output variable is exposed as its own placeholder in webhook payload
  templates, matching prompt and in-call tool behavior
  (#608).
• Configurable extension availability mapping - check_extension_status
  now classifies multiple ARI device states per extension (including
  operator-configured custom states for DND/away) via a configurable
  free/busy/unavailable mapping, with a fail-closed default and new Admin UI
  editors for per-extension and global state mapping
  (#577).
• check_extension_status reliability fixes - availability fields
  survive JSON sanitization across all tool adapters, live-transfer channel
  activity is cross-checked against stale device state, and an unmapped
  device_state_id can no longer bypass restrict_to_configured_extensions
  (#577).

See the v7.5.5 changelog,
migration notes, and
validation matrix.

v7.5.4 is an in-place reliability and privacy release. It does not migrate
databases, reassign Agents, or change Audio Profiles.

• Diagnostics are truly opt-in - disabled playback taps and full-call RCA
  capture perform no per-call conversion, locking, file creation, write, or
  cleanup deletion. Enabled paths reject symlinks, unsafe writable ancestors,
  and foreign ownership before audio is written.
• ARI silent failures recover promptly - 10-second WebSocket ping and timeout
  defaults make readiness fail in about 20 seconds before normal reconnect logic
  takes over.
• Deepgram telephony choices are coherent - Flux and Nova receive the right
  language fields, the UI exposes the commonly used telephony models and seven
  end-to-end Aura languages, and incompatible language/model/voice combinations
  fail before a remote session opens.
• Docker status is accurate again - Docker SDK 7.1 restores Admin UI socket
  compatibility and detection follows rootful, rootless, TCP, and named-pipe
  endpoint configuration.
• Updates preserve optional Local AI state - absent and unselected Local AI
  stays absent, while an installed stopped service can be refreshed without
  being started, including rollback.
• Call History privacy is operator-controlled - strict, routing-visible, and
  explicit off modes make the redaction boundary visible without rewriting
  historical records (#589).

See the v7.5.4 changelog,
migration notes, and
validation matrix.

v7.5.3 focuses on getting an installation back to a known-good configuration
without undoing the operator's unrelated work.

• Restore audio defaults in context - Providers, Audio Profiles, and
  modular Pipelines each expose their own restore action in the Admin UI.
  Provider restores keep credentials, models, voices, prompts, enabled state,
  and provider identity; profile restores keep Agent assignments; pipeline
  restores keep STT/LLM/TTS provider selections and non-audio options.
• Backend-owned baselines - restore values come from the same canonical
  registry used by validation, including the supported OpenAI Realtime GA
  linear16/24 kHz contract. Environment-owned overrides remain visible and
  are never silently rewritten.
• Explicit apply guidance - each restore reports whether no action, a hot
  reload, or an AI Engine restart is needed before new calls use the baseline.
• Fail-closed dialplan transfers - extension, queue, and ring-group
  transfers validate known-missing targets, require a confirmed ARI handoff,
  and preserve ownership safely when Asterisk's response is indeterminate.
  FreePBX queues use the standard ext-queues context by default (#577).
• Query what the agent actually did - completed in-call tools now expose a
  stable tool_call_id, normalized success/failure status, action, and
  reconcilable target_id in Call History and its API without mixing telemetry
  into the transcript (#587).

These recovery actions are intentionally narrow: they do not provide a global
factory reset and do not change secrets or Agent routing.

See the v7.5.3 changelog for implementation and
compatibility details.

v7.5.2 adds a call-scoped wideband path without changing existing Agent
profiles or the established 8 kHz compatibility defaults.

• Native 16 kHz AudioSocket - assign wideband_pcm_16k to an Agent to use
  Asterisk slin16 and rate-specific AudioSocket framing in both directions.
• Provider and pipeline alignment - Grok, Google Live, Deepgram, OpenAI,
  ElevenLabs, Local Hybrid, and Full Local retain truthful per-call media
  contracts, including retries, tool continuations, interruption, and cleanup.
• Fail-closed compatibility - wideband requires Asterisk 20.17+, 21.12+,
  22.7+, or 23.1+ and a genuinely wideband endpoint or SIP trunk path such as
  G.722. ExternalMedia RTP and PSTN/G.711 calls remain on an 8 kHz profile.
• Simple rollback - switch the Agent back to telephony_ulaw_8k or
  telephony_enhanced_8k; no global transport or provider-default change is
  required.

