AI Voice & Chat
How-to guides from the Revamp365 Help Center.
How to Build AI Voice Campaigns for Real Estate | Revamp365.ai
Outbound calling is the lifeblood of real estate wholesaling and investing. With Revamp365.ai, you can automate your entire cold calling and follow-up process using the AI Agent Campaign Builder. This tool allows you to deploy your custom AI voice agents to call hundreds of seller leads automatically, qualifying them and booking appointments directly to your calendar.
This guide will walk you through the 5-step Campaign Builder, including crucial best practices for list management and phone number health.
Step 1: Basic Information
The first step is to name your campaign and select the AI agent that will be making the calls.
- Navigate to AI Voice & Chat > Campaigns.
- Click the Create Campaign button.
- Campaign Name: Give your campaign a clear, descriptive name (e.g., "May 2026 - Absentee Owners").
- Select AI Voice Agent: Choose the specific agent you built (from a template or from scratch) that is trained for this specific list.
Step 2: Select Recipients (The "Continue Calling" Strategy)
This step determines who your AI agent will call. You can filter your CRM contacts by Tags, Lead Status, or Lead Source.
Crucial Best Practice: The "Continue Calling" Tag
If you intend to run this campaign more than once (which is highly recommended for outbound lists), you must manage your list carefully so you don't repeatedly call leads who have already been qualified or disqualified.
- Before creating the campaign: Go to your Contacts CRM and tag the entire list of leads you want to call with the tag "Continue Calling".
- In Step 2 of the Campaign Builder: Under the Tags dropdown, select the "Continue Calling" tag (along with any other tags that segment your list, like "Absentee Owner").
- How it works: When you use the Default Call Flow (see Step 4), the system is pre-configured to automatically remove the "Continue Calling" tag from any lead that the AI speaks to and either qualifies or disqualifies.
- Rerunning the campaign: When you duplicate and rerun this campaign next week, the AI will only call the leads who still have the "Continue Calling" tag (i.e., the ones who didn't answer previously).
Step 3: Calling Schedule & Limits
Real estate outbound calling must respect legal calling hours and phone number reputation limits.
- Select Days and Times: Use the grid to select the exact hours your AI agent is allowed to make calls (e.g., Monday-Friday, 9:00 AM to 5:00 PM). The AI will automatically pause calling outside of these windows.
- Timezone: Ensure the timezone matches your target market.
- Max Calls Per Hour: You can throttle the speed of the campaign here.
Crucial Best Practice: Phone Number Health
To prevent your phone numbers from being flagged as "Spam Risk" by carriers, we strongly recommend a maximum of 150 to 200 calls per phone number, per day. If your campaign has 1,000 recipients, you should either spread the campaign over 5 days or purchase a Number Pool with 5 different phone numbers to distribute the call volume.
Step 4: Call Flow (Automated Follow-Up Rules)
The Call Flow dictates what the system does after a call ends, based on the outcome (Call Disposition).
- The Default Flow: We highly recommend selecting the Default Flow from the "Saved Rule Sets" dropdown unless you are an advanced user who wants to customize the flow entirely on your own.
- What the Default Flow does:
- No Answer: Automatically schedules a retry call (e.g., Retry in 4 hours, maximum of 3 attempts).
- Qualified/Disqualified: Automatically removes the "Continue Calling" tag (as mentioned in Step 2) so they aren't called again if you rerun the list.
- Callback Requested: Pauses the campaign for that specific lead and schedules a callback for the requested time.
- Visual Builder: You can click the Visual tab to see a flowchart representation of your rules.
Step 5: Settings & Launch
The final step allows you to review any custom rules and launch the campaign.
- Review Rules with AI: Click this button to have our AI Copilot scan your Call Flow for any logical errors, unreachable rules, or CRM sync gaps before you launch.
- Click Create Campaign to set it live. The AI will immediately begin dialing based on your Step 3 schedule.
Frequently Asked Questions (FAQ)
How many calls can an AI voice agent make per day?
While the AI itself can make unlimited concurrent calls, carrier restrictions apply to the phone numbers. To keep your phone numbers from getting burned or marked as spam, the recommended maximum is 150-200 calls per number per day.
How do I stop the AI from calling someone who already answered?
Use the "Continue Calling" tag strategy. Tag your initial list with "Continue Calling". The Default Call Flow will automatically remove this tag once a lead is spoken to. When you rerun the campaign targeting that tag, the answered leads will be excluded.
Can I change the schedule after the campaign starts?
Yes. You can pause an active campaign at any time, adjust the calling schedule or max calls per hour, and then resume it.
Quick Reference: Campaign Best Practices
| Strategy |
Recommendation |
Why it Matters |
| Call Volume Limits |
Max 150-200 calls per number / day |
Prevents carriers from flagging your numbers as "Spam Risk". |
| List Management |
Use the "Continue Calling" tag |
Ensures you don't repeatedly call leads who have already been qualified or disqualified. |
| Call Flow Rules |
Use the "Default Flow" |
Automatically handles retries for unanswered calls and manages the tag removal process seamlessly. |
How to Build an AI Voice Agent for Real Estate (Templates) | Revamp365.ai
Speed to lead is everything in real estate investing. Revamp365.ai provides pre-built AI voice agent templates designed specifically for real estate use cases. These templates come pre-configured with the right prompts, logic, and settings to handle common scenarios like seller follow-ups or inbound receptionist duties, allowing you to automate your lead qualification instantly.
This guide will walk you through creating a powerful real estate AI agent using a template.
Step 1: Navigate to the AI Voice Dashboard
- From the main Revamp365.ai dashboard, click on AI Voice & Chat in the left navigation menu.
- Click on the Agents tab at the top of the page to view your existing agents.
- Click the Create an Agent button in the top right corner.
- Select Voice Agent from the dropdown menu.
Step 2: Select a Real Estate AI Template
The Create Voice Agent modal will appear, displaying several options tailored for investors and wholesalers.
- Browse the available templates. Popular options include:
- Seller Lead Follow-Up: Follow up with seller leads to qualify motivation, timeline, and book appointments.
- Inbound Seller Lead: Answer inbound seller calls, discover lead source, and qualify through the full acquisition process.
- RE Business Receptionist: Professional inbound receptionist that routes sellers, buyers, and general inquiries.
- Click on the template that best fits your needs (e.g., Seller Lead Follow-Up).
- Click the Create button at the bottom of the modal.
Step 3: Review the Pre-Built Conversational Flow
Once created, you will be taken to the Agent Editor. The template automatically populates the canvas with a complete conversational flow designed to convert real estate leads.
- Review the Nodes: The canvas displays the logical steps of the conversation (e.g., Start/Identify Lead, Property Confirmation, Mini Intro).
- Examine Prompts: Click on any node to see the specific script the AI will use at that stage of the conversation.
- Check Transitions: Notice how the nodes are connected. These transitions dictate how the AI moves from one topic to the next based on the seller's responses (e.g., "Confirmed Identity", "Not Interested").
Step 4: Configure Your AI Agent Settings
The right-hand panel contains the global settings for your new agent. While the template provides a strong foundation, you should review and adjust these settings to match your specific wholesaling or investing business.
- Agent Name: Give your agent a recognizable name at the top of the panel.
- Voice: Select the voice profile you want the agent to use.
- LLM Model: Choose the underlying AI model (e.g., a leading AI model).
- Execution Mode: Templates typically use Flex Mode, allowing the agent to act flexibly based on the flow, or Rigid Mode for strict step-by-step adherence.
- Post Call Analysis: Review the data extraction settings to ensure the agent captures the right information from every call (e.g., Call Disposition, Interest Level, Call Summary).
Step 5: Test and Publish Your Real Estate AI
Before deploying your agent to call live seller leads, it's crucial to test it.
- Click the Test Agent tab in the right panel to simulate a conversation and ensure the flow works as expected.
