> For the complete documentation index, see [llms.txt](https://docs.reachplatform.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.reachplatform.com/customizations-and-bolt-ons/ai-voice-activation-agent.md).
# AI Voice Activation Agent
Launch a branded inbound voice AI agent that handles mobile activation (new number and port-in) end to end
{% hint style="info" %}
#### A proven, configurable solution
The agent, conversation flow, and APIs are already proven.
A new brand needs configuration, testing, and controlled go-live.
{% endhint %}
### What the AI agent does
The AI voice activation agent is an inbound voice solution that completes the full activation journey for a mobile subscriber without human involvement. The caller dials the support number, once the agent identifies activation as the use case, they get transferred to the activation number, is identified and verified, chooses how they want their line activated, pays the plan charge, and the line is provisioned live on the call.
| **Capability** | **What the agent does** | **Notes** |
| ------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------- |
| **Caller identification** | Identifies the caller from the inbound number, or collects and confirms a registered phone number or an order number | Transfers to a human if no record is found or if unable to identify the customer |
| **Status-based routing** | Reads account, order and line status and routes the call on that basis — already active, ineligible, order not yet delivered, or ready to activate | Prevents invalid activation attempts |
| **Identity verification** | Sends and verifies a one-time passcode before activation (For customers calling from unregistered numbers only) | Retry handling on incorrect codes (OTP) |
| **New number activation** | Assigns and activates a new mobile number on the line | eSIM and physical SIM |
| **Number port-in** | Collects the number and current carrier, runs the portability check, collects port-in credentials (account number, zip and pin) and submits the request | Includes in-progress porting and failure handling |
| **Payment collection** | Retrieves the plan charge, confirms it with the caller and charges the card on file | No card details are captured on the call |
| **Multi-line activation** | After completing one line, offers any remaining eligible lines on the account and repeats within the same call | Driven by live line status, not a fixed order |
| **Device / SIM guidance** | Walks the caller through eSIM install or physical SIM insertion and confirms the line is live | QR code and activation email guidance |
| **Knowledge Q\&A** | Answers activation related questions from a curated knowledge base | Scoped to an approved KB |
| **Escalation & transfer** | Warm-transfers to a human on request, on repeated failure, or on an unrecoverable error, with a summary | Business-hours aware; voicemail outside hours |
| **Reporting** | Writes structured post-call data for every call — outcome, failure reason, transfer reason and sentiment | Repoting on demand |
{% hint style="success" %}
**Exit criteria:** every in-scope journey completes end to end, and every failure path terminates in a defined outcome.
{% endhint %}
### Brand configuration
Every item below is a configuration value and must be confirmed in writing before build starts.
| **Item** | **What is required** | **Owner** |
| ----------------------------------------------------- | -------------------------------------------------------------------------------------------------- | ------------------ |
| **Brand and Agent name** | Exact name as the agent should say it for both brand and agent, including pronunciation if unusual | Brand / Marketing |
| **Brand website** | Self-service URL the agent refers callers to | Brand / Marketing |
| **Greeting & tone** | Opening line and the tone of voice the agent should use | Brand / Marketing |
| **Closing script** | How the agent should close the call and sign off | Brand / Operations |
| **Business hours for warm transfers to human agents** | Days and hours, with time zone, that define in-hours behaviour | Operations |
| **Transfer target number** | Destination for warm transfers during business hours | Contact centre |
| **Out-of-hours behaviour** | Voicemail message or alternative routing outside business hours | Contact centre |
| **Escalation policy** | When the agent should transfer — on request, after how many failures, and on which errors | Operations |
### Pilot and go-live
#### Pilot activities
* Brief the support team on what the agent handles, how escalations arrive, and what context they receive on transfer.
* Launch on a limited share of activation traffic first, reviewing calls daily.
* Ramp up to full traffic as results hold, retaining the ability to route calls back to human agents at any point.
#### Go-live readiness checklist
Confirm the following before opening the agent to live traffic.
**Flow tested end to end**
* New number activation — primary line.
* New number activation — secondary line.
* Port-in — primary line.
* Port-in — secondary line.
* OTP verification.
**Brand configuration confirmed**
* Brand name.
* Greeting.
* Website.
* Business hours.
* Transfer number.
* Closing script.
**Ready to launch**
* Reporting validated against test calls.
* Support team briefed.
#### Hypercare (1 week)
* Daily review of failed and transferred calls, with same-day tuning of prompts and routing.
### Frequently asked questions
Q: Can the voice of the agent be changed, and if so, how?
