The AI Fix for Vanishing Customer Journeys - Tata Communications

Christopher Carey sits down with Gaurav Anand, Global Head of Customer Interaction Suite at Tata Communications, to explore why customer journeys break down in real enterprises—not because of channels, but because backend systems aren’t interconnected and critical context gets lost. Gaurav unpacks the “customer journey black hole,” explains why AI pilots are creating POC fatigue, and outlines the need for an “AI operating system” that connects systems, restores context, and enables multi-agent coordination.

The conversation also covers what true “journey completion” means operationally (relevant context, low latency, and strong guardrails), how AI agents differ from bots and IVR, where TX Hub fits in the enterprise architecture as an orchestration layer on top of the contact center ACD, and why governance controls like audit trails and compliance are non‑negotiable as AI starts taking action across systems.

Transcript

SPEAKER_01 0:07

Hello, welcome to CX Today. Today we're looking at a problem many enterprises feel but struggle to many. Customer journeys that disappear between channels, systems, and teams. If you're trying to improve CX without ripping out your platforms, you're going to want to stay tuned to this. I'm joined by Gamarov Ahmad from Tatama Communications. Gamarov, uh thanks for joining me today. How are you?

SPEAKER_00 0:33

Great, Chris. Thanks for having me again. Good to see you again.

SPEAKER_01 0:37

So just in a sentence or two, how do you explain what TabTama Communications is trying to solve in CX right now?

SPEAKER_00 0:47

Yeah, Chris, just building up on what we talked about last time, right? I think what we're trying to do is kind of build intelligence at the core. So we're going from what we said, you know, fragmented experiences and siloed enterprise systems to more AI-led, context-aware engagement on the customer side and within enterprise productivity, to really honestly reimagine how humans, how machines and agents sort of collaborate, execute, and deliver outcomes securely and at scale.

SPEAKER_01 1:29

And where do customer journeys most often break down in real enterprises?

SPEAKER_00 1:36

So I think the best way to start to answer that question really is from looking at just the customer journey itself, right? So it's not the channels that break, it's the systems that break. And I think one of the things that, you know, I'll emphasize a lot in our conversation today is about the lost context, right? And that's really where things are starting to break down. And by context, I mean uh not just the past history of the customers' interaction and the decisions that they took in terms of what their buying patterns were, not just looking at you know what products they were interested in or what products the brands are trying to promote, but also looking at what are some of the decisions, some of the unstructured data that kind of exists uh in the systems that, you know, systems of record, right? And and in in um and systems of knowledge that exist within an organization where you know certain decisions may have been made in the past about what type of discounts are are agreeable to a company or what sort of credits they can give away, or what sort of changes in policies that they can make, and so on and so forth. So there's some subtle you know differences and and those kind of get lost where you know a an interaction begins, it goes well up to a certain extent, even if it's you know cross-channel, but then it fails in delivering the final outcome because it becomes too complex, or because the systems of the back end are not as interconnected, or the context is lost. So those are some of the main problems I think we're finding today.

SPEAKER_01 3:08

It's an interesting way to frame it. And how does fragmentation, what does it do to customers and agents day-to-day?

SPEAKER_00 3:19

So I think one of the biggest uh uh problems that that we see when it comes to uh fragmentation is that you know, systems in the back end, like I was saying, are not um uh interconnected, right? So, for example, a customer might begin a service request through a voice or a chat interaction, um, and the request may need an interaction with an order management or a billing or a fulfillment system, and that workflow may not be as well connected. Uh, you know, this is something we discussed in our previous conversation as well, right? Where the customer journey black hole, uh, it's a point where the customer intent, the customer context or their history, you know, effectively disappears as the interaction moves between systems or teams, right? And when that happens, right, then um AI may understand the request perfectly, right? But the back-end execution layer is disconnected, right? So the conversation then gets handed off to a human agent who often lacks the full context and that transfer is not happening. In other words, you know, you the journey doesn't really quite get to the outcome that it was intended to achieve. So um and then that gap really exists, you know, because conversational intelligence is is not connected to execution intelligence, right? Uh and when those two layers are are disconnected, uh then the experience for both customers and agents kind of breaks down.