See the v7.5.2 changelog,
v7.5.2 migration notes,
and v7.5.2 validation matrix.

The v7.5.1 hotfix focuses on recovery and observability without changing audio
profiles, provider transport, or fresh-install defaults.

• Recoverable Apply Changes - the Admin UI prepares its updater runner
  before touching a live service and restores the previous image and container
  environment if a Compose replacement fails or does not become healthy.
• Complete realtime transcripts - OpenAI and Grok keep assistant transcript
  state separate from interleaved caller-final events, preventing clipped
  prefixes in Call History and post-call consumers.
• Apply instead of unnecessary restart - tool-only edits advertise and use
  hot reload for new calls. Provider, environment, and process-level changes
  remain on the restart/recreate path.

No database migration or audio-profile reassignment is required. Existing
stored transcripts are not rewritten.

See the v7.5.1 changelog and
v7.5.1 migration notes.

v7.5.0 improves narrowband call audio without changing the established
8 kHz Asterisk wire contract, and adds a production-oriented VICIdial Remote
Agent integration.

• Opt-in enhanced telephony audio - assign telephony_enhanced_8k to an
  Agent to use stateful band-limited downsampling for cleaner G.711 playback.
  Existing profiles keep their compatibility behavior, and switching back to
  telephony_ulaw_8k is the immediate rollback.
• Consistent provider and pipeline policy - hosted providers and modular TTS
  pipelines inherit the Agent's Audio Profile by default, expose narrow
  troubleshooting overrides, and validate incompatible encoding, rate,
  resampler, overlap, and segmentation combinations before apply.
• Safer interruption and teardown - resampler state is isolated per call and
  reset across responses, interruptions, and cleanup; replaced streams cannot
  be removed by stale cleanup; and late pipeline output is blocked after call
  teardown takes ownership.
• VICIdial Remote Agents - VICIdial remains authoritative for campaigns,
  customer channels, reporting, dispositions, DNC, callbacks, and transfers,
  while AAVA supplies the mapped AI Agent with fail-closed ownership checks and
  sanitized lifecycle evidence.
• More recoverable upgrades - the host recovery script handles mixed Git
  ownership, stale updater images, /root traversal constraints, and tracked
  local edits while preserving bounded backups and exact release targeting.

See the v7.5.0 changelog,
Audio Profiles, and
VICIdial Remote Agent setup for details.

Outbound campaigns are easier to prepare, safer to schedule, and much easier
to troubleshoot from the Admin UI.

• Simpler lead intake - import validated CSV or Excel .xlsx files, or add
  individual leads manually. Samples and new campaigns use the canonical
  AI_AGENT/agent routing model while legacy AI_CONTEXT/context inputs
  remain compatible.
• Safer campaign scheduling - scheduled calls consistently receive the lead's
  called number, malformed timezone or calling-window settings fail closed,
  campaign concurrency is counted correctly, and stale attempts recover through
  one validated timeout policy.
• More reliable human handling - human-first AMD defaults reduce false
  voicemail classification, and terminal farewell/hangup handling prevents new
  caller input from reviving a call that is already ending.
• Better HTTP-tool workflows - pre-call, in-call, and post-call HTTP tools
  enforce method/body compatibility; pre-call output variables remain available
  for enriched greetings; and bounded, sanitized tool responses and diagnostics
  are visible in Call History and Scheduling.
• Safer upgrades - updater recovery now handles older Git installations,
  Docker Compose access after privilege drops, and mixed-ownership checkouts more
  predictably without sacrificing local tracked changes.

See the Outbound Calling guide and
v7.4.1 changelog for details.

Each Agent can now receive only the transfer destinations, calendars, and
voicemail mailboxes it should be allowed to use.