A: Yes. The agent voice is a configuration setting in the voice platform. The brand can select from a library of available voices or supply a custom voice recording. Voice changes are applied during Stage 1 (Configuration) and take effect immediately on the next call. No code or flow changes are required.
Q: Can the tone of the agent be changed, and if so, how?
A: Yes. The agent tone — friendly, formal, direct, conversational — is defined in the conversation flow prompts. Tone changes are made during Stage 1 by updating the agent instructions and re-deploying the flow. Examples include greeting style, how the agent responds to caller questions, and how it frames choice. Changes are testable immediately on the next call and do not require any other configuration.
Q: How does the QA team start the evaluation process, and who receives the QA results?
A: The QA team evaluates agent calls using a manual quality audit process and a defined set of evaluation criteria. The approach is:
**Sampling & Selection**
Calls are filtered from the call log by duration: calls over 5 minutes, calls between 3–4 minutes, and calls under 2 minutes. From each duration category, 2–3 calls are randomly selected for review.
**Evaluation Process**
The QA team listens to each selected call, reviews all node transitions and tool results, and evaluates them against criteria including sentiment analysis, intent understanding, system health, resolution, repetition, hold time, language flow, ticketing, escalation recognition, and transfer success.
**Results Delivery**
Findings are documented with pass/fail outcomes and shared with the implementation lead and brand stakeholder for next steps. During pilot and hypercare, results are delivered daily.
### Call analytics and performance data
The AI agent captures structured data for every call. The following table shows the metric categories Reach provides to brands through automated reporting and insights.
| **Field group** | **Captured** | **Reporting & Insights Value** |
| -------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Outcome** | Activation successful, port-in submitted, failed, or not applicable — with the failure reason where relevant | Reach reports on activation outcomes with complete visibility into success/failure status, failure reasons, and final disposition. Enables brands to track activation completion rates and troubleshoot issues. |
| **Journey** | Activation preference chosen, whether it was a primary or additional line, SIM type | Reach provides customer journey insights including primary vs. secondary line activation paths, preferred SIM types, and activation method selection. Enables understanding of customer preferences and patterns. |
| **Caller** | Identification method used, verification outcome, account and order reference | Reach tracks caller verification outcomes and account validation success. Enables brands to monitor identity verification effectiveness and account matching accuracy. |
| **Escalation & AI Attribution** | Whether a transfer occurred, who initiated it, the reason and the category; whether the call was fully handled by AI or transferred to a human, and who completed the final activation | Reach reports on escalation frequency, reasons (e.g., AI unable to resolve, customer requested, post-activation issue), and who initiated handoff (AI or customer). Reach also tracks AI handling attribution — whether the AI fully handled the call and who completed the final activation (AI / Human / AI+Human / None). Enables measurement of AI containment and AI-only completion rates, plus escalation driver analysis. |
| **Quality & Reliability** | Caller sentiment at call start and end, whether the call needs human review, any question the agent could not answer, technical errors encountered, overall call efficiency and resolution quality | Reach provides call quality metrics including customer sentiment (start and end of call), whether calls require human review, and unanswered questions, alongside operational metrics such as call disconnect reason, retry counts, AI containment status, technical errors, call efficiency, and resolution quality. Enables assessment of call handling quality, customer satisfaction, and overall system reliability. |
| **Customer & Account Data** | Customer identification, brand account status, line information (primary and secondary), total lines activated, call duration metrics | Reach reports on complete customer account information, brand identification, line activation status (primary and secondary), and call duration. Enables accurate billing, account reconciliation, and multi-line activation tracking. |
| **Intent & Activation Tracking** | Call intent (activation, port-in, support), activation method chosen, final activation status, whether AI completed or human handoff required, port-in eligibility results | Reach provides complete activation journey insights: call intent classification, activation method chosen, final activation status (Success / Failed / Pending / Not Attempted), API response codes, timestamps, and transaction IDs. Enables end-to-end visibility into activation outcomes and customer intent fulfillment. |
| **Call History & Records** | Complete call data available for review including all metrics and outcomes captured above — accessible via data export | Reach maintains complete call records including all captured metrics, transcripts, and recordings available for review. All data is accessible through Reach's automated reporting system for compliance, quality assurance, and analytics purposes. |
Reach delivers comprehensive insights on these metrics to brands on demand, enabling data-driven decisions on agent performance, activation success rates, customer satisfaction, and operational improvements. No platform access is required; all reports and data summaries are provided by Reach.
{% hint style="info" %}
Questions or clarification? Reach out to your respective account manager or email us at for help and support.
{% endhint %}
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:
```
GET https://docs.reachplatform.com/customizations-and-bolt-ons/ai-voice-activation-agent.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.