- Once you are satisfied with the agent's performance and settings, click the Publish button at the bottom of the settings panel to make the agent active.
Frequently Asked Questions (FAQ)
What is an AI voice agent for real estate?
An AI voice agent is an automated system that can have intelligent, two-way voice conversations with your potential and current clients. In real estate, they are used to qualify seller leads, answer inbound calls, and book appointments 24/7.
Can I customize the pre-built templates?
Yes. Once you load a template in Revamp365.ai, you have full access to the Agent Editor. You can change the scripts, add new logic nodes, or adjust the post-call analysis fields to fit your exact business model.
How does the AI know if a lead is qualified?
The templates include "Post Call Analysis" fields that instruct the AI to evaluate the conversation and categorize the lead based on their motivation, timeline, and interest level.
Field Reference: Post Call Analysis
Templates often include pre-configured Post Call Analysis fields to help you quickly sort leads. Here are common examples:
| Field Name |
Type |
Description |
| call_disposition |
Selector |
Categorizes the outcome of the call (e.g., Qualified - Interested, Wrong Number, Do Not Call). |
| interest_level |
Selector |
Classifies the contact's interest level (Hot, Warm, Cold) based on explicit statements. |
| call_summary |
Text |
A structured summary of the call, including owner details, motivation, timeline, and price expectations. |
| spoke_with_owner |
Yes/No |
Indicates whether the actual property owner or decision-maker participated in the conversation. |
Build a Custom AI Voice Agent from Scratch | Revamp365.ai
For complete control over your AI's conversational flow, Revamp365.ai allows you to build a custom AI voice agent entirely from scratch. This "Start from blank" option is ideal for unique real estate use cases, highly specific wholesaling scripts, or complex lead qualification processes that don't fit standard templates.
This guide will show you how to initialize a blank agent and begin building your custom flow using our intuitive node-based editor.
Step 1: Navigate to the AI Voice Dashboard
- From the main Revamp365.ai dashboard, click on AI Voice & Chat in the left navigation menu.
- Click on the Agents tab at the top of the page.
- Click the Create an Agent button in the top right corner.
- Select Voice Agent from the dropdown menu.
Step 2: Select "Start from blank"
The Create Voice Agent modal will appear, showing various real estate templates.
- Locate the Start from blank option (usually the first option on the top left).
- Click on the Start from blank card to select it.
- Click the Create button at the bottom of the modal.
Step 3: Understanding the Blank Canvas
You will be taken to the Agent Editor. Unlike templates, the canvas will be mostly empty, containing only a single "Begin" node and an initial "Conversation" node.
- The Begin Node: This is the starting point of every call your AI agent makes or receives.
- The Initial Conversation Node: This node contains the very first thing your agent will say when the call connects (e.g., "Hi, is this [Name]?").
Step 4: Building Your Custom Conversational Flow
To build your custom real estate AI agent, you will add nodes and connect them with transitions to guide the conversation.
- Add Nodes: Use the left-hand panel to drag and drop different types of nodes onto the canvas.
- Conversation: For standard dialogue and prompts.
- Extract Variable: To capture specific information from the seller (e.g., a property address or asking price).
- Logic Split: To branch the conversation based on conditions (e.g., "If asking price > $500k, go to Node A; else go to Node B").
- Call Transfer: To route the call to a human acquisitions manager.
- Configure Nodes: Click on a node to open its settings. Here you will define the Prompt (what the agent says) and the Transitions (how the agent moves to the next node based on the user's response).
- Connect Nodes: Draw lines between the transitions of one node to the input of the next node to create the conversational path.
Step 5: Configure Global AI Settings
Once your flow is built, configure the agent's overall behavior in the right-hand panel.
- Agent Name: Give your custom agent a name (e.g., "Cash Buyer Outreach Bot").
- Voice & LLM: Select the voice profile and the underlying AI model.
- Execution Mode: Choose between Flex Mode (allows the agent to handle off-script questions more naturally) or Rigid Mode (forces strict adherence to your node structure).
- Global Prompt: Enter overarching instructions that the agent should follow throughout the entire call, regardless of which node it is currently on (e.g., "Always maintain a polite, professional tone and never guarantee a specific purchase price.").
Step 6: Test and Publish Your Custom Agent
- Use the Test Agent tab in the right panel to simulate calls and verify that your custom logic and transitions work correctly.
- Once testing is complete, click Publish to make your custom agent live and ready to call leads.
Frequently Asked Questions (FAQ)
Why should I build an AI voice agent from scratch instead of using a template?
Building from scratch gives you absolute control over the conversation. It's best for investors who have highly refined, proprietary sales scripts or complex qualification criteria that require specific logic splits and variable extraction.
Can I transfer a call from the AI agent to a live person?
Yes. You can use the Call Transfer node in the builder to seamlessly hand off a qualified lead to your acquisitions team or a live receptionist.
What is the difference between Flex Mode and Rigid Mode?
Flex Mode allows the AI to handle unexpected questions or tangents from the lead while still trying to guide them back to the main flow. Rigid Mode forces the AI to stick strictly to the script and node structure you defined, which is useful for highly structured surveys or intake forms.
Node Types Reference
| Node Type |
Category |
Description |
| Conversation |
Core |
The primary node for dialogue. Contains the prompt the agent will speak and the transitions based on user replies. |
| Extract Variable |
Core |
Designed specifically to listen for and save a piece of data (like a date, number, or name) into a variable. |
| Logic Split |
Logic |
Evaluates a condition (e.g., "Is the property value > $100k?") and routes the conversation down different paths. |
| Call Transfer |
Transfer |
Initiates a transfer of the live call to a designated phone number (e.g., a human sales rep). |
| End Node |
Control |
Terminates the call gracefully. |
AI Assisted Voice Agent Builder for Real Estate | Revamp365.ai
The fastest way to create a custom AI voice agent on Revamp365.ai is using the AI Assisted Builder. Instead of manually connecting nodes and writing scripts, you simply describe what you want the agent to do in plain English, and the AI generates the entire conversational flow for you.
This guide explains how to use the AI Builder to generate a custom agent draft tailored for real estate investing and wholesaling.
Step 1: Access the AI Assisted Builder
- From the main Revamp365.ai dashboard, navigate to AI Voice & Chat > Agents.
- Click the Create an Agent button and select Voice Agent.
- In the template selection modal, click on the AI Assisted Builder card (usually located at the bottom left).
- Click Create.
Step 2: Describe Your Real Estate Agent's Goal
You will be taken to Step 1 of the AI Builder process: Describe.
- In the text box provided, clearly state the primary objective of your voice agent.
- Example: "Follow up with seller leads to qualify their motivation and timeline, confirm property ownership, gauge interest in a cash offer, and book an appointment if they are qualified and interested."
- Be as specific as possible about the desired outcome of the call (e.g., booking an appointment, transferring a call, gathering property details).
- Click Continue.
Step 3: Clarify the Details
In Step 2: Clarify, the builder asks specific questions to tailor the agent's behavior and knowledge to your real estate business. Fill out the form completely.
- Agent Persona: Enter the agent's name and your company name (e.g., "Sarah / Direct Homebuyers").
- Tone: Select the desired conversational style from the dropdown (e.g., "Friendly & conversational", "Professional & direct").
- Qualification Criteria: Define exactly what makes a lead "qualified" for your business (e.g., "Confirmed ownership, motivated to sell within 6 months").
- Cash Offer Pitch: Provide the brief pitch or key points the agent should use when discussing an offer (e.g., "We buy houses as-is for cash, close in 7 days, no repairs needed").
- Objection Handling: List common objections (e.g., "Not interested", "Price too low") and how the agent should respond to them.
- Booking Action: Select what should happen when a lead qualifies (e.g., "Book via calendar link", "Live transfer to agent").
- Compliance: Enter any required legal or compliance language (e.g., "This call may be recorded...").
