“AI Product Management for the Enterprise Consumer”: Savita Kini, Director of Product Management, Speech & Video AI at Cisco (Video + Transcript)

In this session, Savita Kini discusses the emerging area of enterprise consumerization and the impact of AI interventions in both enterprise and consumer settings. Kini highlights the three layers of transformation happening in AI product management (PM) roles in the enterprise, and discusses the opportunities and challenges in leveraging AI in the enterprise, including the need to balance personalization with privacy concerns.

Transcript:

Savita Kini: Hello. Thank you everybody for joining and good morning, good afternoon, good evening, wherever you are. I’m going to talk a little bit about a new emerging area around enterprise consumerization, and there is also AI interventions that are happening both in enterprise and consumer. So there are three layers of transformation that’s happening to the AI PM roles. And in the enterprise, how that’s changing, along with the consumerization of the enterprise. So there’s couple of themes and I’m trying to go through it quickly.

Okay, so what is really enterprise consumerization or the enterprise consumer? I think one of the things that happened over the last decade with the consumer apps is all of us who work in enterprise have expected that same kind of personalization of experience; like how we use app, how we do our performance reviews, how we file our expenses in the enterprise. How do we collaborate with our colleagues? That trend was already happening, even before the pandemic started. And then the pandemic happened and all of us worked from home. We were extremely reliant on the network, on the collaboration, talking to our colleagues via chat. And over the last two, three years, really that whole trend that began before the pandemic only got accelerated.

And then what happened? We had the whole LLM and generative AI explosion, and we are now getting into a whole new generation of enterprise SaaS where we as enterprise users and you, me, all of us, right? We want that same simplicity, the delightfulness, the creativity, the intelligence, the personalization. All of this experience that we see in the consumer domain, we want that in our enterprise experience, but not at the cost of losing the privacy: privacy of our customer data, privacy of our employee data. So there is a very unique transformation that’s happening, and I’m going to speak to it from the perspective of an AI PM in the enterprise.

I have a couple of little nuggets of transformation data that some of the research firms have been talking about. So like I said, hybrid work is here to stay. The Future of Work Research from IDC has put forth some very interesting data points around how our offices are transforming in the post hybrid, post pandemic era, because workplaces are becoming interesting watering holes. We are not going into work for … And this is true much more in the IT sector and since this forum is of women in tech, I will speak to it from the IT sector. It does not maybe apply to education or healthcare or retail. I think I’ll touch upon it a little bit later in my presentation.

But specifically we are going into work more to collaborate with our colleagues. We expect our workplaces to be, again, something of a draw, but not for our regular work, not for our regular mundane jobs. We are going in to collaborate. We are going in because we want to ideate, we want to create. And how do we augment that experience? A lot of companies are spending, according to IDC, over trillion dollars just in 2023 to redesign those workplaces. Now that’s the physical, but how do you create that same 10x better experience when you’re working remotely? And I think those two different trends are kind of colliding.

Now, let me just go specifically into what’s happening in the enterprise. Now this transition of AI in the enterprise actually started before the current generative AI efforts, and so there was speech recognition. I mean we all know about Alexa, Siri, and so on. But there were voice assistants already in the video conferencing space. There were computer vision models in the video conferencing space as well. That’s some of the experience that I come from, so I can speak to it. But what’s happening with the natural language-based model explosions is that that whole transformation is only becoming even more pronounced.

And there’s a huge opportunity. I think a lot of the AI talk with ChatGPT and so on, you talk about all the new opportunity to create your own video, write your own storyline script. That’s still consumer, but how does that change how we work on a day-to-day basis? What productivity gains are likely to happen? And there’s a lot of prediction. You can see everything. Like I looked at it, I was looking at some of the numbers. They’re changing anything from 130, 155 to 200 billion dollars by 2030, and that’s like a huge explosion of investment.

So where are these investments really going? They are going in different categories around AI infrastructure, AI chip sets, and neural accelerators. How they fit into the enterprise infrastructure; software, which is again, enterprise software. And I talked a little bit about the video conferencing space, and the collaboration space is another one with the large language models that we are seeing.

So it’s a gigantic opportunity. And how are we all prepared for capturing that transition and making impact? I think those are the key themes here, as for AI product managers in the enterprise.

Just a quick note that this transition, again, did not start today. It was already happening. There were machine learning models being used to optimize IT for robotic process automation and manufacturing in healthcare, in pharma, lots of different places where there were smaller models and innovations happening. What deep learning is transforming is in sort of the cybersecurity space, further optimization of enterprise infrastructure, sales, and marketing. So that’s where we are seeing some of the newer more game changing innovations.

Again, just to touch upon some of the industries where generative AI is accelerating that trend, you’ll see a lot of innovations in legal services, consulting, consumer and retail. How we personalize the experience for end customers, for example, in retail. Personalized healthcare. You’re going to see a lot of this kind of innovation in the next decade, which is just kind of starting. We are in the infancy zone as some of these viability of some of these products and business models gets fleshed out. So we are still on the hype curve. We have to get to this mainstream, what you say, viable business models, viable use cases, viable experiences. Because remember back to the original premise, enterprise is different from consumer because of just the data privacy concerns. And I’m going to go a little deeper into that in the next couple of slides.