• Per-Agent resource access - configure the global inventory on Tools, then
  choose Inherit, Selected, or None under Agents → Edit Agent → Tools
  for the transfer family, Google Calendar, Microsoft Calendar, and voicemail.
• One enforced call snapshot - provider schemas, prompt guidance, execution,
  deferred transfers, and audit metadata all use the same effective resource set.
  Empty or stale selections fail closed, and a globally disabled tool always wins.
• Restart-free tool updates - Tools → Save & Apply validates and publishes a
  new tool generation for new calls. Active calls keep the generation they started
  with; a failed build leaves the previous generation running.
• Contexts retired - runtime persona routing now reads Agents from agents.db.
  Legacy YAML Contexts are imported atomically on upgrade, and AI_CONTEXT remains
  a deprecated compatibility alias while dialplans move to AI_AGENT.
• Cleaner first run - empty installations start with Receptionist, Sales, and
  Support instead of a collection of demonstration Contexts.
• Call History compatibility - tool names remain google_calendar,
  microsoft_calendar, and leave_voicemail, so existing filters and reports keep
  working.

Before upgrading-especially from v7.3.0-v7.3.3-read the
current upgrade procedure
and Contexts → Agents migration guide.

Callers no longer wait through silent provider or pipeline startup.

• Per-agent ringback control - enable Play ringback while connecting in
  the Agents UI; tone:ring is supplied as the default repeating Asterisk tone.
• One implementation for every call path - full-agent providers and modular
  pipelines share the same caller-only lifecycle, without sending setup audio to
  the AI provider.
• Clean audio handoff - ringback stops on the first provider or pipeline
  greeting audio and is also cleared on no-greeting readiness, startup failure,
  disconnect, or call cleanup.
• Safe and opt-in - existing agents remain unchanged until the setting is
  enabled. YAML/API users may configure an Asterisk-local tone:, sound:, or
  recording: media URI.

See Connection Audio / Ringback
and the v7.3.5 changelog.

v7.3.3 is a Local-AI-only stabilization release. It adds no providers and keeps
the cloud-provider call paths unchanged.

• Calls are isolated by session - agent prompts and conversation state no
  longer mutate shared Local AI Server configuration or leak across reused
  WebSocket connections. AI Engine and Local AI Server should be upgraded
  together; the legacy unscoped switch remains temporarily compatible.
• Barge-in abandons interrupted output - late LLM/TTS work is quarantined,
  the interrupted exchange is removed from weak-model history, and the
  replacement turn stays focused on what the caller just said.
• Farewells finish exactly once - Local hangup_call speaks the selected
  Kokoro/Piper/etc. farewell without a second LLM rewrite, drains partial
  AudioSocket or RTP tails, records agent_hangup, and then disconnects.
• CPU/GPU deployment is safer - dependency pins, CUDA/cuDNN validation,
  optional llama.cpp architecture targeting, and idempotent preflight checks
  reduce first-build and rerun failures.
• Community GPU evidence - Tesla V100S testing passed Faster-Whisper CUDA
  float16, Llama 3.1 8B Q4_K_M, Kokoro, AudioSocket, ExternalMedia, barge-in,
  terminal hangup, concurrent session isolation, and restart recovery.

See the Local AI community test matrix and the
Unreleased changelog for the complete scope.

v7.3.2 is a stabilization-only patch release built from the supervised
AudioSocket and ExternalMedia validation cycle.

• No new providers - scope is limited to reliability, deployment safety,
  documentation, and contributor-facing CI.
• Grok ExternalMedia repaired - clean barge-in, cancelled-output quarantine,
  named-instance runtime inheritance, complete replacement turns, and exact
  inactivity announcements through xAI force_message.
• AudioSocket and modular pipelines hardened - terminal playback, pipeline
  producer ownership, talk-detect echo, and inactivity-grace regressions are
  covered by focused tests and supervised calls.
• Updater and provider-failure recovery hardened - safer ownership,
  rollback/stash handling, readiness validation, and an opt-in dialplan redirect.
• PR quality gates expanded - Admin backend/frontend checks and CLI
  cross-compilation now run before merge.

