Conversations in Collaboration: Zingly.ai’s Gaurav Passi on Disrupting the CX SpaceConversations in Collaboration: Zingly.ai’s Gaurav Passi on Disrupting the CX Space

Welcome to No Jitter’s Conversations in Collaboration, in which we speak with executives and thought leaders about the key trends across the full stack of enterprise communications technologies.

In this conversation, we spoke with Gaurav Passi, CEO and co-founder of Zingly.ai (Zingly). Prior to Zingly, Passi was the President of Cloud Business at Avaya Inc., and was CPO/EVP of Products, Technology, and Marketing at Five9. He also spent over a decade at communications software and service provider Amdocs, most recently as Vice President of Products for the Customer Experience business unit.

Source: Zingly.ai

In this conversation, Passi provided some context for where his company fits in the CX landscape –nestled in the confluence of three major solution categories: systems of engagement, systems of record and systems of action. Passi says he launched Zingly, in conjunction with several other partners, because the CX space needed to evolve. “The industry needs not just incremental help, but disruptive help,” Passi said. “If disruption is not needed, then startups should not be born to begin with.”

No Jitter (NJ): Can you please provide some context on where Zingly fits within the contact center space / customer experience space?

Gaurav Passi (Passi): There are many different players in the customer experience sector. One is the contact center players; these are the companies providing technology for telephony [who mostly came] from an on-prem background. Then, the cloud companies [came in and] said, ‘I can move you from capex to opex.’

From a technology standpoint, the best way to understand [contact center as a service] players,’ is that their ‘system of engagement’ – how you engage with them – is live communication.

Then there are the ‘system of record’ players: they have all the historical information of the customer residing with them – in the form of cases, opportunities, leads or the like. Another area is what we call the ‘system of action’ companies, like a ServiceNow, which provide workflows.

Lastly, and I think it's more an adjacency, are the unified communication players like Zoom, Webex, RingCentral, Dialpad and these kinds of companies. They are built more for communication, but they're also system of engagement designed for employee-to-employee communication and maybe some known customer communication.

[With that context], over the last 10 or more years we’ve all gone through, and are still going through, a digital transformation where we can see our bills and the like online. That really caused traffic to move away from 1-800 phone calls and into digital traffic – apps, Web sites, etc.

At a high level, there’s about ten times more traffic on websites and apps as compared to voice calls. As an example, one company we recently met with gets about 250 million unique visitors on their website every year. Of that 250 million, about 50 million authenticate themselves, meaning that they tell the company who they are. Of those, 12 to 14 million [access] their support portals for customer service, sales and onboarding.

By contrast, that company gets about 3.8 million calls to their call center. So, [there’s a] dichotomy in how their traffic moved online while the calling slowed down.

When someone calls a call center, the clear data we have now is that 55 to 60% of people who were trying to get help on the support portals did not get the help [they needed], so they ended up calling the contact center. That helps explain why they are disgruntled by the time they call in. And when they do call in, they punch numbers [on the IVR] and wait on hold and end up more frustrated. And then the brand tries to sell them something when they have a problem? Good luck.

That's where the CCaaS market has absolutely hit a brick wall. The CCaaS players would not say it, but CCaaS is dead, at least the CCaaS in the way we understand now.

NJ: OK, so where does Zingly fit?

Passi: I mentioned the system of engagement being CCaaS and the system of record being CRM companies and then the workflow companies. The funny thing about the CRMs is they do not understand the real-time world. They only know ‘case object’ when the case is created, [for example]. They know historical information. They don’t know real time information.

The CCaaS industry has always understood the real-time world, but what happens in the real world is that after the phone call is done, agents get about 90 seconds for their after-call work. So, they go to the CRM system and, based on what they understood during the call, enter their notes. And then 90 seconds later, the CCaaS serves them the next call. So the ‘system of engagement’ players do not have a ‘system of record’ – [other than] the agent who’s actually entering the notes, the disposition of the call, etc., into the CRM solution.

Now, most people in this country don't want call center jobs, and even if they wanted them, companies can’t afford them because call center jobs are expensive. So, companies offshored those jobs to BPOs in the Philippines and India, etc. The cost per interaction did decrease by moving call centers to BPOs, but they [the BPOs] never understood the business. Companies have figured out that even going from $10 cost per interaction down to $5.60 cost per interaction is still not enough, because their [customers are] still disgruntled while [their company] board meetings are about generating more revenue.

And companies are struggling with lower CSAT because customer patience is low – they want help now, for example, so after hours service has become a problem, etc.

CCaaS players are hitting a brick wall because while their customers are moving to the cloud, those companies [their customers], still have the same [human agent] head count. That's where the world is becoming a very interesting place. CRMs are seeing the opportunity – if there’s 250 million people on the website, then maybe they can offer a chat solution for that website. But they don't understand the world of real time, so this is where the industry is at an impasse.

Zingly was created to do a ‘consumer inversion,’ meaning that if we can help customers self-serve then they can be self-sufficient. They don't need a person unless they need a person. When customers ask a brand a question, and they pause and come back within 10 minutes or even the next day, they don’t want to restart [from the beginning]. They want the brand to remember everything they’ve talked about – and do that in a collaborative space in which they can engage with the brand, which is not an 800 number.

The first thing Zingly did is provide an interface for those consumers. Second, we went after the Fortune 500 companies, as well as those in highly regulated markets like financial and healthcare companies, because they're very complex in nature. We went after them to solve not only the problem of scale, but also the problem of security and risk.

Third, and most important, is [changing] call centers into profit centers. We do that by [introducing a lot of] service automation, meaning that customers don't need to talk to a human being for their low value or repeat interactions. We use AI for those and by doing that we help companies move their budgets into more revenue creation opportunities.