SPEAKER_01 4:45

So interesting there, what you said about the the disconnect. I mean, is this more urgent today than say one or two years ago?

SPEAKER_00 4:55

Yes, absolutely. It is. Um I think first of all, what's happening is that you know there are a lot of investments folks are making into AI, and particularly, let's say, voice AI as an example. And and those types of investments are uh kind of spotty in nature right now. There are pilots that are running in you know uh pockets, and and and not only are these pilots running in pockets, there are agents that are running within these pockets that are running sort of independently. Um and and I think there is a little bit of uh pilot POC fatigue that's you know sort of going on, and there's a lot of restlessness to kind of get a return on this AI investment. Um so realizing the value of AI and having the ability to you know convert these pilots to scale and to actually have the impact across the organization, amongst all of the different ways in which these agents can coordinate, not only within these specific you know, areas, but also between each other. So kind of become a multi-agentic uh framework. Um, I think that is where um you know the urgency is coming from, is that restlessness to sort of get to you know that value realization. Uh, and I think where uh we are thinking uh that the companies and brands are looking to find a solution is to almost look for an operating system, like an AI operating system. It's something that we're uh sort of promoting quite heavily now. Um because for us, it's that interconnection and the bridge between the different systems and connecting them and bringing that context layer back into play is what really is the solution to you know getting to that sort of uh uh AI value realization stage.

SPEAKER_01 6:54

You mentioned there just about the uh bridging. I mean, what does true uh journey completion mean operationally?

SPEAKER_00 7:04

Uh that's a good question. Uh because you know, typically and historically, you know, journey orchestration was, you know, hey, let's just make sure that you know the context is you know not lost and we have all the history and the intent is captured. Um but operationally it's a lot you know different. So um it it's it's not just capturing the intent while there is the interaction uh ongoing live, uh, but also knowing how to deal with that interaction. So the history matters, but relevant history matters, right? It's uh a lot of times the what we're seeing is is that there is you know um uh generative uh AI type solutions that are getting put in place that are wasteful of not only uh the resources, the GPU resources, but also from a latency standpoint, and getting to the you know responses in the right time and in the right um you know context. Um and then the connectivity between like I've you know was just talking about you know the downstream systems, uh but but also very importantly, as I mentioned in my you know starting statement, it's it's not just driving AI-led context-aware customer engagement and productivity, but it's also doing it within the right guardrails. Right? You need to have the right audit trails in place. So operationally that becomes really relevant to make sure that you know there is that you know uh oversight and and there there is that trackability or the audit trail that you could you could you could have while at the same time keeping the experience at a level where you know it's low latency, it's relevant, and it's timely, and it's more effective.

SPEAKER_01 8:57

Just just expanding on one thing you said there. How should leaders think about AI agents versus say bots or IVR?

SPEAKER_00 9:08

Oh man, that's a big that's a good question because uh that's pretty um uh fundamental, right? In the sense that look, IVRs uh back in the day, uh there's still IVRs around, but they were very rule-based, right? And uh it's decision tree, you say X, you go to Y, and so on and so forth. Um, and and then traditional bots, you know, were a little bit more uh I would say uh flexible, but I won't uh argue that I would argue that not by much. Uh they could use like common questions, so they'll have a FAQ that would you know uh be fed with and then standard you know answers and what very reactive in some ways, right? But AI agents, on the other hand, right? I mean, they're uh very context aware, number one, right? And then I told you it was used context a lot in this conversation because it is fundamental to what you know our approach is, and I think fundamentally what enterprises are looking for. Um but they're also action-oriented, right? So the agents they interpret the intent, they understand the prior interactions, if they have the right, like I said, AI operating system, feeding them the relevant information in a timely manner, in low latency, low resource use, in a safe and guard-railed, governed fashion within the policies of the enterprise. I I mean, there is uh you know a progression then from not just being able to answer questions or produce or generate information, but also then go to the next level of completing the task, taking the actions. So the leaders you know can think of you know IBRs as you know handling the routing of a call, bots possibly handling FAQs, right? But AI agents handle outcomes.