Release evidence and remaining gates are tracked in the
v7.3.2 validation matrix.

AVA now protects silent calls and finishes every terminal message before disconnecting.

• 30-second inbound inactivity protection by default - AVA asks “Are you still there?”, waits 15 seconds for a reply, then speaks a configurable final warning and ends the call. Outbound agents remain opt-in.
• The agent keeps its configured voice - check-ins and final warnings are synthesized by the active Google Live, OpenAI Realtime, Grok, Deepgram, ElevenLabs, local full-agent, or pipeline voice.
• Transport-safe hangup - watchdog and hangup_call farewells drain AudioSocket or ExternalMedia/RTP streaming buffers and ARI file playback before ARI disconnects the caller. Fixed sleeps no longer clip long final sentences.
• Deepgram and ElevenLabs lifecycle fixes - Deepgram control frames no longer split greetings, and ElevenLabs response-completion plus hosted-silence handling keeps AVA's watchdog authoritative.
• Global and per-agent controls - configure defaults under Advanced Settings → Voice Activity Detection → Caller Inactivity, then optionally override them per agent. Call History labels watchdog endings as No input timeout.

See Caller inactivity configuration, ElevenLabs setup, and the full v7.3.1 changelog.

Voice now belongs to agents. Configure one provider, create multiple agents that share it - each with its own voice.

• Provider-aware voice picker in the Agent form: a dropdown of OpenAI's 10 GA voices, suggestions + custom clone IDs for Grok, Google Live's 30 prebuilt voices, Deepgram's Aura models - the control adapts to the agent's selected AI Engine.
• Provider-specific safety - the provider-level voice becomes the default voice; agents without one behave exactly as before. OpenAI and Google log and fall back for unknown values. Deepgram preserves a configured Aura value for review but fails the call before connection when the voice is unknown or its language does not match the Deepgram Agent language.
• Observable - every call logs the resolved voice and its source, and Call History shows "Voice: marin (from agent)" per call.
• Agent voice changes apply instantly - no engine restart.

Thanks @foytech for seeding this feature (#497). Full guide: docs/VOICE_SELECTION.md.

Real-time system status for the Admin UI - pushed, not polled.

• Live-status hub - a single /api/live-status snapshot endpoint plus an SSE stream (/api/live-status/stream) aggregates AI Engine health, Local AI connectivity, active sessions, audio directories, platform checks, and Asterisk ARI into one normalized status feed.
• Push-first - ai_engine and local_ai_server push their own readiness to the Admin UI (POST /api/live-status/publish, authenticated with LIVE_STATUS_PUSH_TOKEN), so the dashboard converges in sub-second time after a restart instead of waiting on staggered polls. Legacy /api/system/* probes remain as fallback/enrichment.
• Configurable - LIVE_STATUS_POLL_INTERVAL_SECONDS (default 30 s, min 2 s) and LIVE_STATUS_INITIAL_PROBE_TIMEOUT_SECONDS (default 2 s), read live from .env.

Full notes in CHANGELOG.md.

A focused quality release across the Admin UI - no call-path changes.

• Dashboard reliability - the Asterisk status pill no longer flaps on a transient ARI blip: it reads the engine's authoritative, reconnect-supervised ARI state and applies hysteresis. The system endpoints the Dashboard polls every 5s no longer block the admin event loop, the heaviest is TTL-cached, polling backs off on errors, failed polls surface in the error banner, and a single bad poll no longer flashes cards to "Loading…".
• No more "Loading configuration…" flash - ~11 config pages now seed from a shared stale-while-revalidate cache of the config document, so revisiting a settings page is instant.
• Accessibility (WCAG AA) - form labels programmatically associated with inputs, a focus-trapping modal, a navigation landmark + "skip to content" link, accessible names on icon-only buttons, non-colour status cues on the topology, a visible dark-mode toggle on-state, and light-mode contrast fixes. Debug console.logs (including one that leaked the auth token to the browser console) were removed.
• Prompt editor - configured tool names are colour-coded by their in-call status (enabled / global / not-enabled) as you type.
• Fix (#436) - a canonical google_live: { type: full } provider can be edited and saved again.

