How AI Text Summarization Saved Us From Wading Through a 32-Page Call Transcript
- Andrew Kita
Kicking off a project with a new client always feels good. The account is shiny and new; both sides are optimistic about the great things they will achieve together. The kick-off meeting starts with friendly banter before diving into all those vital “getting to know you” questions.
For complex projects or companies with niche offerings, the kick-off can last a lot longer than a typical meeting. After all, for us to deliver innovative strategy and compelling marketing content, we need to fully understand our clients: their offerings, their audiences, and all the little niggly details of both. And it was after a recent meeting with an exceptionally technical client that we discovered the benefits of AI text summarization.
Let me set the stage.
Meet Torchlight AI
The team behind Torchlight AI started out with a mission to bring the power of behaviour-based intelligence and analytics used in the defense industry to commercial customers. Torchlight helps users monitor activity at different locations to identify and guard against potential threats.
Focusing on multinational brands with interests overseas — a growing audience in today’s era of globalization — the company provides advanced, proactive operational risk solutions that can help customers safeguard their assets and supply chains.
Even in terms of a TPM client — our customers are primarily in the B2B technology space, so complex, technical subject matter isn’t unusual for us — Torchlight offers something truly unique. Our team spent close to two hours on an initial kick-off call with them, getting a picture of their team, their business, their use cases, and their target audience. And then it was time to start sifting through all that new information to build a basis for our strategy.
AI Summarization to the Rescue
After trimming out pleasantries and goodbyes, we had one hour and 45 minutes of recorded content from our kick-off call. Woof. This is where AI text summarization tools get put to the test. There are loads of transcription tools out there. But there are several problems with straight transcription.
- Too long to be useful
Our 1.75-hour call yielded approximately 32 pages of transcription. No one was ever coming back to that unless it was to search for specific keywords and pull out quotes. We were more likely to rely on notes taken by hand during the call than to wade through 32 pages of dense, technical text.
- Meetings aren’t linear
No meeting has ever gone from topic to topic in the ideal order for storing and absorbing information — let alone a first meeting. Questions emerge as conversation goes on, subjects are revisited. Sometimes a complete understanding of a client’s solution comes after piecing together explanations that occurred at various points in the call. We had to dissect the content fully to find all the essential information.
During our experiments with marketing tools that use generative AI, we discovered Parrot AI — and decided to put it to the test on this task. (Full disclosure: our go-to AI meeting summary tool has, for a long time, been Otter.AI. We’ve made good use of it over the years and will also be exploring the improvements they’ve made as technology has advanced.)
AI Tools for Efficient Marketing
Parrot AI offers many of the same features as Otter, including live or post-transcription. (Currently, both tools provide five hours of transcription per month on their free plans, but Otter limits transcriptions to 30 minutes.) But it has a few other AI text summarization features that we had never tried on Otter — that went a long way in helping us organize the information we had gathered on Torchlight.
1. Meeting Summary
Through the magic (OK, science) of generative AI, we had a complete summary of the entire recording within minutes. The copy was in natural language, making it easy to understand — a significant improvement over copying and pasting from a summary and dealing with all the “ums” and silences.
2. Answers to Onboarding Questions
- What is the scope of the work our new clients expect from us?
- What marketing assets do they currently have?
- What are the most common use cases for their offering?
- Who is their target audience?
- What kind of brand image do they have in mind?
- What differentiates them from their competition?
- What key words or phrases does the client use to describe their offering?
- Who are the key points of contact we should know?
3. Ways to Leverage Recordings
More than just an AI summarizer, Parrot was also able to provide timestamps for when key topics or words were mentioned during the recording, allowing us to go back to exact points to revisit exactly what the client said. Parrot also lets users watch recordings with closed captions at up to three times the standard speed.
4. Repurposing Content
With all our ducks in a row, we decided to explore the other ways we could leverage AI text summarization. Repurposing content is always valuable in the marketing world, so we asked Parrot to help us create an ebook based on this call.
To our surprise, we found it was able to provide a clear ebook outline based on the transcript, covering why Torchlight’s product makes sense for a commercial market, the difference between government and commercial use of its technology, and comparisons between its products and use cases. Unfortunately, Parrot’s language model is still fairly slow compared to tools like Jasper or ChatGPT. We determined asking it to create a fully fleshed-out ebook was not a practical use of the tool. But Parrot’s robust privacy policy could help ease client concerns about uploading confidential recordings to a third-party platform.
Accelerate Marketing Efforts with AI and TPM
Honestly, Parrot’s failure to produce a comprehensive ebook based on our first meeting with a client is no surprise — and a bit of a relief. (After all, if AI can do everything, why are we here?) But this experiment did reveal the value of today’s upgraded AI summarizers and how we can use them to make marketing more efficient.
Our team continues to explore new AI marketing tools and how they can help us better serve our clients. Want to know more about what we’re doing? Drop us a line.