SPEAKER_01 10:52

And one thing you mentioned there just about the approach. I mean, I'm sure our viewers will be interested to know where does TXHub sit in the enterprise architecture.

SPEAKER_00 11:05

Yeah, that's a great question. So uh so where we sort of look at uh you know CX journeys, like if you think about it, you know, there's there's the outbound uh side of things, right? Where you're making calls and and whether it's marketing outreach or uh you know um uh lead generation, uh you know, uh whatever demand uh generation calls and so on and so forth, or appointment reminders, whatnot. But then there's the inbound side of it. Um and the outbound ends up being a little less, in my opinion, productive because it's a bit of let's throw something against the wall, unless it's very specific task-oriented outbound call. Um uh, but inbound is is more valuable, right? Because that's where the customer is calling in, and I think that's when you really want to make sure that that call is treated uh or that interaction is treated with the best uh you know um uh outcome that we can achieve for that interaction end-to-end, not just from the experience of the interaction, but also how that you know ends up with the final outcome. Um so um quite honestly, um when it comes to TX, I think there's a combination of you know using AI agents in our what we call our you know voice AI capability, right? So with the commotion acquisition that we recently did, um we are building, we have built a voice AI, which is a speech-to-speech um uh uh solution. Uh, that on the front end is taking a call and it's integrated now with our platform, which is what we call TX Hub, which is Total Experience, because it looks at not just the customer experience, but also the agent experience. And it's really acts as an orchestrator between the AI agent and potentially the uh human agent, right? And that orchestration between those two, uh, whether it gets to the human agent or not, the TX hub essentially is sitting on top of the contact center uh ACD, if you would, and is potentially um looking to automate the workflows that are associated with this agent that's talking to the voice AI agent that's talking to the end customer, helping and aiding with them with the context, making them make the decisions, but also at the same time referencing the knowledge bases and allowing for that agent and that transcription to kind of flow through within whatever the brains of the organization is, or whether it's within commotion or another CRM, to have that sort of uh a flow of information go through, right? So TX offers the ability to orchestrate between workflows, has the connectors to all of the different CRMs that are out there, obviously also to commotion. So, in context of inbound calls that are coming in, addressing and combining inbound with voice AI, so you can get to self-serve, lower the you know, initial sort of uh deflection, lower the average whole time, yet at the same time provide the bridge or the orchestration layer for the agent to be communicating with the brains of the organization, throwing out whatever the context is, what is the next best action, what is the you know, uh the the the generative sort of context uh aware outcomes that you know um and and information that the agent can be sort of surfaced with. And then when the call and if the call gets transferred to the human agent, to actually then allow for that context, that intent, all of it to then be transferred to the human agent, so it's not a start of a new conversation, it's a continuation of it. So it's a very powerful uh combination of that trio, if you would, of the voice AI substituted with uh complemented by the orchestration layer, which is TX with the connectors into the CRMs and then into commotion, and then having that whole sort of uh flow uh be orchestrated uh uh in a manner that's that's that's truly effective.

SPEAKER_01 15:14

And just picking up on something you said there, just about the the area of AI, I mean, uh as AI starts triggering actions across systems, what governance controls become non-negotiable?

