How To Summarize Dialogue With AI

The Forefront Team
August 17, 2022
How To Summarize Dialogue With AI

If you want to get anything done, chances are you have to communicate with someone else. Transcripts are artifacts of our conversations with others and the ability to distill transcripts into short form summaries can unlock all kinds of previously unattainable analyses.

Customer support organizations can use summaries of conversations their customers have with their agents or chatbots to track evolving customer queries. They can even cluster these summaries to create visual representations of their customers’ needs. People can recap interviews with their favorite celebrities and role models to stay up to date with their lives. Economists can summarize interviews with key figures to keep up with rapidly shifting policy and news.

Forefront Summarize

Modern large language models excel at creating high-fidelity abstractive summaries that can freely paraphrase source transcripts into easy to read text snippets with varying degrees of compression. These models are powerful but can be hard to set up and get started with. Forefront Summarize is an easy to use abstractive text summarization API that can generate high-quality dialogue summaries of any length. Using the API to summarize dialogue is easy and only requires two inputs:

  1. dialogue - a structured list of conversation turns.
  2. compression_level - an integer from 1 to 5 where 1 means the summary will be longer and have more information about the transcript whereas 5 will mean the summary will be as short as possible. This defaults to 3 if not given.

Let’s see the Forefront Summarize API in action.

Example 1: Recapping celebrity interviews

Interviews are often people’s favorite portions of TV shows and podcasts but with so many out there it can be daunting to try and keep up. Forefront Summarize can reduce entire interviews down to the most important points to give us what we need. Let’s use Forefront Summarize to summarize a 60 minute interview with Trevor Noah.

Summarizing a 60 Minutes Interview with Forefront Summarize

With a compression level of 5 we see over 80% of the character count replaced with the key facts of the interview. Imagine reading the insights from 5 interviews in the same time it would take to read the entire transcript of 1 interview.

Example 2: Summing up the Fed

Forefront can tackle dialogue summaries from nearly any field. Let’s use Forefront Summarize to sum up a recent interview with current Fed Chairman Jerome Powell.

Summarizing a Jerome Powell interview with Forefront Summarize

The interview saw a reduction in character count by nearly 90% and still conveys the key economic points that Chairman Powell is trying to get across.

Example 3: Summarizing Customer Support Transcripts

Forefront Summarize can transform a multi-turn dialogue between a customer and a customer support agent into a brief or longer-form summary by setting the desired compression level.

Summarizing a customer support conversation with Forefront Summarize

We can see that the lower compression level of 1 gives us more information about the customer’s sentiment where the highest compression level of 5 just gets straight to the point about what happened.

Dialogue summaries made easy

It’s clear from our examples that Forefront Summarize can summarize all kinds of dialogue from different domains. The possibilities with what we can do with Forefront Summarize are virtually limitless. Documentation for the Summarize API can be found here.

Ready to sign up and get started? Sign up

Ready to get started?

Start fine-tuning and deploying language models or explore Forefront Solutions.

Transparent, flexible pricing

Pay per token or per hour with flat-rate hourly GPUs. No hidden fees or confusing math.

pricing details
Start your integration

Get up and running with your models in just a few minutes.

documentation