Smarter AI Call Summaries

Overview

The babelforce Call Summarizer uses AWS Bedrock - Amazon Web Services' generative AI platform, to automatically generate structured summaries from customer call transcripts. AWS Bedrock provides access to advanced foundation models capable of understanding, analyzing, and generating human-like text.

Within babelforce, the Call Summarizer transforms call transcripts into structured, actionable summaries that can be automatically used in tickets, emails, CRM systems, and other workflows. Because you have full control over how summaries are generated, and because it uses high-quality, well-formatted full call transcriptions as input, you can instruct the AI to focus on exactly the information your business needs.

Benefits

The babelforce Call Summarizer helps your team:

• Generate summaries of customer calls automatically
• Extract key information from conversations
• Identify customer issues, actions taken, and next steps
• Create consistent documentation across teams
• Reduce time spent reviewing call recordings or transcripts
• Use AI-generated content in tickets, emails, and internal workflows

Different teams can tailor summaries to their own processes. For example:

• A support team may want: customer issue, actions taken, and next steps
• A quality management team may focus on: customer sentiment, resolution status, and escalation risks
• International teams can request summaries and transcripts in multiple languages

Example: Instead of listening to a complete call recording, a support agent can review an AI-generated summary like this:

• Issue: Customer reports repeated login failures after password reset.

• Root Cause: User account synchronization failed between systems.

• Actions Taken: Support agent verified account status and triggered synchronization.

• Next Steps: Customer will test login again within 24 hours. Support team will follow up if the issue persists.

The generated summary can then be automatically added to a ticket, shared internally, or included in customer communications.

Prerequisites

A call recording is required before a transcript can be generated, and a transcript is required before a summary can be generated.

Note: To activate the babelforce Call Summarizer for your account, please contact your Customer Success Manager or reach out to the Customer Success team.

Setup Guide

Step 1: Enable Call Recording

Before a transcript can be created, the call must actually be recorded. This is configured in the ACD, not in Global Automations:

• In the ACD, switch on the Recording toggle.
• Set a condition for when recording should happen (e.g. "Always", or a custom condition depending on your workflow).

Without this step, no recording will exist, and the automation below will have nothing to transcribe.

Step 2: Create a Transcript

Once call recording is active, the transcript is generated via a Global Automation:

• Event: Voice Recording finished
• Action: babelforce.transcripts — transcribe recording

This automation only triggers once a recording has actually been produced (Step 1). Ticket creation or other actions are not required for this step — they are optional and depend on your own workflow.

Once the transcript has been created successfully, it can be used for summarization.

Step 3: Generate a Summary

After the transcript is available, configure the Call Summarizer automation:

• Event: Transcription Finish
• Action: babelforce.summarize : summarize a call

Configuration options:

• Language — Select the language of the conversation.
• Model ID — Choose the AI model that should generate the summary. You can select from multiple models using babelforce's AWS Bedrock integration.
  • Higher model IDs generally provide higher-quality output.
  • Lower model IDs may be faster and more cost-efficient.
• Instructions — Define what information the AI should extract and how the summary should be structured.

Example Instructions:

Example 1: Full Conversation Analysis

• Provide: issue summary, root cause, actions taken, resolution status, and follow-up actions. Also provide the transcript in Italian and German.

Example 2: Resolution Tracking

• Generate: conversation summary, follow-up section, and resolved/unresolved status. Also provide the transcript in Turkish and German.

Example 3: Support Ticket Summary

• Focus on: customer request, actions taken, next steps, and additional follow-up instructions.

Output Variable
The generated summary is stored in: {summarization.summary}
This variable can be used throughout your automation workflows, but only once this step has completed and with the Event "Summarization Finish" 

Step 4: Use the Summary

Only after the summary has been generated (Event: Summarization Finish) can {summarization.summary} be referenced in a follow-up automation. Common use cases include:

• Send via Email: Include the generated summary in customer or internal emails.
• Add to Tickets :Automatically attach the summary to support tickets to provide context and improve documentation.
• Integrate into Other Workflows: The summary can be used anywhere variables are supported, including CRM updates, internal notes, follow-up tasks, reporting processes, and custom integrations.

At the end, your setup will look like this:

Best Practices

• Ensure transcripts are accurate before generating summaries.
• Use clear, specific instructions to guide the AI output.
• Start with simple summary formats and refine your instructions over time.
• Tailor summaries to your team's specific requirements.

Pricing

The babelforce Call Summarizer is an additional feature and may incur extra costs. For pricing information, please contact your Customer Success Manager or reach out to the Customer Success team.