AgentChief Series: How I Automated Testimonial Mining with AI

First in a series on practical AI applications for real business problems

Ever had that moment when your website team casually mentions, "Hey, we need client testimonials for the new campaign... by yesterday"? Yeah, me too. Last weekend.

Instead of panicking or calling clients at 11 PM on a Saturday (a strategy I don't recommend), I turned to my AI toolkit to solve this problem. The result? A treasure trove of authentic client testimonials extracted from years of meeting recordings – without the tedious manual work.

Quick Takeaways:

• Problem: Needed testimonials immediately for a new marketing campaign

• Traditional Solution: Manually review 1,500+ meeting recordings (aka "The Weekend Killer")

• AI Solution: Automated extraction of positive client feedback from 5 years of meeting transcripts

• Result: Authentic testimonials ready to use, plus a reusable dataset for future marketing needs

The Testimonial Hunt: Why Manual Methods Make No Sense

Let's be honest – there was absolutely no way I was going to manually comb through 1,500 meeting recordings from the past five years. I could have sent out requests for new testimonials, but that wouldn't solve my immediate need.

I've been recording almost all my client meetings via Zoom and Teams for years, using Grain.com's meeting agent to transcribe everything. That's a goldmine of client feedback just sitting there, waiting to be discovered.

So how do you sift through thousands of hours of conversations to find the gems? This is where AI becomes not just useful, but essential.

The AI-Powered Solution: My Weekend Workflow

Here's exactly how I tackled this problem:

1. First, data consolidation:

   • Created a Google folder and spreadsheet to store all transcriptions

   • Set up a Zapier workflow (sorry PowerAutomate fans) to trigger when meetings were updated in Grain.com

   • For each recording, the Zap created a Word document in Google Drive and logged details in a spreadsheet

   • End result: A folder with 1,500 documents containing meeting transcripts

2. Enter the AI:

   • Zipped the folder and uploaded it to ChatGPT

   • Used this prompt: "I need you to read through all these transcriptions and store each speaking comment in a row with all the details: Date, Speaker, Text/Speech, grain recording url"

   • ChatGPT processed everything and organized every speaker comment into a structured dataset

3. Mining for gold:

   • Fed the dataset to both ChatGPT and Claude with this prompt: "Pull out the most positive comments that are complimentary to CRMChief or me and I need really good quotes for testimonials for my website. Please provide the speaker, date, quote and grain recording url"

   • Interesting observation: ChatGPT was more efficient at data consolidation, but Claude had the edge on extracting the most meaningful quotes and sentiment analysis

The Results: Client Love in Their Own Words

The AI-extracted testimonials were exactly what we needed. It took all of 2 hours.  Here's the complete collection:

• "We love working with you dude. This is one of the best contracting relationships we've ever had. You've heard me say it many times." — Ben, SnyderTech (June 2023)

What's Next: From Weekend Hack to Push-Button Solution

While my weekend AI solution worked beautifully, it's not yet scalable across my team. But that's the beauty of this rapidly evolving tech landscape – I'm confident that a year from now, this entire process will be a single-button operation from an AI service vendor.  For now, I have both the testimonials I needed AND a comprehensive dataset of all client comments from all meetings. That's marketing gold that goes well beyond my immediate need.

Stay tuned for the next installment in the AgentChief Series, where I'll share another real-world business problem solved with practical AI applications. Got questions about how I did this or ideas for future topics? Drop me a line!Transforming Meeting Transcriptions

AI has the power to turn hours of meeting recordings into valuable marketing assets. By leveraging tools like Grain's AI, you can automatically transcribe and analyze your meetings, uncovering hidden gems of customer feedback and testimonials.  This process not only saves time but also ensures you don't miss out on crucial insights. Instead of manually sifting through hours of content, AI can quickly identify key themes and standout quotes.

The transformation of raw meeting data into actionable marketing content is a game-changer for businesses looking to showcase authentic customer experiences.

Comparing ChatGPT and Claude Capabilities

When it comes to AI analysis, both ChatGPT and Claude offer unique strengths. Here's a quick comparison:

While ChatGPT excels in speed and flexibility, Claude often provides more nuanced analysis and better quote extraction. The best choice depends on your specific needs and the volume of data you're processing.

Consider using both tools in tandem for optimal results. Zapier's Claude integration can help you seamlessly incorporate both into your workflow.