Step 4: Build the Agent Draft
Once all fields in the Clarify step are filled out:
- Scroll to the bottom of the page.
- Click the Build agent draft button.
Step 5: Review and Refine Your AI Agent
The AI will process your inputs and generate a complete node-based conversational flow.
- You will be taken to the Agent Editor, where you can see the generated nodes and transitions.
- Review the Flow: Click through the generated nodes to ensure the prompts and logic align with your expectations.
- Make Adjustments: You can manually edit any prompt, add new nodes, or change transitions just as you would if you had built the agent from scratch.
- Test and Publish: Use the Test Agent tab to simulate a call. Once satisfied, click Publish to make your AI-generated agent live.
Frequently Asked Questions (FAQ)
How does the AI Assisted Builder work?
The AI Assisted Builder uses advanced language models to interpret your plain-English description and structured inputs (like your pitch and objection handling) to automatically generate a complete, node-based conversational flow in the Agent Editor.
Can I edit the agent after the AI builds it?
Absolutely. The AI Builder creates a draft. Once generated, you have full access to the Agent Editor to tweak prompts, add logic splits, or change the flow entirely.
What are the best practices for the AI Builder?
Be as specific as possible in Step 1 (Describe). In Step 2 (Clarify), take the time to list out your most common objections and exactly how you want the AI to handle them. This makes the resulting agent much more robust and capable of handling real-world real estate conversations.
Tips for the AI Builder
- Be Specific in Step 1: The better your initial description, the more accurate the generated flow will be.
- Comprehensive Objection Handling: Take the time to list out your most common objections in Step 2. This makes the resulting agent much more robust and capable of handling real-world conversations.
- Always Review: The AI Builder creates a draft. Always review the generated prompts and test the flow before publishing to ensure it meets your quality standards.
Place a Test Call to an AI Voice Agent
The agent builder has a built-in test panel that places a live web call to your AI voice agent straight from the browser, so you can hear the agent and read its responses before it ever touches a real lead. The call uses your microphone and your published agent version.
Where the test-call control lives
Open the agent builder at /agent and select the agent you want to test. Inside any node sidebar (the Begin, Default, End, or node-settings panel) is a test strip with three tabs:
- Test Audio — the live web call (voice agents only).
- Test Chat — a text-only conversation with the agent.
- Train Agent — coaching tools, shown only on Pro+ and Dealmaker plans.
Click Test Audio, then press Start Test Call.
Prerequisites before you call
A few things must be true or the call will not connect:
| Requirement |
Why it matters |
An agent_id in the page URL |
The button reads it from the URL. With none present it shows No Agent ID in URL and stays disabled. |
| A published agent version |
The call forces your latest published version, not draft edits. Publish first or unsaved changes will not be exercised. |
| Microphone permission |
The panel calls getUserMedia before connecting. Approve the browser prompt when it appears. |
| A secure (HTTPS) origin |
Browsers block the mic on plain HTTP, which surfaces as a microphone error. |
Note the in-panel reminder: call transfer is not supported in Webcall. To test a warm transfer, place an outbound phone call instead.
What happens during the call
- The button changes to Call Active... and a red End Call button appears.
- The status line walks through
Connecting..., then Connected.
- Speak into your mic. The live transcript builds below, labeling each turn Agent: or You:.
- For conversation-flow agents, the canvas pans to and highlights the node the call is currently on, following the conversation node by node as the agent advances.
Press End Call to hang up. The call also tears down automatically if you switch to the Test Chat or Train tab, navigate away from the builder, or close the tab, so a session never keeps running unseen.
Reviewing the result
The transcript in the panel is your immediate record of how the agent handled the conversation. The status line reports the final state (Call ended) or any Error: message if the connection dropped. For the full saved recording, summary, and structured transcript, open AI call transcripts, recordings, and summaries.
Free-plan limits
On the Free plan, test calls are capped at three minutes and run a credit pre-check first — if your balance cannot cover a worst-case call, an upgrade or top-up prompt appears before our voice AI engine is contacted. Free users can also place an outbound phone test call (the Test Audio web call is the default); that path is limited to a set number of test calls per day.
Next steps
Once the agent sounds right, assign it to an inbound number so live callers reach it. If you have not provisioned a number yet, provision a phone number for SMS and AI voice first.
Assign an AI Voice Agent to Answer an Inbound Number
Inbound routing in Revamp365 connects a phone number to the AI voice agent that picks up when someone calls it. You set this on the Phone Numbers tab inside AI Voice & Chat, and it requires a published agent plus a number that is linked to AI Voice.
Prerequisites
Three things must exist before a call can reach an agent:
| Requirement |
Where it lives |
How to confirm |
| A number linked to AI Voice |
Phone Numbers tab |
The number appears under My Phone Numbers with a monthly cost |
| A published voice agent |
Agents tab |
Open the agent and click Publish in its Global Settings tab |
| Pro+ or Dealmaker plan |
Account plan |
The Call Flows tab and the Inbound flow column are visible |
Call Flows and the inbound-flow control are gated to Pro+ and Dealmaker. Free-plan accounts use a shared test number for outbound test calls only and cannot route inbound calls.
Open the Phone Numbers tab
- In the left sidebar, click AI Voice & Chat.
- Click the Phone Numbers tab.
- Find the number under My Phone Numbers. If you have none, search by a 3-digit area code and click Purchase, or link an existing SMS number — every Revamp365 number is voice + SMS capable.
Assign the agent through an inbound flow
The current routing model points a number at a call flow that contains one AI Agent step. The agent inside that flow answers the call.
- In the number's row, open the Inbound flow dropdown (it lists — None — plus every call flow you have built).
- Pick the call flow whose AI Agent step uses the agent you want.
- The selection saves immediately. Behind the scenes, Revamp365 binds that flow's agent to the number with our voice AI engine and updates the number's voice webhook.
To build or edit a flow first, use the Call Flows tab. A flow needs exactly one AI Agent step — if it has more than one, the system asks you to open the flow editor and choose which step the agent belongs to.
Legacy direct-to-agent routing
Some numbers route straight to an agent without a flow. These show an amber Legacy: routes directly to agent badge under the Inbound flow column. To move a legacy number onto the flow-based model, click Switch to Call Flow in that row.
After you assign
- Place a test call to confirm the agent answers. See place a test call to your AI voice agent.
- If you have not bought or linked a number yet, start with provision a phone number for SMS and AI voice.
Troubleshooting
- No Inbound flow column or Call Flows tab — your plan is below Pro+. Upgrade to Pro+ or Dealmaker.
- The agent never picks up — confirm the agent is published, and that the flow you selected actually contains that agent's step. To check publish state on the Agents tab, open the Columns dropdown and enable the Published column; published agents show a green Published badge (drafts show Draft).
- Dropdown only shows "— None —" — you have no call flows yet. Build one on the Call Flows tab, then return and select it.
Provision a Phone Number for SMS and AI Voice
Every phone number in Revamp365 is a single carrier number that works for both SMS texting and AI Voice calling — you buy it once and it is wired for both. Numbers are managed on the AI Voice & Chat page under the Phone Numbers tab.
Where Numbers Are Managed
Open AI Voice & Chat from the left sidebar and click the Phone Numbers tab, or go straight to /ai-voice?tab=numbers. The page header reads "Phone Numbers — Purchase and manage phone numbers for your AI agents." Three stat cards at the top show Total Numbers, Active, and Monthly Cost.
Search for an Available Number
In the Search Available Numbers card:
- Enter a 3-digit Area Code (the field requires exactly 3 digits, e.g.
415).
- Click Search.
- Matching numbers appear as cards in a "Found N numbers" grid. Try a different area code if you see "No results found."
Search only returns US local numbers that are SMS-capable, so anything you buy can text and call.
Buy the Number
Click the green Purchase button on a result card. On success you get a "Purchased and linked to AI Voice" toast, and the number drops out of the search list and into your table. Behind one click, Revamp365:
- Buys the number from our telecom carrier and wires up its voice webhook.
