They may be considered sleek and some would even say sexy. But in more ways than one, Siri and Alexa do not have the skills to be contact center agents.
Thanks to Anthony Scodary of Gridspace for sharing his insight at Customer Contact Week.
Here is an edited transcript of my interview with Anthony:
Jim Rembach: Hey, this is Jim with Call Center Coach and the Fast Leader Show. And I’m here with Anthony Scodary of Gridspace. Anthony, how do you help folks get over the hump?
Anthony Scodary: So, we build technology that understands conversational long-form speech from scratch. So, everything from what people say, why they’re calling in, and also what they mean. But additionally we listened to how they say things, the emotional content of their speech, the cadence of their speech, the prosody, and pull that all together to talk about how a call went, whether the call concluded properly and what can be done to improve those interactions.
Jim Rembach: So when I think about the understanding and the interpretation piece, there’s a lot that goes into that. And there’s a lot of companies that are here talking about call guidance, AI, chatbots, when you start talking about what you’re doing, how does all of that fit?
Anthony Scodary: Sure. So, the two things that we do (Gridspace) that I think are relatively unique, one is we build everything in-house, everything from telephony to speech recognition, down to NLP and search. And the reason we do that is partly because we specialize on just conversational speech in the contact center. And so, all our models that we build, all our machine learning models are all trained on realistic call center conversations, oftentimes decades or centuries of data of people talking conversationally. So the model is sensitive to realistic call center data, rather than traditional speech recognition systems that were built for command driven systems like Siri or Alexa.
Jim Rembach: So, when I think about you saying that, I start thinking about…oh my gosh, eight months to deploy, seven figure type of investment, I now have to forklift everything–that’s what starts running through my head. Is that’s what we’re talking about here, a professional services type play?
Anthony Scodary: No. The integration path is actually several, they are very simple. One, you can just fork SIP traffic to us through the cloud, it’s a one-day integration. If you’re able to adjust your SPC to do that sort of integration or your cloud telephony provider. Another choice, is because we have no external dependencies everything is provided by Gridspace. We can literally just provide a box that doesn’t need the internet, if you want an on-prem solution, especially if you have maybe privacy or data concerns that require you to do that. But both of these paths are very simple integrations.
Jim Rembach: When I start thinking about the work that you’re doing, gosh, I can think that it can be quite complex, but ultimately we need to be able to get it back down to something that’s actionable and impactful, so the supervisor is a critical part of that happening. So how are you making a supervisor’s job easier?
Anthony Scodary: Well, if you think about it, a supervisor, no matter how prolific they are, they could only listen to a very small percentage of the calls that their supervisees generate. And a lot of times the questions they want to ask are aggregate questions, how often does this occur? Show me the one or two times this needle in a haystack occurred so that I can make or take a corrective action, whether it’s for coaching or training or for compliance enforcement.
And so, from the supervisor’s perspective, especially if they’re managing a large number of agents, it’s very important to have tools with very high accuracy, 95 plus percent accuracy identify exactly what they’re looking for. And that’s because they’re not machine learning experts. They need to be able to define declaratively what they want to search for in the call center.
And the way we’ve done that is we built something called scanner, it’s a semantic search query engine, and we have a builder that you just describe kind of the story of a call. This happens and then the agent does this, and then the caller does this and the neural networks do all the heavy lifting to figure out, okay, let’s compress it down into something that actually can be searched against. Hundreds of millions of calls in some cases instantly, interactively.
Jim Rembach: When I start thinking about what you’re saying is, am I looking at this as a quality assurance tool or is it a coaching tool, or is it like, yes, all of that?
Anthony Scodary: I don’t think we’re in the case where every single major problem that a call center can be solved using language tools like this. But I think for quality assurance, for compliance, for coaching, it’s an excellent tool, it’s also completely real-time.
And so, we actually monitor the call as it’s happening. So, whether the agent or their supervisor or the call center manager or the CEO of the company that’s being represented wants to monitor a call they can hop into the call, they can see the transcripts being generated. You can write tens of thousands of alerts if you want, defined semantically. Anytime someone is angrily talking about our TV commercials or every time someone is really relieved and talking about a loan extension and get an alert as soon as that happens. You can build things around that, inform the agent, I was like, okay, well this obscure case that we couldn’t train you for and our short training period has occurred you need to know this, this is our institutional policy.
There’s no way we could have trained you and know this in advance, but because we’ve programmatically coded it in our alerts it takes a lot of load off of training of the agents because you can just focus on the core stuff. And then, whether through coaching or through automated systems help them get through these kinds of edge cases or challenging issues whether it’s compliance or customer service or just representing your company.
Jim Rembach: So Anthony, how do folks learn more?
Anthony Scodary: You can learn more, you can go to gridspace.com. We have a lot of information videos and, white papers that explained what we’ve done for our customers. This year we’re processing about 2000 years of conversational speech. It would take us 20 centuries to listen through all the audio that we train our models on and we use to build our systems.
Jim Rembach: Anthony, thanks for sharing your knowledge and wisdom and we wish you the very best.
Anthony Scodary: Yeah. Thank you very much.
Please Share
- [VIDEO] Why You Shouldn’t Hire Siri and Alexa in Your Contact Center – Click to Tweet
- [VIDEO] Siri and Alexa are not SLEEK and SEXY enough for the contact center – Click to Tweet
- How to leverage conversational speech in the contact center – Click to Tweet
Additional Resources
- WEBINAR: 3 Common Questions Contact Centers Should NEVER ask about Speech Analytics
- WEBINAR: How Coca-Cola® Adds Life to Contact Center Employee Engagement
- WEBINAR: How Do Your Call Center Supervisors Measure Up?
Jim Rembach is the Editor in Chief of the Customer Service Weekly and it’s Podcast host. He is President of CX Global Media and the creator of the Call Center Coach Virtual Leaders Academy. As the host of the Fast Leader Show Podcast, he has interviewed hundreds of experts, authors, academics, researchers, and practitioners on various angles, viewpoints, and perspectives for improving the customer experience. He has held positions in retail operations, contact centers, customer support, customer success, sales, and measured the customer experience. He is a certified Emotional Intelligence practitioner, Employee Retention Specialist, and recipient of numerous industry awards.