Top Voice Features in Ace AI for Follow Up Boss

What is voice chat in Ace AI and how is it different from text chat?

Ace offers two ways to interact with your CRM assistant: text chat (available on Regular and Pro plans) and voice chat (Pro-only). The underlying AI and CRM data access are identical in both modes — the difference is the interaction modality.

In text mode, you type a question into the Ace chat interface embedded in Follow Up Boss, and the AI responds in text. In voice mode, you speak through your microphone, the AI responds through your speaker in a natural conversational voice, and the session runs over a real-time WebRTC connection. You can interrupt, ask follow-up questions, and hold a back-and-forth conversation exactly as you would with a colleague on the phone.

Voice mode is designed for agents who are between appointments, in the car, or otherwise occupied and cannot type. It brings the full power of the Ace assistant into a hands-free workflow.

Who can use voice chat in Ace AI?

Voice chat requires a Pro seat. This is enforced at two levels in the code:

• The voiceChatEnabled flag is set to true only when a user's planType is 'pro' (embed.js, line 2647)
• The voice token endpoint on the companion returns a 403 pro_required error for any seat that is not Pro (companion.js, line 778)

Regular and Free plan users see the voice interface as locked with an upgrade prompt. Admins who have purchased Pro seats for their team can enable voice for individual agents. See the pricing page for current plan details.

How does the voice session work technically?

When a Pro agent starts a voice session in Ace:

1. The client requests a short-lived voice session token from the server (30-minute expiry, HMAC-signed).
2. The server verifies Pro status and generates a conversation token from ElevenLabs Conversational AI.
3. The client connects directly to ElevenLabs via WebRTC (LiveKit), using the token for authentication.
4. The AI assistant receives the current contact context (Ace Score, history summary, memory facts, Ace fields) before the session starts, so it already knows who you are talking about.
5. Voice is streamed bidirectionally in real time. The AI uses Voice Activity Detection to know when the agent has finished speaking before responding.

The server-side ElevenLabs API key is never exposed to the browser — the token approach keeps credentials secure while allowing the browser to connect directly for low-latency voice.

What voice IDs can agents choose in Ace AI?

Agents can choose their preferred AI voice for text-to-speech output. The available voice IDs are: onyx, alloy, echo, fable, nova, and shimmer. The default voice is onyx.

To change your voice preference: go to your user settings in the Ace interface and select from the voice options. The preference is saved to your user profile and persists across sessions. Each voice has a distinct character — onyx is deeper and more measured; nova is higher-pitched and conversational; alloy sits in the middle. Choose based on what feels natural for a working conversation.

What can you do in a voice session with Ace AI?

Because voice chat has the same CRM access as text chat, anything you can ask in text, you can ask by speaking. Common use cases in a voice session:

• Contact briefing before a call. "Brief me on Sarah Johnson before I call her." The AI reads from the contact's Ace Score, recent activity, notes, and memory facts and gives you a spoken summary — without you touching your keyboard.
• Next-action guidance. "What should I do next with this lead?" The AI considers the contact's Ace Tier, days since inbound, preferred channel, and recent activity and suggests a specific action.
• Draft a follow-up message. "Draft a text to send to Marcus after today's showing." The AI uses the contact's history and your conversation to draft a personalized follow-up that you can review and send.
• Pipeline queries. "How many of my active buyers have not been contacted in the last week?" The AI queries your live FUB data and reports the count and names spoken back to you.
• Compliance check. "Is there anything in this draft that could be a Fair Housing issue?" The AI scans the draft text and flags any protected-class language before you send.
• Lead status update. "Mark this contact as Active Buyer and note that we toured three properties today." The AI executes the CRM write-back and confirms it verbally.
• General CRM questions. "When did I last talk to this person?" or "What's their Ace Score?" — answered instantly by voice.

How does Ace AI voice compare to typing for CRM tasks?

Voice is most valuable in mobile scenarios — driving between showings, walking a property, or in any situation where typing is inconvenient. Text remains preferable when you need a written output you can copy directly, like a draft email or a formatted summary.

Does voice chat work on mobile devices?

Voice chat is supported on mobile browsers (iOS Safari and Android Chrome) with some important caveats. The WebRTC connection requires microphone permission granted at the browser level. On iOS, Safari must be the browser — other iOS browsers do not have WebRTC microphone access. Network changes (switching from WiFi to cellular mid-session) may interrupt the voice session and require reconnecting.

The Ace companion popup is the recommended entry point for voice on mobile, accessed via the FUB embed. Make sure your device's microphone is enabled for the FUB site in your browser settings before your first session.

What plan do you need and how do you enable voice in Ace AI?

1. Ensure your seat is Pro. Admin users who have purchased Pro seats see the voice button unlocked. If you see a lock icon on the voice button, contact your team admin to upgrade your seat to Pro.
2. Choose your voice preference. In user settings, pick from onyx, alloy, echo, fable, nova, or shimmer. The default is onyx.
3. Open the Ace companion on a contact. The voice session is most useful when opened on a specific contact — the AI pre-loads that contact's context before you speak.
4. Click the microphone / voice button. The session initializes, requests mic permission if not already granted, and connects you to the AI voice agent within a few seconds.
5. Speak naturally. The AI uses voice activity detection to know when you've finished a thought before it responds. You do not need to say "Hey Ace" or use any wake word — just speak.

For more on how Ace works with FUB data generally, see our ultimate guide to AI voice and chat tools and the Agentic page for how the 215-tool MCP surface extends Ace to Claude and ChatGPT. For compliance considerations specific to voice, see our compliance overview.