- Auto-links it to AI Voice so it can run voice agents immediately.
- Auto-enrolls it for SMS if you already have a verified MARKETING A2P campaign — that step adds the number to your messaging sender pool, which is what enables texting.
The default cost is about $3.45/mo per number, shown in the Monthly Cost column.
What a Number Is Used For
The My Phone Numbers table is where each number is configured. Key columns:
| Column |
Purpose |
| Nickname |
A friendly label (pencil icon to edit inline) |
| Inbound flow |
Route inbound calls to a call flow (Pro+/Dealmaker) |
| Lead Tracking Campaign |
Attribute calls/texts to a campaign |
| Caller ID |
CNAM + Branded Calling (Pro+/Dealmaker) |
| SMS (A2P) |
Texting registration state |
The SMS (A2P) column shows Registering…, Enabled, Failed (SMS), or Pending review. When you have a verified campaign, an Attach to [campaign] action enables texting on that number. See activate texting and the Trust Center checklist to get A2P approved first.
Edge Cases and Limits
- Free plan: No number to buy — outbound test calls use a shared test number, capped at 25 test calls per day (3 minutes each), billed against your credit balance. Upgrade to Pro+ to own a number.
- Isolated accounts must purchase numbers through our voice AI engine directly; the in-app buy returns a permission error.
- Already-bought SMS number: purchasing it again simply links it to AI Voice instead of charging twice.
- Release a number: click the trash icon in the Actions column and confirm — this removes it from our telecom carrier and your account.
Once a number is provisioned, assign an AI Voice agent to it to start answering inbound calls.
Read AI Call Transcripts, Recordings, Summaries, and Post-Call Analysis
Every AI voice call your Revamp365 agent places or answers is logged with a transcript, an audio recording, a conversation summary, and structured post-call analysis fields. This page shows you where that history lives and what each field on the call-detail screen means.
Where call history lives
Open the standalone Voice AI Logs page (its heading reads "Revamp 365 AI Call Logs"), or go to AI Voice & Chat and click the Call History tab. Both render the same unified log, which combines AI calls, human-dialer calls, and Zoom meeting transcripts into one table.
By default the table shows these columns: Date, Call ID, Agent Name, Call Type, Status, AI Summary, Sentiment, Disposition, Interest, Duration, Cost, Contact Name, Property Address, plus Automations and Actions. More columns — Campaign Name, From/To Number, Tags, Follow-up Notes, and any post-call analysis fields — are available through the column picker. Use the search box to match across the transcript text, phone numbers, and call ID. Archived calls are hidden by default; the sidebar Archived view surfaces them. Calls are never deleted, because each row carries billing and usage data.
Click any row to open the full call-detail dialog.
Reading the transcript and recording
The detail dialog opens on the Transcription tab by default. AI calls render a turn-by-turn chat view (your agent's bubbles on the right, the prospect on the left) with per-line timestamps. If only a flat transcript exists, you see the plain text instead — and a "Whisper transcription — no speaker labels" note appears when the transcript came from the recorded-audio fallback rather than the real-time stream.
The Call Recording panel in the left sidebar gives you an inline audio player plus a Download link. Human-dialer recordings are streamed through Revamp365 so your browser never needs carrier credentials.
The AI Keywords tab highlights terms you care about (for example "appointment" or "cash offer") wherever they appear in the transcript, with a match counter. Add or remove keywords in that tab's settings and choose whether to scan both speakers or the prospect only.
Post-call analysis fields
The left sidebar and summary pane surface the structured analysis our voice AI engine returns after the call:
| Field |
What it shows |
| Conversation Summary |
AI-written recap of the call (your custom summary is preferred over the default summary) |
| Sentiment |
Prospect sentiment: positive, negative, or neutral |
| Call Successful |
Whether the call met its goal (Yes / No) |
| In Voicemail |
Whether the call hit voicemail |
| Disconnection |
Why the call ended (for example user_hangup) |
| Call Disposition |
Outcome label from your post-call config |
| Interest Level |
Seller interest captured by the agent |
| Price Node Reached |
Whether the price step was reached |
Any custom fields you defined for the agent appear under Extracted Data. The sidebar also reports Duration, Cost, the conversation balance (agent vs. user talk time), latency (end-to-end and LLM), LLM token usage, the Call ID, and the Agent ID.
Tips and limits
- List rows include the lightweight
transcript for search and export; the word-level transcript and raw payload load only when you open a specific call.
- The Actions tab lists automations the call triggered (tags, status changes, callbacks). See reading automation run logs.
- Data & Webhooks and Raw JSON tabs are admin-only.
- No log yet? Confirm your agent is live by placing a test call.
Write a System Prompt That Keeps Your AI Voice Agent On Track
Your AI voice agent's behavior starts with the Global Prompt, the system prompt that frames who the agent is, what it's trying to accomplish, and the lines it should never cross. A focused prompt is the difference between an agent that books appointments and one that rambles.
Where the Global Prompt lives
- Open AI Voice & Chat in the left sidebar, then go to the Agents tab.
- Create a new voice agent or open an existing one to load the agent builder.
- In the agent panel, find the Global Prompt box (placeholder text reads "Enter your global prompt here").
- Type your prompt, then click Publish for a new agent or Update when editing one.
The Global Prompt is sent to the voice engine as the agent's general_prompt. It applies across the whole call, on top of any per-node instructions in your conversation flow.
What to put in the prompt
Structure the prompt in clear sections so the agent can prioritize. Keep sentences short, the agent reads them literally.
| Section |
What it does |
Example |
| Role |
Names the persona and company |
"You are Kelsey, an acquisitions rep for South Bay Investment." |
| Goal |
The single call outcome |
"Confirm the seller still wants an offer and book a callback." |
| Guardrails |
Hard limits |
"Never quote a price. Never promise a closing date." |
| Tone |
How it speaks |
"Warm, concise, one question at a time." |
| Fallback |
What to do when stuck |
"If asked something you can't answer, offer to have a person follow up." |
Use variables to personalize
Insert dynamic variables with double curly braces so the agent greets contacts by name and references live CRM data:
{{first_name}} pulls the contact's first name into the opening line.
{{contact_id}} and {{user_id}} pass CRM identifiers to downstream tools and function nodes.
Set fallback values for these variables so the call doesn't break when a field is empty.
Global Prompt vs. node prompts
The Global Prompt sets the overall mission. Individual steps in your conversation flow have their own prompts with an Instruction Type toggle:
- Prompt lets the LLM phrase the step in its own words.
- Static Text speaks an exact, fixed line (useful for disclosures or the begin message).
Note: a prompted begin message on calls under 10 seconds still bills as 10 seconds, so use Static Text for the opener to avoid extra fees. Every node prompt must contain text, an empty one shows "Prompt cannot be empty" and blocks publishing.
Keep facts out of the prompt
Don't paste long policy details or pricing into the Global Prompt. Instead, attach reference material so answers stay accurate and the prompt stays lean. See add knowledge base documents to your AI agent for how to upload that content.
Test before you trust it
After you publish, place a test call to your AI voice agent and listen for drift. If the agent over-promises or wanders, tighten the guardrails section, republish, and call again. Two or three rounds usually lock in reliable behavior.
Configure Post-Call Analysis Fields for an AI Voice Agent
Post-call analysis tells your AI voice agent what structured data to pull from each conversation, from a call disposition to a custom "asking price" number. You configure these fields directly in the agent editor.
Open the Post Call Analysis section
- Go to the AI Voice surface and open the agent you want to edit (or create a new one).
- In the agent editor, find the Post Call Analysis accordion and expand it.
- Pick an Analysis Model from the dropdown. This is the model that reads the transcript and fills your fields. It defaults to a leading AI model (Free); Free-tier options are labeled
(Free).