SPEAKER_00 15:29

That's uh a fundamental uh question, right? I mean, I think at the top of mind for uh almost uh every enterprise, right? As we start to get into you know more and more agents not only, as I said before, generating information, but taking action, right? And if those actions, if you think about it, are you know, hey, there's gonna be a change in the contract, or there's gonna be, you know, we're gonna update uh the system and give you a certain amount of credit. Um, or you know, it runs through um making changes to um uh you know uh future policies and how you handle. So there are there are different agents that are talking to each other that are making changes, and you need to be able to trace and you need to have an audit trail, like who did what when, and then you need to have guardrails and what they can and cannot do. Um so so it it definitely you know from a governance standpoint, measuring um you know uh the effectiveness of the agents is critical, right? Um compliance to certain rules and regulations and certain geographies is paramount, right? So have that be built in. And this is where the AI operating system comes in, right? So you can build in these rules, and and these this AI operating system is not sort of like kind of you know one uh for all, um one size fits all, but it's it's bespoke, right? So every you know organization can you know essentially build that um uh for their own set of rules and their own set of LLMs and their own set of SLMs and so on and so forth, right?

SPEAKER_01 17:02

And just one final question. How does the the the Commotion acquisition strengthen what you're building?

SPEAKER_00 17:11

Right. So uh see, right from the get-go, I think we've been working with Commotion for some time and then they finally uh acquired a 51% stake in December. Uh fundamentally an AI native company, um, they're doing a lot of AI um uh development for us, right? So they've built the one of the leading voice AI uh solutions. Uh we're looking at some independent benchmarks, but at the moment, you know, from a latency perspective, uh they're top of the charts, um, multiple languages, you know, 40 different languages so uh that that we can you know sort of have uh uh the the the voice agents uh converse in. Um and what's more important is that that AI operating system that I was telling you about, essentially uh you know, some of it is built on our own infrastructure, but some of it is also available through from an infrastructure layer standpoint through the public clouds. Um we can do private instances, dedicated instances hosted in customer premises and so on. But essentially, uh, you know, building that framework of AI and making it into a uh a multi-agent uh environment that actually looks at the context layer, which is like sort of getting the connectors in place to make sure that you know all of the different systems are integrated. So you have the access and you pull in all of the context and the the structured data, so to speak. But then also we go into, like I was saying earlier, into the unstructured part of the data, right? Which is sitting in our in the enterprise system of you know records, knowledge like Teams and uh you know notes in in uh CRMs and so on, picking that up, taking notes out post-call, right? All of those uh you know elements that actually add now back to the feedback loop, right? So all of that uh sort of uh capability uh feeds into uh you know the the orchestration layer uh of all of these different agents, which you know, voice AI obviously is one of the big ones, but uh uh you know, we're cutting across the entire enterprise and we're looking at agents that are helping from an HR perspective, legal and whatnot. And and and quite essentially, right, this combination of what I was saying before with the voice AI, TX, and and the ACD, and quite honestly, the ACD can be you know anybody, right? Because TX sits on on over the top. And and uh, you know, so while we do have a CCAS service as well, but we don't need really need to be managing that CCAS platform, it can still um you know be connected and be you know the orchestrator uh on top. So combination of all of that, uh the AIOS uh is is a fundamentally new concept that um you know we're we're launching uh that you know commotion allows us to uh you know kind of leapfrog uh into uh sort of bringing to the enterprises. A lot of conversations at the sea levels that we're having are you know they're taking this you know quite to heart because like I said before, there are a lot of pilots in islands that are happening, uh point solutions, a lot of experimentation, POCs. But to get to that scale, uh I think is where you know we think that commotion is helping us, you know, and the enterprises to uh get to uh uh the the outcome level uh with the guardrails and security and our channels you know sort of integrated into it to give like a full holistic story. So everything sort of comes together in a very nice way, and commotion driven by you know being AI native is is really helping us to execute sort of the end-to-end tasks for for uh for a lot of these uh uh workflows.

SPEAKER_01 21:06

Sounds interesting. Looking forward to see what's gonna happen next. Garov, thank you very much for joining me today. Um, if people want to get in touch, what's the best way?

SPEAKER_00 21:17

Well, the best way is to look us up on our website, tadacommunications.com, uh, or you're welcome to look me up on my LinkedIn. I'm always uh available to talk to you guys personally as well. Thanks very much, Garov.

SPEAKER_01 21:30

If you like this video, please give us a like and a share on social media, and we'll see you next time. I've been Christopher Carey for CX Today. Thanks for watching.