Launch approval

Page Summary

• Agent launch involves a review process including brand verification and asset assessment managed by Google, carriers, or both.

• Google-managed launches typically take 1-3 business days if brand approval and all required assets are provided promptly.

• Before submitting a launch request, you must complete pre-launch tasks such as filling out agent information, testing, implementing opt-out, and providing an agent preview video.

• Submitting a launch request initiates the review and requires providing details about countries/carriers, contact information, message triggers, interactions, and opt-out responses.

• For Google-managed launches, a mandatory brand approval email must be responded to by the brand point of contact to expedite the review process.

Note: To launch an agent using the RBM Management API, refer to the Submit agent for launch, Look up an agent's launch status, and Add additional carriers to an agent launch documentation.

When you request to launch an agent, it undergoes a review that includes brand verification and an assessment of your agent's assets. This process is managed by Google, carriers, or both, depending on which carriers you've selected for launch. For more information, see Google-managed vs. Carrier-managed launches.

Google-managed launches take 1–3 business days, provided these conditions are met:

• Brand approval is received within 24 hours.

• The following agent assets are accurate, formatted correctly, and publicly accessible:

  • A working website
  • Privacy Policy and Terms of Service links

• Inputs are provided for all fields in the Agent experience section of the launch questionnaire, and the detailed flows are clearly demonstrated by either supplying an agent preview video or writing detailed, step-by-step text transcripts within those fields. For more information, see Describe your agent experience.

Important: Launch requirements vary for carrier-managed networks. Check with each carrier directly for their specific criteria.

After your agent is approved for launch, you'll receive a confirmation email. At this point your agent can begin sending messages.

Prerequisites

Before you can launch your agent, you need to complete the following:

Pre-launch tasks
Fill out all required agent information
Test the functionality of your agent and its associated infrastructure
Implement the STOP (or opt-out) flow
Prepare inputs describing your agent's triggers, interactions, and opt-out flow that match the use case declared when the agent was created. To clearly demonstrate these flows and the STOP functionality, prepare either an agent preview video or detailed, step-by-step sample text transcripts. For more information, see Describe your agent experience.
For Google-managed launches: Launch the agent on at least one carrier-managed network in each country where you intend to launch on Google-managed carriers. For more information, see Google-managed vs. Carrier-managed launches.

With those tasks complete, gather the following information:

Launch information
List of countries and carriers you want your agent to operate in
Name, email address, and phone number for the points of contact at your company who are responsible for the agent
(Optional) How you obtain opt-in to message users
What actions or events trigger messages to users
A step-by-step transcript of the interactions your agent will have with users, or an agent preview video showing these interactions at a publicly accessible URL
The opt-out keywords your agent supports and the exact message the agent responds with when a user opts out of communications, or an agent preview video showing this opt-out interaction at a publicly accessible URL
(Optional) Instructions for how reviewers can directly access your agent on their devices
(Optional) Screenshots at publicly accessible URLs
(Optional) Agent preview video at publicly accessible URLs

Describe your agent experience

You must provide either an agent preview video or detailed, step-by-step text transcripts for the following fields:

• Trigger description
• Interaction description
• Opt-out description

Make sure your descriptions match the use case that you specified when creating your agent.

The following sections provide best practices and sample templates for each field. The templates are examples only. Adapt them to accurately describe how your agent works.

Trigger description

In the Trigger description field, explain the exact user action or event that initiates the first interaction between your agent and the user. This helps confirm that your outreach is legitimate and aligns with the user's consent.

Describe all possible triggers for your agent. For example, if your agent handles both A2P (application-to-person) and P2A (person-to-application) conversations, provide the triggers for both.

When you write your trigger description, follow these best practices.

For A2P:

• Specify when and why your agent sends the message.
• Be specific about the event or transaction that triggers the message. For example, instead of writing "When a user signs up," describe how and where the user signs up.
• Make sure the trigger matches your agent's use case.

For P2A:

• Specify the exact discovery surfaces or entry points that the user interacts with, such as Google Search results or your brand's website.
• Describe the exact user action that starts the conversation, including the text of the pre-populated message.

The following sample templates show how to describe the trigger for each agent use case.

One-time passwords (OTP)

Sample template:

The message is triggered immediately when a user requests a one-time password on the [brand name] [website or mobile app] to [authenticate a login, verify a phone number, or authorize a transaction].

Example: "The message is triggered immediately when a user clicks 'Send code' on the Acme Bank portal to authorize a wire transfer."

Transactional

Sample template:

The message is triggered by a specific customer transaction, such as [an order confirmation, a shipping update, an appointment reminder, or a flight delay].

Example: "The message is triggered automatically by our backend logistics system as soon as a customer's package is handed over to the courier service."

