Ultimate Guide to Voice and Video Features in Follow Up Boss
Follow Up Boss includes built-in calling and text through its native dialer integration. The Ace AI layer adds a WebRTC voice chat assistant (Pro plan required) powered by ElevenLabs and LiveKit that runs directly inside FUB — so agents can dictate notes, talk through contact context, and log interactions hands-free without leaving the CRM.
What voice features does Follow Up Boss actually include?
Follow Up Boss ships with a built-in calling integration that lets agents dial contacts, record calls, and log call outcomes directly to a contact record. It also supports text messaging through its inbox. What FUB does not include is an AI-powered voice assistant — a real-time voice agent that can answer questions about a lead, write notes by voice, or coach you through a conversation while you are on a call.
That gap is where the Ace AI layer comes in. Ace adds a conversational voice assistant that lives inside Follow Up Boss rather than alongside it.
How does the Ace voice chat assistant work?
Ace voice chat uses a WebRTC connection via the ElevenLabs Conversational AI SDK over a LiveKit room. Here is the technical path, verified in the source code:
- The agent clicks the voice button inside the FUB embed. The server mints a short-lived voice session token and renders the voice popup.
- The browser calls
/api/voice/session, which exchanges the token for a per-session conversation token from the ElevenLabs API — the API key stays server-side and is never exposed to the client. (chat-app/routes/voiceConvai.js:75) - The browser SDK opens a direct WebRTC connection to an ElevenLabs LiveKit room using the conversation token. (
chat-app/public/js/voice-convai.js:317) - Voice activity detection (VAD) using Silero ONNX models (
silero_vad_legacy.onnx/silero_vad_v5.onnx) determines when the agent is speaking so the assistant knows when to respond. (chat-app/index.js:3913) - The conversation is contextualized with the open contact record and the agent's account data, so Ace can answer questions like "when did I last call this person?" without the agent having to look it up.
The agent can also select a preferred AI voice — the system supports six voices (onyx, alloy, echo, fable, nova, shimmer), configurable per user. (chat-app/routes/embed.js:3347)
What plan do you need for voice chat in Ace?
Voice chat is a Pro plan feature. When an account admin upgrades a user to Pro, the system sets voiceChatEnabled: true on that user's Firestore record. If the user is downgraded back to Regular, the flag is set to false — and if the webhook is delayed, a self-healing check at the /voice-chat route revokes access automatically. (chat-app/routes/embed.js:2647, chat-app/index.js:3111–3134)
| Plan | AI Chat | Voice Chat | AI Suggestions |
|---|---|---|---|
| Free | Limited | No | No |
| Regular | Yes | No | Yes |
| Pro | Yes | Yes | Yes |
Plan tiers in the codebase are free, regular, and pro — the normalized identifiers used throughout the system. (chat-app/routes/embed.js:2595)
Does Ace support video features inside Follow Up Boss?
No. As of June 2026, Ace does not include any built-in video calling, video messaging, or screen-share functionality. The voice layer is audio-only. If you need video communication with clients, you would use a separate tool (your phone, Zoom, or Google Meet) and log the outcome back to FUB manually or via an integration.
This is worth stating plainly because "voice and video" is a common search phrase in the real estate CRM space. The answer for Ace is: voice yes, in-CRM video no.
What can I actually do with voice inside Follow Up Boss?
Once you are on Ace Pro and have the FUB embed installed, voice lets you do several things that would otherwise require you to stop, type, and navigate:
- Ask about a lead by voice. "When was the last time someone reached out to this contact?" or "What stage is this deal in?" — Ace has the contact record in context and can answer without you opening another tab.
- Dictate call notes. After a call, speak your notes aloud. Ace transcribes and logs them.
- Work hands-free on mobile. The voice interface is designed to work on mobile browsers, not just desktop — important for agents driving between showings.
- Choose your AI voice. Six voice options (onyx, alloy, echo, fable, nova, shimmer) let agents pick a tone that feels natural for their workflow.
- Get speed-to-lead coaching. Ace can check whether a portal lead is still within the critical 5-minute response window via the
lead_speed_to_lead_checkMCP tool, which is accessible from the same session. (mcp-server/src/zillowTools.ts:377) See our Zillow Playbook for the full workflow.
How does voice connect to the rest of Ace Trove?
Voice is one surface of Ace, not a standalone product. Underneath it, the same Ace Trove that powers written chat is available during a voice session. That means:
- ACE scoring data — engagement, intent, recency, quality, velocity, and conversion probability scores — is available as context when you ask about a contact. (
chat-app/config/aceIntelligenceConfig.js:194–199) - The agentic tools (200+ MCP tools exposing FUB data to Claude and ChatGPT) are the same tools the voice session can invoke.