So where are some of the innovations like I talked about? So manufacturing, supply-chain, you’ll start see some of the automation that was already started, but how to predict and make that even more informed and more intelligent.

Where the enterprises have the biggest advantage, which is lacking in consumer, is really the data. If you think about consumer, like let’s say take Google example, or Alexa; they rely a lot on our data, what we have produced. Even ChatGPT for example, there are huge concerns about copyright violations. That ChatGPT is trained on content of the writers and it has not credited them for their contributions, right? It’s just using that data, crawling the network and internet, and just using it to train the models. And that’s not okay.

In the enterprise, however, we are sitting on treasure trove of data from users coming to our website, who’s coming, what are they buying? There’s so much information across the customer journey that sometimes today sales is not able to make informed decisions. What should I upsell? What should I do better? HR, recruiting, there are so many of these interventions that are possible. One of the data points, for example, that I was reading about Copilot is that it has increased 30% productivity for developers. Our hiring practices, how we [inaudible 00:10:26] candidates, how we interview, how we train our interviewers, how can we do that better to make the hiring process simpler, more ethical, and unbiased.

AI can actually help us. There is a lot of talk about how AI has influenced bad hiring practices because of the data, but the other flip side can also be true. It can help us in detecting our own prejudices and biases. I think that’s where some of the interesting ways in which AI can help us do better, is what I think are some of the interesting interventions.

Anyway. So the big advantage for enterprise is that they have treasure trove of business data, which can be capitalized hugely to create very customized experiences for both internal employees as well as their end customers.

I want to show a quick video here about just an example of how we are doing it in Webex. Hopefully this will play through.

Narrator: In today’s fast-paced world, collaboration is key. Bring teams together effortlessly with real-time communication, no matter where they are. From home to office, or across the world. AI powered interactions break down barriers and make virtual collaboration immersive. Integrated meetings, messaging, calling, and events give you the tools needed to reach a global audience. Easily manage from a single place for uninterrupted productivity. Experience the power of seamless collaboration with the Webex suite.

Savita Kini: Okay, so now let me talk about the gory details. I presented a nice view of what the opportunity is out there when it comes to collaboration, business workflows, sales and marketing, healthcare. But what’s unique and different about the enterprise use cases is enterprises serve two stakeholders, ideally speaking. It’s the customers and then employees. Employees help us build the best products to serve our end customers. Right? I mean, that’s the bottom line. If you have good employees, good culture, you create the best customer experiences. And so when you think of enterprise apps particularly, it would be for one or the other stakeholder primarily.

Now, the second thing that I want to highlight that I have learned over the last five years of dealing with AI in the enterprise, is the issue around data governance and privacy. Unlike in consumer where you can get away by doing things like ChatGPT, just crawl the internet and release something, in enterprise we can’t do that. Because we are governed by stricter laws, our customers expect. We sell to both public sector and private sector. Like for example, if you sell into the federal government, you have to go through specific certifications. If you’re selling into healthcare, you are going through a lot of healthcare related regulatory compliance and certifications.

And so there are very strict governance policies that enterprise software and hardware and infrastructure vendors have to adhere to. And so that flows into how the apps have to be developed when we create these experiences for the enterprise use cases.

The other question is training data. So if you are building an app and you are building it for an enterprise, how do you acquire the data? If I’m sitting, I don’t have the data of a large bank. I might not have exposure to the conversations that they have internally in their meetings. How do I create a summarization using an LLM? Those are very interesting challenges that are unique in the enterprise space. So you’ll see a lot more of large enterprises actually building their own AI tools and experiences. So the opportunity for AI PM in the enterprise is both from an external vendor, but also internally in large enterprises. You’ll see AI PMs coming in to actually help with their own internal business workflow and optimizations.

There are restrictions to using third party and open source tools as well. Like at Cisco, we have very strict guidelines and tools and processes as we build our products on what third party or open source tools we can use. Then finally, the complexity of AI models, how they are deployed, transparency and aligning to public and private sector concerns.

Finally, in closing, let me just say, the big transformation here for AI product managers is not only do they have to do the enterprise PM role, but also there is this whole challenge of balancing personalization versus privacy when it comes to AI models. Because unlike in the consumer, in the enterprise, we have to disclose what we are doing with our models and what models are built into our features.

I know I’m running out of time, but-

Amanda Beaty: Yeah, I’m sorry. We do have a hard stop.

Savita Kini: Anyway, so those are the key takeaways. Final words, there’s enormous opportunity, but high ambiguity and chance of failures. Because finally, AI models are statistical models. The role and the concern and the focus for AI PMs will be how we bring the productivity gains through AI and deliver a more personalized and creative experience for the enterprise consumer. And it’s an interesting and challenging, but very hugely impactful opportunity in the enterprise.

So I’m happy to take questions offline. Please feel free to connect with me offline. And thank you for listening.

Amanda Beaty: Thank you so much. And thanks everybody for joining us. We’ll see you in the next session.