The header reminds you that results are extracted automatically and delivered via webhooks after each call ends.
System fields you cannot change
Every agent ships with locked System fields (marked with a lock icon) that you cannot edit or remove. They keep data consistent across all your agents:
| Field |
Type |
Purpose |
call_disposition |
Selector |
One of 9 outcomes (Qualified – Interested, Do Not Call (DNC), No Answer / Voicemail, etc.) |
interest_level |
Selector |
Hot, Warm, Cold, or na |
call_summary |
Text |
Structured summary of motivation, timeline, and price |
spoke_with_owner |
Yes/No |
Whether the owner actually joined the call |
gatekeeper_screen |
Yes/No |
Whether a screener answered instead |
price_node_reached |
Yes/No |
Whether pricing was discussed |
extracted_first_name, extracted_last_name, extracted_property_address, plus phone/email/mailing fields |
Text |
Feed contact auto-updates from what the lead disclosed |
Add a custom extraction field
- Click the dashed Add button to open the Add Extraction Field dialog.
- Choose a Type:
- Text — names, summaries, free-form values (supports optional Examples)
- Yes/No — a true or false determination
- Number — a numeric value
- Selector — a fixed choice list (at least 2 options required)
- Enter a Name. It is auto-formatted to snake_case (for example,
Asking Price becomes asking_price).
- Write a clear Description of what to extract. Be specific — the model follows this instruction literally.
- For Selector fields, add each option; for Text fields, add Examples if helpful.
- Click Add (or Update when editing). Hover any custom row to reveal the edit (pencil) and remove (trash) buttons.
Fields save when you publish or update the agent. Reopen the editor anytime to see your saved items.
How fields surface later
After a call, the analysis model populates every field and the results post to your webhook. The extracted_* system fields also drive automatic contact updates from disclosed names, phones, and addresses.
When you build automation rules across multiple agents, only custom fields that match by name, type, and option list on all selected agents appear as filters. To act on the captured outcomes, set a contact's status from the result — see Set Lead Status and Temperature.
Add Knowledge Base Documents to an AI Voice Agent
A Knowledge Base gives your AI voice or chat agent a set of facts to pull from during a call, so it answers from your documents instead of guessing. You build the Knowledge Base once, then attach it to any agent.
Where to find it
- Open AI Voice & Chat from the main navigation.
- Click the Agents tab.
- Click the Knowledge Base sub-tab.
The left sidebar lists every Knowledge Base on your account. Select one to see its ID, type, and creation date in the detail panel.
Create a Knowledge Base
- Click the + button at the top of the sidebar to open the Add knowledge base dialog.
- Enter a Name (40 characters or fewer) — for example,
Property FAQs.
3. Click Add and pick a source type, or drag files onto the drop zone:
| Source |
What it does |
Limit |
| Upload files |
Pulls text from your documents |
100MB per file |
| Add text |
Paste content manually with an optional title |
— |
| Add URL |
Crawls a public page (http/https only) |
— |
- Add as many sources as you need; each appears in a removable list.
- Optionally tick Re-fetch URL content periodically (24h) to keep crawled pages fresh.
- Click Save. The Knowledge Base is stored in Voice AI and appears in the sidebar.
You need a name plus at least one file, text block, or URL before Save activates.
Supported file types
Files are processed for their text content. Use .pdf, .docx, .doc, or .txt — these extract reliably. A file with no readable text (such as a scanned image PDF) is skipped, and the save fails if nothing usable is found.
Attach it to an agent
A Knowledge Base does nothing on its own — it has to be linked to an agent.
- Open the agent you want in the builder, or create a new one.
- Find the Knowledge Base picker, click Add, and select your Knowledge Base.
- Save the agent. The selected Knowledge Bases now ground its answers.
One Knowledge Base can be reused across several agents, so a single Property FAQs set can power both a seller-outreach voice agent and a buyer chat agent.
Edit or delete
The original files, pasted text, and crawled pages are not shown back to you in Revamp365 — the panel only confirms what is stored. To change content, build a new Knowledge Base with the updated sources and attach it.
To remove one, select it, click the trash icon, and confirm Delete knowledge base?. Deletion removes it from your account and from Voice AI and cannot be undone, so detach it from any live agents first.
Tips for grounded answers
- Keep each Knowledge Base focused on one topic so retrieval stays accurate.
- Pair it with a clear system prompt — see write an AI voice agent system prompt.
- Run a test call afterward to confirm the agent quotes the right facts.
Knowledge Base management is part of the AI Voice & Chat feature set; if you do not see the Agents tab, your plan may need an upgrade.
Publish, Unpublish, Duplicate, and Archive AI Voice Agents
Once you have built an AI voice agent in Revamp365, you manage its lifecycle from one place. This covers how an agent becomes live, how to copy it, how to roll back changes, and how to remove it.
Where to manage agents
- Open AI Voice & Chat from the sidebar.
- The page opens on the Dashboard tab — click the Agents tab next to it.
- The All Agents table lists every voice agent, with Calling and SMS type badges.
Click any row to open the builder. Select rows with the checkboxes to reveal the Actions menu.
Publishing happens on save
Revamp365 does not use a separate Publish button. When you save an agent in the builder, it is published to the live calling engine automatically so your changes take effect on the next call. Both new agents and edits to existing agents are published the same way.
To see publish state, open the Columns dropdown (top right of the table) and enable the Published column:
| Badge |
Meaning |
| Published |
A live version exists and answers calls |
| Draft |
Saved locally but no live version yet |
There is no one-click "unpublish." To stop an agent from taking calls, remove its phone number assignment in the builder, or delete the agent.
Duplicate an agent
Duplicating clones the full configuration — flow, prompt, voice, and post-call analysis — into a new agent so you can iterate without touching the original.
- Select exactly one Calling agent using its checkbox.
- Open Actions and choose Duplicate.
- The copy appears named with a (Copy) suffix.
Notes:
- Duplicate is available for voice agents only, one at a time.
- The new copy starts without phone numbers assigned, so it will not take calls until you wire it up and save.
Export and back up
Use Actions → Export to JSON to download the selected agents as a .json file. This is the portable backup format; you can re-create agents later with the Import JSON button in the table header.
Roll back with version history
Each save creates a new version. Inside the agent builder, open the version history drawer to restore an earlier configuration. After you restore, the editor reloads with the restored setup, which you then save to make live again.
Remove an agent
There is no archive state for voice agents — removal is a delete:
- Select one or more agents.
- Open Actions → Delete and confirm.
Deletion is permanent, so export a JSON backup first if you might need the agent again.
Before you go live
Always place a test call to confirm tone, flow, and transfer behavior before pointing real traffic at an agent. See place a test call to an AI voice agent to verify the published version end to end.
Set Up Transfer-to-Human Rules for an AI Voice Agent
When your AI voice agent reaches the limit of what it can handle, a Call Transfer node bridges the live caller to a human's phone. You decide the destination number, whether the handoff is warm or cold, and what happens if the transfer fails.
Add the Call Transfer node
- Open the agent builder at /agent and select your conversation-flow voice agent.
- In the node palette, find the Transfer section and click Call Transfer. (The neighboring Agent Transfer hands off to another AI agent, not a person.)
- Connect the node to the point in your flow where the agent should escalate.
- Click the node to open the Call Transfer settings panel on the right.
Set the destination and transfer type
- Destination Number (required): pick a country code and type the number; the panel shows the E.164 result. To route to different people, enter a dynamic variable such as
{{seller_phone}} instead of a fixed number — the name must match a variable configured on the agent.
- Transfer Type: choose one.
| Type |
Behavior |
| Warm |
The AI introduces the caller and whispers a private brief to the human before bridging. |
| Cold |
Connects the caller directly, with no introduction. |
New nodes default to Warm. Switch to Cold when you want the caller connected straight through with no introduction.