- Pipeline health and lead nurture checks (
pipeline-health-checkatmcp-server/src/index.ts:4198,lead-nurture-optimizeratindex.ts:4230) can surface insights mid-call. - Fair Housing compliance scanning runs on any message drafted through Ace, including drafts triggered by voice. The
scanForComplianceViolations()function (chat-app/utils/complianceGuard.js:293) checks text against protected-class language rules before anything is sent. See our Compliance guide for details.
How do I set up voice chat in Follow Up Boss with Ace?
- Install the Ace embed in Follow Up Boss. The embed is a URL integration that places Ace inside the FUB sidebar. You configure it once per account in the FUB admin settings.
- Upgrade the user to Pro. An account admin navigates to user management in the Ace admin dashboard and upgrades the agent's seat to Pro. The system sets
voiceChatEnabled: trueautomatically. - Open a contact record in FUB. The Ace panel appears in the sidebar. Click the microphone button to start a voice session.
- Grant microphone permission. The browser will prompt for microphone access on the first session. This is a one-time step per browser.
- Start talking. Ace connects via WebRTC within a few seconds. VAD handles turn-taking — you do not need to press a button to speak and release.
- Select your preferred voice (optional). In Ace settings, choose from the six available voice options. The preference is saved to your user profile.
Does voice work on mobile browsers?
Yes, with caveats. The voice interface is built to handle mobile browser constraints — the code includes specific handling for network changes (a common failure point when switching between cell and WiFi on a phone) and microphone permission dialogs on iOS and Android.
The VAD ONNX models are served via a dedicated route with explicit MIME-type headers so they load correctly across browsers including mobile WebKit. (chat-app/index.js:3725)
In practice, performance depends on your phone's browser support for WebRTC and the stability of your connection. Desktop or laptop use tends to be more reliable for extended voice sessions.
How does the Ace Trove tier affect voice access?
Voice chat (per-seat Pro plan) and the Ace Trove (account-level subscription) are two separate billing dimensions. You can have Pro voice without Ace Trove, and you can have Ace Trove without upgrading individual agents to Pro.
Ace Trove is billed by total contact count across the account's FUB database, with six flat monthly tiers:
| Tier | Contact cap | Monthly price |
|---|---|---|
| Starter | Up to 5,000 | $49 |
| Growth | Up to 20,000 | $99 |
| Team | Up to 50,000 | $199 |
| Brokerage | Up to 100,000 | $349 |
| Scale | Up to 250,000 | $549 |
| Enterprise | Up to 500,000 | $899 |
Tiers auto-upgrade when your contact count crosses the next threshold (with a 10% buffer and a 7-day notice period). (chat-app/config/accountAiBillingConfig.js:64–70)
What is the difference between voice via Ace and FUB's built-in calling?
These are two different things and they complement each other rather than compete.
| Capability | FUB native dialer | Ace voice chat |
|---|---|---|
| Dial a lead's phone number | Yes (via integrations like Kixie, CINC) | No |
| AI assistant for agent questions | No | Yes |
| Dictate notes by voice | No | Yes |
| Contact context awareness | Limited (call log) | Full ACE scoring context |
| Real-time VAD (no push-to-talk) | N/A | Yes (Silero ONNX) |
| Fair Housing compliance check | No | Yes (on drafted messages) |
Frequently asked questions
Can Ace voice chat dial a lead's phone number?
No. Ace voice chat is an AI conversation assistant — it does not connect to the public telephone network. For outbound dialing, you use FUB's native calling integrations (Kixie, CINC phone, etc.). Ace voice chat is for the agent-to-AI conversation that happens before, during, or after a call to a lead.
Is the conversation recorded or stored?
The voice session connects your browser directly to ElevenLabs via a WebRTC/LiveKit room. The server does not record or store the audio stream. The conversation token is per-session and cannot be reused.
What happens to voice access if I downgrade from Pro to Regular?
When a Pro user is downgraded, the system sets voiceChatEnabled: false on the user record. A Stripe webhook enforces this at the billing period end. If the webhook is delayed, the voice-chat route self-heals by checking downgradeEffectiveAt on every session request. Access is revoked without requiring manual intervention. (chat-app/index.js:3116–3134)
Does Ace connect to Claude or ChatGPT for voice sessions?
Ace exposes Follow Up Boss data to Claude and ChatGPT via the MCP (Model Context Protocol) connector — that is the agentic workflow layer. The voice assistant itself runs through ElevenLabs Conversational AI. These are complementary surfaces: you can use MCP tools to analyze your database in Claude, and use the in-CRM voice assistant for real-time hands-free work in FUB.
How does voice help with Fair Housing compliance?
Voice does not bypass the compliance layer. Any message or note drafted through Ace — including content generated in response to a voice prompt — is subject to the same scanForComplianceViolations() check that covers written chat. Protected-class language is flagged before the message reaches the agent's outbox. See the full Fair Housing compliance guide for what the scanner detects.
Try Follow Up Ace in your Follow Up Boss
Free to start, no sales call. Connect Follow Up Boss in one click and Ace works inside your CRM.
Get Started Free