Full notes in CHANGELOG.md.

The biggest release yet: manage your AI agents from the Admin UI, not a config file.

• 🤖 Agents tab - create, edit, and manage agents in the UI. Start from a template (receptionist, after-hours, appointment booker, and more), set the prompt and provider, and copy a ready-to-paste dialplan snippet.
• 📊 Multi-agent dashboard - live KPIs (active agents, active calls, calls routed, transfers), per-agent stats, and routing breakdowns at a glance.
• ☎️ New AI_AGENT dialplan variable - route a call to an agent by name. Your existing AI_CONTEXT dialplans keep working unchanged.
• 🔄 Automatic migration - your existing contexts move into a local agents database on first start. Back up agents.db before later major-version upgrades; see the operator migration guide for rollback boundaries.
• 🔒 Security hardening - no more admin/admin: a one-time admin password is generated and must be changed at first login. Config exports no longer bundle your .env by default.

⚠️ Major release - please read the Upgrade Notes before upgrading from 6.x.

Follow-up to the v6.5.3 hotfix. v6.5.3 only flipped config/ai-agent.yaml; v6.5.4 brings the rest of the codebase in line:

• Pydantic defaults in src/config.py now default to api_version: ga + model: gpt-realtime (so fresh wizard installs are correct).
• Admin UI "Add Provider" template for OpenAI Realtime no longer seeds the sunset preview model.
• Model dropdown removes the 5 sunset preview options and adds 3 new GA models - gpt-realtime-1.5 (best audio-in/audio-out quality), gpt-realtime-2 (reasoning voice model, GPT-5-class), and gpt-realtime-mini (cost-optimized) - alongside the existing gpt-realtime.
• Legacy preview values in operator YAML now render in a "Custom (legacy - will not connect)" optgroup with a yellow warning banner above the form so the broken state is visible without silently swapping the operator's config.
• Engine emits a one-shot warning when api_version: beta is detected in config (exactly once per provider lifetime, not per reconnect attempt).
• Docs: full rewrite of docs/Provider-OpenAI-Setup.md model section + fix to docs/TROUBLESHOOTING_GUIDE.md.

OpenAI sunset the Realtime Beta API on 2026-05-12 and removed the gpt-4o-realtime-preview-2024-12-17 model on 2026-05-07. Shipped config/ai-agent.yaml still pinned api_version: beta + that preview model, so every operator using OpenAI Realtime hit error.code: beta_api_shape_disabled and the WebSocket closed immediately. Two-line config flip - no code change required. The provider's GA wire-protocol path has shipped since v6.0.0; v6.5.3 just makes it the default everyone gets:

• api_version: ga (was beta)
• model: gpt-realtime (was gpt-4o-realtime-preview-2024-12-17)

If you have an ai-agent.local.yaml that explicitly pins api_version: beta, remove the override or change it to ga. Refs: OpenAI deprecations, gpt-realtime.

🆕 xAI Grok Voice Agent realtime provider (NEW, v6.5.2)

• Fifth full-agent realtime provider - structurally parallel to OpenAI Realtime and Google Live, built on a multi-instance foundation from day one
• μ-law @ 8 kHz caller input with no input resampling; observed xAI output is PCM16 @ 24 kHz and AAVA converts it to the configured Asterisk transport format
• Five named voices (eve, ara, rex, sal, leo) plus custom voice ID free-text for cloned voices
• Custom function-tools identical to OpenAI Realtime; xAI-native tools (web_search, x_search, file_search, mcp) accepted via YAML extra_tools escape hatch
• Conservative long-session warning at 28 minutes for compatibility with older xAI limits; xAI's current Voice Agent model page lists a 120-minute maximum session
• Setup guide: docs/Provider-Grok-Setup.md

🏢 Multi-instance full-agent providers (NEW, v6.5.2)

• Run multiple instances of the same full-agent provider type with isolated credentials (e.g. acme_google_live + globex_google_live both using type: google_live)
• Per-instance credential files at /app/project/secrets/providers/<provider_key>/{api-key,agent-id,vertex-json} - the new per-provider Vertex upload path does NOT mutate .env
• Route via AI_PROVIDER, an Agent's provider selection plus AI_AGENT, or DID-based dispatch with Asterisk Gosub
• Setup guide: docs/Multi-Instance-Full-Agent-Providers.md
• Breaking for multi-instance setups: short aliases AI_PROVIDER=openai, AI_PROVIDER=google, provider: deepgram_agent now fail validation - use exact provider instance keys instead. Single-instance setups using the canonical block names are unaffected.