Promotional

Sample template:

Messages are sent periodically to users who have opted in to receive marketing communications. Triggers might include [seasonal sales, abandoned shopping carts, or loyalty program milestones].

Example: "The message is triggered 2 hours after a registered user leaves items in their shopping cart without completing the purchase."

Multi-use (A2P)

Sample template:

This agent handles both transactional and promotional events. Triggers include [a transactional event, such as a monthly billing statement] and [a promotional event, such as a seasonal upgrade offer].

Example: "Messages are triggered when a user reaches 90% of their monthly data limit (transactional) or when a new bundle discount becomes available to the user (promotional)."

Multi-use (P2A)

Sample template:

The conversation is user-initiated through [discovery surface, such as the brand website, Google Search, or a QR code]. The user interacts with the [entry point name], which opens the native messaging app with a pre-populated message that reads "[pre-populated text]". Sending this message triggers the agent's conversation flow.

Example: "The conversation is user-initiated through a chat widget on the Acme Tech Support website. The user taps the 'Chat with Support' button, which opens the native messaging app with the pre-populated message 'I need help with my account'. Sending this message triggers the agent's customer service flow."

Interaction description

In the Interaction description field, provide a step-by-step transcript of your agent's conversation flow. Include the exact text of your agent's standard messages and any rich elements, such as suggested replies, suggested actions, rich cards, and carousels.

When you write your interaction description, follow these best practices.

For A2P:

• Format the description as a script or a step-by-step flow.
• List all buttons, suggested replies, suggested actions, and media links included in each message.
• Clearly show the conversation path that users follow.

For P2A:

• Start the transcript with the user's initial inbound message.
• Clearly map out the two-way conversation flow, showing how the agent responds to the user's intent and guides them toward their goal.

For complex agents, provide transcripts for the primary user journeys and one or two common alternative paths. You don't need to document every possible error state or fallback.

The following sample templates show how to write a transcript for each agent use case.

One-time passwords (OTP)

Sample template:

Agent: [OTP code] is your verification code for [brand name]. It will expire in [number] minutes. Don't share this code with anyone.

Example:

• Agent: 554123 is your verification code for Acme Bank. It will expire in 10 minutes. Don't share this code with anyone.

Transactional

Sample template:

Agent: Hi [customer name], your order [order number] has been confirmed! You can track your delivery status below.
    

      
Rich card image: [image description]
      
Suggested action: "Track order" (Open URL action: [tracking link])
      
Suggested action: "Contact support" (Dial action: [phone number])
    
    

Example:

• Agent: Hi John, your order #XY7765 has been confirmed! You can track your delivery status below.
  • Rich card image: An image of a package with a green check mark.
  • Suggested action: "Track order" (opens the tracking page in a browser)
  • Suggested reply: "View invoice"

Promotional

Sample template:

Agent: Flash sale alert! Enjoy [X]% off all [product category] this weekend only. Use code [code] at checkout.
    

      
Carousel card: Image of [product]. Title: [product name]. Description: [description]. Suggested action: "Buy now" (Open URL action: [product link]).
      
Suggested replies: "Browse categories", "Unsubscribe"
    
    

Example:

• Agent: Flash sale alert! Enjoy 20% off all sneakers this weekend only. Use code KICKS20 at checkout.
  • Carousel card 1: Image of Runner X shoes. Title: Runner X Shoes. Suggested action: "Shop now".
  • Suggested reply: "View deals"

Multi-use (A2P)

Sample template:

    

      
Agent (transactional flow): Your bill for [month] is ready. Amount due: [amount].
      
Agent (promotional flow): Thank you for your payment! Upgrade to our Premium plan today and get one month free.
      
Suggested action: "Upgrade now" (Open URL action: [upgrade link])
    
    

Example:

• Agent (transactional flow): Your bill for September is ready. Amount due: $45.99.
• Agent (promotional flow): Thank you for your payment! Upgrade to our Premium plan today and get one month free.
  • Suggested action: "Upgrade now" (opens the plan upgrade page)

Multi-use (P2A)

Sample template:

    

      
User: [pre-populated entry message or the user's typed question]
      
Agent: [welcome message acknowledging the user's intent]. [prompt for the user to narrow down their request].
      
Suggested replies: "[option 1]", "[option 2]", "[option 3]"
    
    

      
User: [selects option 1]
      
Agent: [provides the requested information or routes the user to the next step].
      
Suggested action: "[action label]" ([link to a secure form, live agent handoff, or other next step])
    
    

Example:

• User: I need help with my account
• Agent: Welcome to Acme Support! To get started, select the topic you need help with today.
  • Suggested replies: "Billing", "Technical issue", "Password reset"
• User: Billing
• Agent: I can help with that. Sign in to your secure account portal so I can pull up your recent invoice details.
  • Suggested action: "Sign in" (opens the account portal)

Opt-out description

In the Opt-out description field, list the opt-out keywords that your agent supports and provide the exact confirmation message that your agent sends when a user opts out.