Tune the warm-transfer handoff
Selecting Warm reveals extra controls:
- Whisper to receiving agent — a private message only the human hears. Pick Static sentence (supports variables like
{{seller_name}}) or Prompt (LLM) to have the AI generate it.
- On-hold music the caller hears while dialing (Ringtone, Relaxing, Uplifting, Silence, Ringback).
- Transfer ring duration (10–60s) and Agent / human detection timeout (10–90s) sliders.
- Checkboxes for Skip human detection, Navigate IVR before human detection, showing the caller's number to the destination, and a three-way handoff message both parties hear.
Note: agentic warm transfer (a separate AI that talks to the destination first) is not yet configurable in this editor.
Handle failures
Every transfer node has a Failure Handling block:
- Max Retries — 0 to 3 attempts.
- Failure Message — what the agent says when the transfer fails after retries, for example "I was unable to transfer your call. Let me help you instead."
The node exposes two transitions: Transfer (terminal — the call left the agent) and Failure (after retries are exhausted). Wire the Failure path back into your flow so the agent keeps helping instead of dropping the caller.
Test it on a real call
Call transfer does not run on the in-browser web call, so place an outbound phone test instead. Review the result, then keep refining the AI voice agent system prompt so the agent escalates at the right moment. Publish the agent for your transfer rules to take effect on live calls.
Transfer an Existing Phone Number to Revamp365
You can bring an existing phone number with you when you move to Revamp365, so the number on your marketing, signs and business cards keeps working. There are two different routes, and which one you take depends entirely on who you are leaving. One takes days and almost no paperwork; the other takes weeks and needs documents that match your current provider's records exactly.
Transfers are handled for you — there is no self-serve transfer screen in the app. Contact support to start one, and someone will open the request and tell you which route your number takes. Buying a brand-new number instead is self-serve: see provision a phone number for SMS and AI Voice.
Before You Do Anything Else
Do not cancel your current account, and do not let the number lapse, until we confirm the number is live on Revamp365. This is the single most common way a transfer fails, and in the worst case it is unrecoverable:
- If the number goes inactive partway through a carrier port, the request is rejected and the process restarts from the beginning.
- If you close an account on a software platform, that platform may release the number back into the general pool. At that point it is gone and cannot be retrieved.
Keep paying your current provider until the switch is confirmed. It is a few extra days of overlap, and it protects the number.
Which Path Applies to You
You do not need to know anything technical — just who sends you the bill today.
| You are currently with |
Your route |
Typical time |
| A software platform: smrtPhone, GoHighLevel, Launch Control, Smarter Contact, Roor or similar |
Fast transfer |
Days |
| A phone carrier: Verizon, AT&T, a VoIP company — anything that sends you a phone bill |
Full port |
Up to 4 weeks |
The reason for the split is that most real-estate calling and texting platforms run on the same underlying network Revamp365 uses. When that is the case, your number never actually leaves that network — ownership just moves from their account to ours, which skips the entire carrier process. If you are with a genuine carrier, the number has to be physically ported between networks.
If you are not sure which bucket your provider falls into, tell support who you are with and they will check before asking you for anything.
Fast Transfer: Coming From a Software Platform
Because the number stays on the same network, there is no Letter of Authorization, no account PIN, no service address and no copy of your bill. There is no port fee either. Both sides simply have to approve the move, so the only work is getting your current provider to cooperate.
Two Things to Request
Ask your current provider's support team for both of these:
- Their Account SID. This is the identifier of the account your number currently sits in. It begins with
AC followed by a long string of letters and numbers. Ask for "the Account SID that holds my number."
- Their written approval of the transfer. The network will not move a number between two unrelated accounts without approval from both sides. They can write in to the carrier's support directly, or reply approving it on the request we open with them copied in.
Send both to support and the move is opened the same day.
How to Keep It Moving
- Be direct with your current provider. You are changing providers and you need them to approve a number transfer. It is routine, and the slowest part of this route is almost always their support queue — worth chasing.
- Settle any outstanding balance with them first. An unpaid account is the most common reason a provider declines to release a number.
If your setup happens to run on your own carrier account rather than your provider's, you already control the losing account and can approve the transfer yourself with no waiting at all. Mention it to support and they will confirm which applies.
Full Port: Coming From a Phone Carrier
What to Gather
A genuine port is checked field-by-field against your current carrier's records. A single mismatch — an old address, a slightly different business name — gets the request rejected and adds roughly two weeks. Pull these from your account rather than from memory:
- The number or numbers you want to bring over. Several can go on one request.
- The authorized user's full name, exactly as it appears on the account, including the business name if that is what is on file.
- The service address on file. It must be a real street address — PO Boxes are rejected.
- Your account number with the carrier.
- Your account PIN or passcode, usually four or five digits. Their support team can give it to you.
- A Customer Service Record (CSR). Ask your carrier for this by name. It is their own record of items 2 through 5 — by far the best way to avoid a rejection. If they will not provide one, send your most recent bill instead, dated within the last 30 days and showing the account number, owner name and address.
What Happens After You Submit
Support then sends you a Letter of Authorization to sign, which formally authorizes the transfer — the last thing needed from you.
The request is then validated, passed to the carrier network, and sits with your current carrier for about a week before they approve it and set a cutover date. Your number keeps working throughout; the switch happens on one specific date. There is a one-time port fee on this route; support will confirm the current amount when they open your request.
Texting Switches On After Calling
This applies to both routes and catches people out, so it is worth understanding before you start.
Calling works as soon as your number lands on Revamp365. Texting does not. The A2P 10DLC compliance registration that lets a number send business texts in the US does not transfer between providers — not even on the fast route where the number stays on the same network. It has to be rebuilt under a Revamp365 campaign.
Support starts that registration as soon as your transfer is underway so it is ready and waiting, but expect a gap of roughly a day or two after your number arrives before texting turns on. Attaching a number to a messaging campaign too early can make the registration fail outright, which is why the wait exists.
Nothing is broken if you see calling work and texting not work on day one. For the detail on what that registration involves, see activate texting and the Trust Center checklist and A2P 10DLC registration.
After the Number Arrives
Support configures the number once it lands, which covers inbound calling, inbound texting, and linking it to AI Voice so it can run voice agents. You will be told when it is live.
Check it yourself with three quick tests:
- Call the number from your cell and confirm it rings through.
- Text the number and confirm the message appears in your conversations.
- Once texting is enabled, send an outbound text to your own cell.
If any of the three fails, contact support rather than waiting — it usually means one piece of configuration still needs attaching.
Common Reasons a Transfer Stalls
| What happened |
What to do |
| Current provider has not sent their approval (fast route) |
Chase their support directly. This is the most common delay. |
| Outstanding balance on the old account |
Settle it. Providers decline releases over unpaid accounts. |
| Details do not match the carrier's records (port route) |
Ask for a CSR and resubmit with their exact values. |
| Number went inactive or the account was closed |
Ask your provider to reinstate it immediately, then tell support. |
| Number is part of a larger block |
Your carrier may need to break the block, or all numbers in it must move together. |
| Calling works, texting does not |
Expected for a day or two. Beyond that, contact support. |
Set Up and Customize an AI Voice Campaign
An AI voice campaign calls a list of contacts with your AI agent, on the schedule you set, and then runs the Outcomes rules you define after every call — tagging contacts, setting lead temperature, marking numbers valid or invalid, retrying no-answers, and more. This guide walks the five-step Create Campaign wizard end to end.
Before you start: publish at least one AI voice agent (it needs an outbound phone number), and have your contacts in the CRM. New here? See Build a Custom AI Voice Agent from Scratch and Provision a Phone Number for SMS and AI Voice.
Open the campaign builder
Go to AI Voice & Chat → Campaigns and click New Campaign. The builder opens as a full-screen wizard with a step rail on the left: Basic Info → Recipients → Schedule → Outcomes → Review. You can jump between steps using the rail at any time; a green check marks each completed step. Nothing is saved until you click Create on the final step, so it is safe to click around.