🎛 Admin UI polish (v6.5.2)

• Uniform per-instance credentials paste-style uploader across all full-agent provider forms (Grok, OpenAI Realtime, Deepgram, Google Live, ElevenLabs Agent)
• EnvPage adds a new "Per-Instance Provider Credentials" status section so operators can audit credential file presence without SSH
• Dashboard System Topology rebuilt: tri-state per-component health with 2-strike debounce (transient probe blips no longer flip dots red), responsive provider grid, multi-instance sub-rows grouped by provider type, Asterisk + AI Engine cards stretched to match Providers height
• Backend probe timeouts bumped (ai_engine 1.5s → 5s; local_ai_server 2.5s → 5s) to stop legitimate localhost probes timing out under load
• ~260 inline help tooltips backfilled across provider forms, Setup Wizard, and System pages - new HelpTooltip is viewport-aware (flips placement to keep popovers visible in scrolled modals)

📞 Call recordings (v6.5.2)

• Browser playback for compact .ulaw recordings (Asterisk's 8 kHz μ-law output, ~10× smaller than PCM WAV) via server-side audioop.ulaw2lin WAV wrapping - no transcode dependency
• Uppercase .WAV, compressed WAV, and .gsm recordings transcode via sox; AAVA_RECORDING_TRANSCODE_TIMEOUT_SEC env var (default 120s) governs the timeout

Previously in v6.5.1

• 💻 CPU-demo profile end-to-end - Faster-Whisper tiny.en + Piper + Qwen 0.5B wired through the Admin UI; runtime Device/Compute selectors with CPU/float16 gating; Filler Audio and LLM/TTS Overlap runtime toggles
• 🛡️ Local provider hot-path hardening - send_audio() no longer blocks on per-frame reconnect; asyncio.Lock serializes _reconnect() against _send_loop's on-ConnectionClosed path
• 🎨 Faster-Whisper verify path tolerates the runtime CUDA→CPU fallback so working CPU/int8 configurations no longer get rolled back as "verification failed"

Previously in v6.5.0

• 🔧 Local LLM tool-gated response (#368) - new WS protocol message types tool_context / tool_result v2; per-WebSocket fail-closed sync prevents cross-call ACL/policy/prompt leakage on reused connections
• ☁️ Gemini 3.1 Flash Live verified compatible (no engine changes); Vertex AI mode is the production answer for #351 barge-in
• 🎤 Deepgram Flux v2 + nova-3 default flip; Admin UI surfaces "Flux Turn-Detection Tuning" panel for flux-* models
• 🩺 Admin UI HTTP-tool-test guard now reads .env first so Environment-page edits to AAVA_HTTP_TOOL_TEST_* take effect without a container restart (#370)

For older releases, expand Previous Versions below. Full release notes in CHANGELOG.md.

v6.4.2 - Microsoft Calendar V1 + Google Calendar overhaul

• 🗓️ Microsoft Calendar - Outlook / Microsoft 365 integration via device-code OAuth, Graph free/busy, legacy per-context account binding, Tools UI Connect/Verify/Disconnect (migrated to per-Agent resource access in v7.4)
• 📅 Google Calendar - multi-account / legacy per-context binding (#338), JSON upload + auto-discover, Domain-Wide Delegation, native free/busy mode (migrated to per-Agent resource access in v7.4)
• 🎯 Reschedule reliability - server-side event_id resolution + 400/404 fallback eliminates LLM-id-hallucination duplicate bookings
• 🔧 Date/time prompt placeholders ({today}, {current_date}, etc.) so models stop reasoning with stale years
• OpenAI Realtime duplicate-events fix (per-response_id async-event gating); per-context tool_overrides now actually take effect on OpenAI Realtime / Deepgram / Google Live; Google Live 30-voice catalog (#349)