When you write your opt-out description, keep the following requirements in mind:

• Your agent must support the keyword STOP and the equivalent keywords in other languages that your agent supports.
• The confirmation message must clearly state that the user is unsubscribed and won't receive any more messages.
• After the confirmation message, your agent can't send the user any more messages unless the user opts in again.

For more information, see Respect when users don't want messages.

The following sample template applies to all agent use cases.

All use cases

Sample template:

Supported keywords: STOP, UNSUBSCRIBE, CANCEL
Confirmation message: You have been successfully unsubscribed from [brand name] alerts. You will receive no further messages. To opt back in, reply [START or RESUME].

Example:

• Supported keywords: STOP, UNSUBSCRIBE, CANCEL
• Confirmation message: "You have been successfully unsubscribed from Acme Bank alerts. You will receive no further messages. To opt back in, reply START."

Submit a launch request

Submitting a launch request initiates the brand verification process and a review of your agent's assets by Google, carriers, or both.

To submit a launch request, follow these steps:

1. Open the RCS for Business Developer Console, sign in with your RCS for Business partner Google Account, and click your agent.
2. In the left navigation, click Launch.
3. Click Get started.

4. Answer some questions about the state of your agent. Your agent should be ready to:

   • Communicate with users who aren't on test devices.
   • Allow users to opt out of communications.
   • Allow reviewers to evaluate your agent's primary, secondary, and opt-out flows through direct testing, an agent preview video, or detailed text transcripts.
   Important: If your agent isn't ready for one or more of these tasks, it isn't ready to launch. To prevent launch delays, your agent must satisfy all listed requirements before submission. You must complete the Trigger description, Interaction description, and Opt-out description fields. You must provide either an agent preview video or exhaustive, step-by-step transcripts in these three fields. Ensure that your submission clearly proves your declared use case and STOP functionality.

5. Click Fill out questionnaire.

   Note: You won't be able to edit the questionnaire after submitting the launch request. If you need to update the questionnaire after launch, contact the RCS for Business support team.

6. Fill out all fields for the following sections with accurate information.

   • Agent information
   • Points of contact

   • Agent experience

     For best practices and sample templates for the Agent experience section, see Describe your agent experience.

   Note:

   If your "multi-use" agent has one use case defined at launch and another planned for a later date, make sure to:

   1. Add the following statement to the Interaction description field in the Agent experience section:
   • "Currently sharing the proof for only the transactional/promotional use case. Selected the agent use case as multi-use because we plan to use the same agent for both transactional and promotional use cases in the future."
   2. Submit proof to the Support team that you've launched the second use case within six months of the agent's initial launch date.
   3. If you anticipate a delay, request a three-month extension from RBM Support within the initial six-month period.
   4. If you take the extension, submit proof that you've launched the second use case within the three-month grace period. If proof of the second launch is not received by the final deadline (including any extension), the agent will be converted to a single-use agent reflecting its initial launch case.
   Important: Make sure the information you input is accurate. It will be reviewed by Google, carriers, or both prior to the launch of your agent. To prevent delays due to missing launch information, make sure your submission includes an active link to the Terms of Service and the Privacy Policy.

7. Click Save.

8. Click Select carriers.

9. Select the carriers whose networks you want to launch the agent on.

   Tip: To demo or test an RCS for Business agent in countries or on carriers where RCS for Business is not available, use a test device.

   When RCS for Business becomes available on a new carrier network, you need to submit a launch request for your agent:

   • Using the Console (the current flow)
   • Using the RBM Management API

   The approval process can vary depending on whether the new carrier is carrier-managed or Google-managed. Carriers marked with an info icon follow their own launch approval process. You must have direct agreements with these carriers before they'll approve your agent to launch on their networks. These carriers reject your launch if they have no contract in place with you, and they may contact you to establish a relationship.

10. Click Confirm.

11. Review the next steps, then click Confirm.

After submitting a launch request

After submitting an agent for launch on Google-managed carriers, the brand POC receives an email from [e-mail] (with the subject line "Please approve brand use for RCS Business Messaging agent: AGENT_ID"). Please reach out to the brand POC proactively and ask them to reply to that email as soon as possible. This step is required, and faster email responses help expedite the launch approval.

For carrier-managed launches, the verification process can differ.

After you submit launch requests, the Launch page displays your agent's status for each selected carrier. This page may take a while to update. After the status changes to Launched, your agent will be reachable after about 30 minutes.