Step 1 — Basic Info
Give the campaign a clear name (this is internal — contacts never see it) and an optional description. Then choose the AI Agent that will place the calls. The agent must have an outbound number; if it does not, you will see a warning here and the step is marked incomplete.
Under number routing you can attach a pool of caller-ID numbers and control rotation:
- Max calls per number per day — caps daily volume on each number to protect its reputation.
- Calls before rotating — how many calls to place from one number before moving to the next in the pool.
Rotating across a small pool and capping per-number volume keeps your caller IDs from being flagged as spam. See Provision a Phone Number for buying more numbers.
Step 2 — Recipients
Choose who gets called by filtering your CRM. You can combine filters to narrow the list:
| Filter | What it does |
| Tags | Include contacts that have (or do not have) specific tags. |
| Lead status | Target a single lead status, e.g. #1 Make Contact. |
| Lead source | Limit to a source such as a specific webhook or import. |
| Saved view | Reuse a CRM view you already built. |
| Phone status | Default is valid — skip numbers already marked invalid. |
Call all phone numbers — when a contact has more than one number, turn this on to attempt each number, not just the primary. Leave it off to call only the primary number.
Step 3 — Schedule
Set the days and time windows the campaign is allowed to dial, plus the timezone calls are scheduled against. Optionally set a max calls per hour to pace the campaign, and a start and end date.
Stay compliant. Keep calling windows within local business hours for the contact's timezone. See Sending Windows, Quiet Hours, and Daily Caps.
Step 4 — Outcomes (the call flow)
This is the heart of the campaign. The Outcomes step defines what happens after each call based on how it went. A complete, recommended Default Flow is already loaded for you when you create a campaign — you can use it as-is, tweak it, or clear it and start over.
Saved Rule Sets toolbar
At the top of the Outcomes step you can manage whole rule sets:
- Default Flow — the system preset. Click it any time to reload the recommended rules.
- Save Current — save your current rules as a named preset to reuse on future campaigns.
- Refresh — reload your saved presets.
- Clear Current Rules — remove every rule and start from scratch (you will get an "are you sure" confirmation first).
All three controls — the toolbar (Save Current / Refresh / Clear Current Rules), the Default Flow chip, and the Help link — are visible at the top of the Outcomes step in the screenshot above.
Anatomy of a rule
Each rule has three parts:
- When — the call outcome that triggers the rule: Call Completed, No Answer, Voicemail, Busy, or Call Failed.
- AI Analysis Conditions (optional, for completed calls) — narrow the rule using what the AI heard, such as Call Disposition = Qualified - Interested, Interest Level = Hot, or Price Node Reached = Yes. With two or more conditions you can match ALL (AND) or ANY (OR).
- Actions — what to do when the rule matches. Add as many as you need.
For example, Rule 1 in the Default Flow reads: When a call is Completed and the disposition is Qualified - Interested → add the ai_interested tag, remove Continue Calling, and mark the dialed number Valid.
Actions you can take
| Action | Use it to |
| Add / Remove Tag | Tag contacts by outcome, e.g. ai_interested. |
| Set Lead Temperature | Mark Hot / Warm / Cold based on Interest Level. |
| Update Number Status | Mark the specific number that was dialed Valid, Invalid, or Unknown — without touching the contact's other numbers. |
| Mark as Do Not Call | Add the contact to your DNC list. |
| Pause Campaign for Contact | Stop calling this contact for the rest of the campaign. |
| Retry Call | Re-dial after a delay (used by the No Answer / Voicemail / Busy / Failed rules). |
| Send Follow-up SMS | Text the contact after the call. |
| Update Lead Status / Notes, Assign to Team Member, Trigger Webhook, Schedule Callback | Route, document, and integrate the outcome. |
Update Number Status beats a "wrong number" tag. The default flow marks the dialed number Valid after a real conversation and Invalid on a wrong number / not-owner outcome. Because it targets only the number that was dialed, a contact's other phone numbers stay callable.
What the Default Flow does
- Interested (or price discussed) → tag
ai_interested, remove Continue Calling, mark the number Valid.
- Not Interested → tag
ai_not_interested, mark the number Valid.
- Wrong Number / Not Owner → pause the campaign for the contact and mark the number Invalid.
- Do Not Call → pause and mark the contact DNC.
- No Answer / Voicemail / Busy / Failed → retry once after a short delay.
- Interest Level Hot / Warm / Cold → set the matching lead temperature.
Form vs. Visual, and Review with AI
Toggle Form / Visual in the top-right to edit rules as a list or as a flow diagram. Use Review rules with AI to get a plain-English audit of your rules and catch gaps (for example, overlapping rules where only the first match runs) before you create the campaign.
Step 5 — Review & create
The Review step summarizes your agent, recipients, schedule, and the number of Outcome rules. Confirm everything looks right, then click Create. You can also Save Draft at any point to finish later.
After launch
Once running, you can pause, resume, and watch progress from the Campaigns list. To read what happened on each call, see Read AI Call Transcripts, Recordings, Summaries, and Post-Call Analysis. To hand qualified leads off automatically, build a follow-up AI Voice Workflow.
FAQ
Do I have to build the Outcomes rules myself? No. A complete Default Flow is loaded automatically. Adjust it only if you want different behavior.
What's the difference between a campaign and a workflow? A campaign calls a one-time list on a schedule. A workflow is an ongoing, multi-step cadence (calls + texts) that contacts are enrolled into over days or weeks.
Will it call outside my hours? No. Calls only go out within the days, time windows, and timezone set on the Schedule step.
Can I reuse my rules? Yes. Use Save Current to store a named rule set and load it on future campaigns.
Set Up and Customize an AI Voice Workflow
An AI voice workflow is an ongoing, multi-touch cadence — a sequence of AI calls and text messages spread over days or weeks — that contacts are enrolled into. Where a campaign dials a one-time list, a workflow keeps following up automatically until the contact connects, opts out, or finishes the sequence. This guide walks the five-step Create Workflow wizard.
Before you start: publish an AI voice agent with an outbound number. If your cadence includes SMS steps, your account also needs A2P 10DLC approval to publish (you can still Save Draft). See A2P 10DLC.
Open the workflow builder
Go to AI Voice & Chat → Workflows and click New Workflow. The wizard's step rail is Basic Info → Schedule → Cadence → Outcomes → Completion. Use the rail to move between steps; nothing is saved until you publish or save a draft.
Step 1 — Basic Info
Name the workflow and choose the AI Agent that places the calls. As with campaigns, attach a number pool and set per-number daily caps and rotation to protect caller-ID reputation.
Step 2 — Schedule
Set the days and time windows the workflow may contact people, and the timezone. You can also set a daily call cap to pace enrollment. Steps in your cadence only fire inside these windows, so a step "due" overnight waits until the next allowed window.
Step 3 — Cadence
The Cadence is the sequence of touches. Add Call and SMS steps, each with a delay from the previous step, to build a follow-up rhythm. Start from a template and customize:
- Aggressive — tightly spaced touches over a few days.
- Moderate — a balanced multi-day sequence.
- Mild (30 days) — slower, longer nurture.
- Drip (6 months) — occasional long-term touches.
Double-tap heads-up. If two call steps land within a few minutes of each other, the builder flags it — a contact would be called twice in quick succession. You can remove the double-tap steps with one click, or keep them intentionally. Note that retries are handled by the cadence here, not by the Outcomes step.
SMS steps require A2P 10DLC approval before a workflow can be published. Until then you can Save Draft, or remove the SMS steps to publish a call-only workflow.
Step 4 — Outcomes (global call-flow rules)
The Outcomes step (labeled "Global Outcome Actions") defines what happens after any call in the cadence, based on how it went — regardless of which cadence step placed it. A complete Default Flow is pre-loaded for new workflows, so you are never staring at a blank screen.