v6.4.1 - CPU Latency Optimization

• ⚡ Streaming LLM→TTS overlap - sentence-boundary token streaming, sub-2s perceived latency on pipelines
• Pipeline filler audio (instant "One moment please" acknowledgment) configurable via Admin UI
• Qwen 2.5-1.5B Instruct recommended for CPU; ~15-30 tok/s vs Phi-3's ~0.8 tok/s
• Direct PCM→µ-law conversion in all 5 TTS backends (10-50ms saved per response)
• Preflight hardening - Buildx detection, RAM/disk/network checks, GPU install gated behind --apply-fixes

v6.4.0 - Attended Transfer & Russian Speech

• 📞 Attended transfer with three screening modes: basic_tts, ai_briefing, caller_recording
• ExternalMedia RTP streaming delivery; provider-agnostic transfer-target tool guidance
• 🗣️ Russian speech backends: Sherpa Offline STT (VAD-gated), T-one STT, Silero TTS (multi-language)
• 🎧 Admin UI: fullscreen dashboard panels, per-message conversation timestamps, JSONPath [*] HTTP-tool wildcards

v6.3.2 - Azure Speech & MiniMax LLM

• Microsoft Azure Speech Service STT & TTS pipeline adapters (REST batch, WebSocket streaming, SSML)
• MiniMax LLM M2.7 via OpenAI-compatible API with tool-calling
• Call Recording Playback in Admin UI Call Details modal
• Azure SSRF prevention, PII logging discipline, input validation hardening

v6.3.1 - Local AI Server & Guardrails

• Backend enable/rebuild flow, model lifecycle UX, GPU ergonomics, CPU-first onboarding
• Structured local tool gateway, hangup guardrails, tool-call parsing robustness
• agent check --local / --remote CLI verification

v6.1.1 - Operator Config & Live Agent Transfer

• Operator config overrides (ai-agent.local.yaml), live agent transfer tool
• Experimental ViciDial community-tested configuration notes, Asterisk config discovery in Admin UI
• OpenAI Realtime GA API, Email system overhaul, NAT/GPU support

v5.3.1 - Phase Tools & Stability

• Pre-call HTTP lookups, in-call HTTP tools, and post-call webhooks (Milestone 24)
• Deepgram Voice Agent language configuration
• ExternalMedia RTP greeting cutoff fix

v4.4.3 - Cross-Platform Support

• 🌍 Pre-flight Script: System compatibility checker with auto-fix mode.
• 🔧 Admin UI Fixes: Models page, providers page, dashboard improvements.
• 🛠️ Developer Experience: Code splitting, ESLint + Prettier.

v4.4.2 - Local AI Enhancements

• 🎤 New STT Backends: Kroko ASR, Sherpa-ONNX.
• 🔊 Kokoro TTS: High-quality neural TTS.
• 🔄 Model Management: Dynamic backend switching from Dashboard.
• 📚 Documentation: LOCAL_ONLY_SETUP.md guide.

v4.4.1 - Admin UI

• 🖥️ Admin UI: Modern web interface (http://localhost:3003).
• 🎙️ ElevenLabs Conversational AI: Premium voice quality provider.
• 🎵 Background Music: Ambient music during AI calls.

v4.3 - Complete Tool Support & Documentation

• 🔧 Complete Tool Support: Works across ALL pipeline types.
• 📚 Documentation Overhaul: Reorganized structure.
• 💬 Discord Community: Official server integration.

v4.2 - Google Live API & Enhanced Setup

• 🤖 Google Live API: Gemini 2.0 Flash integration.
• 🚀 Interactive Setup: agent init wizard (agent quickstart remains available for backward compatibility).

v4.1 - Tool Calling & Agent CLI

• 🔧 Tool Calling System: Transfer calls, send emails.
• 🩺 Agent CLI Tools: doctor, troubleshoot, demo.