(Editor’s note: In an email follow-up, Passi said that because Zingly’s AI can dip into customer information – which he discusses in more detail below – it can make revenue-maximizing recommendations to the customer. That is, “the AI opens the opportunity and assigns it to an appropriate sales rep (loans, for example) or can carry out the upsell itself, depending on the product/service.”)

NJ: Which brings us to Zingly Rooms. Can you explain what it is and how it works?

Passi: There are multiple entry points to a Zingly Room and people can come from any one of those. The first entry point is a brand's website, one where you haven’t signed in and you’re just looking for answers, shopping, buying or onboarding, etc. We call that pre-authenticated. The second is you're authenticated; the brand already knows you. The third is a support portal – a web site, an app – where you are an existing customer and you’re helping yourself.

The last piece is traditional calling. If somebody calls, Zingly is the voice bot. It understands from your phone number who you are and, because it's generative in nature, it converses with you. It understands your emotions. You can interrupt it. But what it does is immediately create a multimodal Room for you.

NJ: So if I’m calling from my mobile phone the Room is sent via text message?

Passi: Yes, that’s one way. We can send a text message and say, ‘for more information, click here,’ and once you do, the Room opens up – and it is secure…SOC 2, fully authenticated, etc. If you’re on a website then Zingly can be a search bar at the bottom of the page [pictured below] which you can use to ask questions [via text] and it [supports] a voice bot, as well. It also welcomes you with the most commonly asked questions. When you do start asking questions, an AI Buddy comes in to assist.

Source: Zingly.ai

NJ: And once in the Room, how does a customer escalate to a human if they need to?

Passi: In the past, people who called in would say ‘agent, agent, agent,’ right? That's the clue [laughs]. People find ways to cut through this stuff, and they will do the same with AI. The opposite is also true; the customer [can] spend too much time with AI when they actually need to talk to a human.

We have a ‘relationship AI’ that understands the natural language, the sentiment, the emotion, etc. And, behind the scenes we have a rating system that knows the confidence level associated with the answers [being provided] to the customer.

Every conversation [with a customer] has a goal. For example, if you want to do a 401K rollover, and you're asking questions about that and you're moving toward your goal, [we know that] based on our rating system – the answers they’re getting are moving the customer toward their goal. Because we know that, we won’t bring a human rep in. But if the conversation is not going toward the goal, our rating system will also know that, and we will then bring a human in the loop.

But what if it's a high value customer coming to your site? Are you going to throw AI at them, or are you going to throw a red carpet in front of them? Businesses can decide differently for themselves, but we have that ability to know it's a premium customer coming in and therefore the business can choose to treat them differently and maybe bring a human in right away.

With Zingly, that human in the loop can talk with the customer on any channel – voice, video, text – and [other media, like] screen shares, media share files, even DocuSign payments – all of that comes together in one single place, the Room. Most importantly, the Room is persistent, meaning it does not leave you. The customer starts and if they come back in two hours, two days or in two years, the data will be there. They can pick up where they left off.

And then when the Gen AI Buddy is in the Room [shown below], it can interact with the customer and because the data is persistent, the ‘buddy’ has more information and more content, not only based on the data it already had, but also from the answers it provided. It also knows the goal the customer is headed toward, understands the sentiment and understands where in the workflow you are.

Source: Zingly.ai

NJ: Where is that Room data stored? How does it work on the back end?

Passi: Everything is in the Room, structured and unstructured data alike. If you just had a call, it will get summarized, and the summary of the call shows up along with any action items or reminders. If you were applying for a mortgage and you had to send a paystub, there’d be a reminder in the Room to upload that pay stub.

We don’t store all the different collateral, sales materials, FAQs, knowledge bases, etc., in a Room. We built an integration framework – like an ‘infrastructure as a service’ – that enables access into [a company’s] tech stack whether it’s on-prem or not, unstructured or structured data.

That’s where the [retrieval augmented generation] comes in. We start with the use case the customer wants and then, based on that, we integrate to a knowledge base, the CRM and then, for example, the core banking [system], and then we build the RAG.

Now, some of that data does not need to be stored all the time in our system. So, to serve the customer we grab the data [we need] and then store the responses [to the customer] so that we can have fast access and lower latency in the future. So, the Room has the answers that were served to the customer – or the actions the customer, or company, needs to take.

CRM companies think about the world [in terms of] historical data. After the call is done, they store the data and then do something about it. They’re missing the real-time nature of it. But the system of engagement companies have always thought about real time. I think the world is headed toward this combination of the ‘system of engagement’ with the ‘system of record’ – and that’s what Zingly has done.

NJ: Can you talk a little bit about how Zingly is working toward flipping the script from cost center to revenue center?

Passi: That's the base reason why we exist. We don't exist to go to companies and say, ‘Hey, we can bring your cost of interaction down.’ That's just a foot in the door that everybody needs. What they really need is, I want my customers to grow. I want my CSAT and NPS to grow, and I want to make more revenue. I'm not a dot org, I'm dot com – a profit-making company.

There used to be clear lines between a service call and a sales call. When a customer called for service, maybe they really needed to buy something [to solve their issue]. In the past, they’d call and get a service agent who didn’t know how to help them. So, the customer was unhappy. The second path was that maybe they got lucky, and the service agent did understand so they transferred the call to a sales rep – but that salesperson had no empathy about who they were as a customer and then started to ‘hard sell’ them on something. That also resulted in an unhappy customer.

What our technology does is understand that the lines are getting blurred. It could be a service call, but it could also be a sales call. It can [detect] a cross-/up-sell opportunity. This is where I think a lot of the excitement is with Gen AI and agentic – not just at the scale of what it can do, [but also] with [respect to] reducing the cost while also improving the NPS.