How Privo uses Amazon Transcribe to deliver a better customer experience

Speech-Transcription-Audio-Video-to-Text-FormatAuthor: Simon Kohnstamm

Things move fast at Privo. We talk to customers all day long about their exciting projects on AWS, and it seems the pace just keeps accelerating. By the end of the day, the conversations can blend together and it can be difficult remember exactly what we discussed with each customer.

Like many sales teams we enter notes about our phone calls directly into our CRM. This helps tremendously but it’s not perfect. First, it’s nearly impossible to capture all of the details from a call. What’s worse, typing notes is a huge distraction. It significantly detracts from our ability to really listen to our customers. We knew there had to be a better way.

At Privo, we feel that if anything needs to be done more than once, it should probably be automated. In fact, our entire sales cycle is highly automated: from scheduling meetings with customers, to the creating Scopes of Work and sending follow-up emails.  Automation not only allows for these activities to be repeated in an error resistant way, but it helps us to be more scalable. But how to automate note taking for sales calls? Enter AWS Transcribe.

The process works like this:  We first record each customer call using our VoIP phone service RingCentral. At the conclusion of the call, a Zapier script is triggered which uploads the call recording into the customer’s folder on Google Drive, and to an S3 bucket for processing. Once uploaded to S3, a CloudWatch event triggers a Lambda script which initiates the Amazon Transcribe process, including AWS Transcribe settings, such as number of speakers, and audio sample rate which allow the transcription to be more accurate.

Once the AWS Transcribe job is finished, a CloudWatch event triggers another Lambda process which stitches together the JSON file produced by Amazon Transcribe, and produces a transcript which looks like this (these are direct call transcriptions from the service but I've trimmed them for sanctity):

spk_0: alright so uh got some of them at least notes from your conversation the other day about what you're looking to accomplish on AWS..

spk_1: yeah I mean we're uh we're a provider uh within the financial services industry um so you know we already have some stuff ah in a database... could we move some of our local file email and some of the services as well um and if so you know how how can we do it in a in a secure manner...

These call transcripts are then automatically uploaded to our CRM so other sales engineers know exactly what was discussed. When the project is ready to be handed off to our engineering team, they can get the full context of what was discussed beforehand!

This solution helps us listen better to our customers and never lose track of the important details uncovered during those initial conversations.  And, importantly, it happens the same way every time with no effort.

About the Author

Nick Underwood

Nick Underwood has over 15 years of experience supporting IT infrastructures for businesses across a broad range of industries.


    Get Your Project Started With a Free Consultation

    Get a Free Consultation