Saved Rule Sets toolbar
Just like campaigns, the toolbar (visible at the top of the Outcomes screenshot above) lets you reload the Default Flow, Save Current as a named preset, Refresh your presets, and Clear Current Rules (with an "are you sure" confirmation) to build from scratch.
Anatomy of a rule
Each rule is When (the call outcome) + optional AI Analysis Conditions (what the AI heard, e.g. Call Disposition = Qualified - Interested, Interest Level = Hot) + Actions. For the full breakdown of conditions and every action, see the campaign guide's rule anatomy — the builder is shared.
What's different from a campaign's Outcomes
| Difference | Why |
| No "Retry Call" action | The cadence (Step 3) handles re-dials. Retry actions are stripped from a workflow's Outcomes, so place double-tap retries in the cadence instead. |
| "Move to Another Workflow" | Hand a contact off to a different workflow after this one completes — useful for routing hot leads into a closer's sequence. |
| "Stop Cadence for Contact" | The pause action is named for workflows — it stops further touches for that contact. |
| Number status | Same as campaigns: mark the dialed number Valid after a real conversation and Invalid on a wrong number, leaving the contact's other numbers callable. |
What the Default Flow does
- Interested (or price discussed) → tag
ai_interested, mark the number Valid, hand off via Move to Another Workflow.
- Not Interested → tag
ai_not_interested, mark the number Valid, hand off.
- Wrong Number / Not Owner → stop the cadence and mark the number Invalid.
- Do Not Call → stop the cadence and mark DNC.
- Interest Level Hot / Warm / Cold → set the matching lead temperature.
Use Form / Visual to switch between a list and a flow diagram, and (once the workflow exists) Review rules with AI to audit your rules.
Step 5 — Completion
Choose what happens when a contact reaches the end of the cadence without an earlier rule stopping them — for example, do nothing, or move them into another workflow. This is the safety net for leads who never connected.
When you are ready, Publish the workflow (or Save Draft). Then enroll contacts from the CRM to start the cadence.
FAQ
Campaign or workflow? Use a campaign for a one-time call blast to a list. Use a workflow for ongoing, multi-touch follow-up (calls + texts) over time.
Why can't I add a Retry Call action in Outcomes? Retries belong in the cadence for workflows. Add a second call step in Step 3 with a short delay to "double-tap" unanswered calls.
Why can't I publish? Your cadence has SMS steps but your account isn't A2P-approved yet. Save Draft, or remove the SMS steps to publish a call-only workflow.
Do I have to build the Outcomes rules? No — a complete Default Flow is pre-loaded. Customize only if you want different behavior.
AI Voice Agents: build, train, and launch your first voice agent
What AI Voice Agents do (BETA)
An AI Voice Agent is a Revamp365-managed phone bot that can place outbound calls, answer inbound calls on numbers you own, hold a real conversation with the caller, and update your CRM with what it learned. You design the agent's voice, persona, knowledge base, and call goals once, then point a campaign or a phone number at it.
Time to first call: ~10 minutes if you already have a phone number provisioned and any A2P/voice compliance done. Result: a published agent that can be triggered manually, attached to an inbound number, or used as the dialer for a voice campaign.
Voice agents and SMS AI agents share the same screen but are different products. SMS AI agents send text conversations, voice agents place / receive phone calls.
How the screens fit together
Open AI Voice & Chat from the sidebar. The page is a single hub split into tabs:
| Tab |
What you do here |
| Dashboard |
Health stats, recent calls, quick actions for "new agent / new campaign / view history". |
| Agents |
Create and edit agents. Sub-tab "Knowledge Base" stores docs the agent can quote from. |
| Campaigns |
Outbound automation. Sub-tabs: voice, sms-blast, sms-drip. |
| Workflows (BETA) |
Multi-step automations triggered by call outcomes. |
| Phone Numbers |
Provisioned numbers and which agent answers which one. |
| Call History |
Every inbound + outbound call with recordings, transcript, and the action log. |
| A2P Registration |
Trust Center / 10DLC for any SMS the platform sends. Separate guide. |
| Analytics |
Volume, duration, sentiment, and conversion rollups. |
Build your first agent
- Open Agents → click New agent.
- Pick a starter template (Lead qualifier, Appointment setter, Voicemail drop, etc.) or "Blank".
- Fill in the four core fields:
- Agent name — internal label, never spoken.
- Voice — the TTS voice. Click play to preview.
- Direction —
inbound, outbound, or both.
- LLM — the underlying model (e.g. leading AI models). Higher tier = slower + smarter.
- Write the greeting (first thing the caller hears) and the system prompt (the agent's persona, goals, and rules).
- Add up to 5 knowledge base documents on the Knowledge Base sub-tab — pricing, scripts, FAQs. The agent can quote these verbatim.
- Click Save draft, then Publish when you're ready to go live.
Keep system prompts under ~800 words. Anything longer slows down the agent's first response. Move long material to the knowledge base instead.
Connect a phone number
Open Phone Numbers. Each row shows a number you own and which agent currently answers it.
- If you don't have a number yet, click Provision number, pick an area code, and confirm — billing kicks in on the next cycle.
- Click Assign agent on the row and pick the published agent.
- Place a test call to that number from your cell phone. The call should be answered by your agent within ~2 seconds.
Inbound voice agents require an active phone number. Outbound-only agents do not need a number assigned to them — the campaign supplies the caller-id at run-time.
Run an outbound voice campaign
- Open Campaigns → voice.
- New campaign. Pick the agent, the caller-id number, and the contact list (any saved CRM list).
- Set the dialing window (e.g. 9am–6pm in the contact's timezone) and the daily call cap.
- Click Schedule. The campaign starts at the next allowed window.
While the campaign runs, every call is appended to Call History with status (completed, no-answer, voicemail, failed) and the agent's structured outcome (e.g. "interested", "not-interested", "wrong-number").
What happens after a call
Every call produces:
- Recording + transcript (searchable in Call History).
- Summary generated by the AI (~3-5 sentences).
- Post-call analysis — structured fields you defined on the agent (e.g.
budget, timeline, motivation).
- Action log — every CRM update made by the agent (tags added, lead status change, note created).
If the call qualified the lead, the matching CRM Contact gets the post-call fields, a timeline note, and any tags the agent decided to add.
Troubleshooting
| What you see |
Likely cause |
Fix |
| Agent doesn't pick up |
Number not assigned to the agent, or agent unpublished. |
Phone Numbers → confirm assignment. Agents → Publish. |
| Agent rambles / ignores instructions |
System prompt too long or contradictory. |
Trim system prompt. Move facts to knowledge base. |
| Voice sounds wrong |
TTS voice cached on previous version. |
Edit agent → re-pick the voice → save → republish. |
| Outbound campaign stuck on "scheduled" |
No allowed dialing window today, or daily cap reached. |
Check campaign settings + campaign action log. |
| Calls drop after a few seconds |
Number's voice capability disabled. |
Phone Numbers → confirm Voice = enabled. Contact support if it remains off. |
FAQ
Can the agent transfer a call to a human?
Yes — define a transfer condition + target number in the agent's call-flow rules. The agent will say its hand-off line then warm-transfer.
Can I record only some calls?
Recording is on by default. To opt out, edit the agent's settings → Recording: off. State / federal call-recording laws still apply.
What languages are supported?
Voice + LLM language is set per agent. English (US/UK), Spanish, French, German, Portuguese (BR) are supported in current voices. Mix-language calls are not supported on a single agent — duplicate the agent and pick a different voice/LLM language.
Why is "Workflows" labeled BETA?
Workflows orchestrate multi-step actions across calls + SMS + CRM updates. They work, but the editor and naming may change. Existing workflows stay running across UI changes.
Why is "Call Flows" hidden in my account?
The full Call Flows editor is a Dealmaker-tier feature. Standard accounts use the simpler agent-level rules. Contact your account manager if you need it.