The Enterprise CPO’s Guide for Winning at AI (with Randhir Vieira, Chief Product Officer at Omada Health)

This is the analogy I think of when comparing established enterprises to startups in the face of major disruptions like AI. So while the business equivalent of a cruise ship may have a lot of resources on board, it's gonna take a lot more time and nautical experience to change course than say the speedboat carrying four people in a case of energy drinks. My guest today is Randhir Vieira, Chief Product Officer at Omada Health.

Omada has been operating since 2011 and as a mature company with established products, they're playing with an entirely different set of challenges and advantages compared to the new startups that have AI embedded in their DNA. Randhir talked me through different approaches that enterprise leaders can take when implementing AI in both their product offerings and their internal culture, and how those decisions can help establish companies navigate into new horizons. Let's jump in.

Oh, by the way, we hold conversations like this every week, so if this sounds interesting to you, why not subscribe? Okay, now let's jump in. Welcome back to the Product Manager podcast. I'm here today with Randhir Vieira. He's the CPO of Omada Health. Randhir, thank you so much for coming and joining us today.

And so I followed my passion, which was photography, and joined a photography related startup called ifi. It was less than 20 employees at the time. One of the first or early internet of things, combination of hardware and software. And through that and through the VC network, met somebody at B2B learning management software called Mind Flash. And I was there for a few years leading product, marketing, sales, and engineering at various points there.

And then I have a huge passion for meditation and I had an opportunity to join Headspace to start their B2B. And eventually led product for all of Headspace. That was my first foray into health and wellness. And then from there to Omada.

So can you tell me a little bit about how you're thinking about these horizons and why you believe that companies tend to get stuck optimizing their existing workflows, rather than thinking about that horizon two, what's possible and what's next?

So summarizing is a great example, whether it's summarizing content, summarizing meetings, those kind of things are great places to start. And those are low hanging fruit, and of course we should absolutely take advantage of it. The bigger opportunities though, lie in the horizon two areas. So what are the things that we could just not do before that this new tool or capability can unlock? I think that's where what holds us back is the inertia.

It's the inertia of our thinking, of existing processes, of our business model, of lots of different things that usually hold especially large companies and successful companies back from moving to that horizon two, or revolutionary updates. And I think there are a lot of standard examples of, if we look at the last big tech shift that we had, which was around mobile and smartphones.

A lot of companies that may have had an online presence, let's say, for booking or other things from the web that moved that content or that capability to the phone because it was more convenient, et cetera. But the companies that really stood out were those that. Use the capability that in a way that just wasn't possible before. So the ride share companies like Uber and Lyft, et cetera, would likely not have existed without a smartphone.

So those are the kind of things that are more challenging for especially successful large companies to reimagine versus being almost victims of their own success with their existing successful business model and process.

So one is an example of currently we have a lot of people applying for jobs and we've always had some kind of algorithms that were helping to screen people, right? At many companies that received lots of applicants. And some of the new wave tools that I've noticed, some of them do full on synchronous, not just audio, but also video interviews of candidates.

Now, what if instead of having those keyword based, scraping, et cetera, anyone who applies for a role is instantly offered the opportunity to interview with an AI agent, would we now have a much better selection criteria that is more objective versus just what people state? In the application, you could actually run a structured interview process completely AI based, especially for that first tier or second tier interview.

So now you don't have the constraint that was forcing you to select a small group of people to put in front of a human interviewer. So I'm fascinated with these kind of companies that I just reimagining what's possible because of that constraint is removed of having to filter down to a smaller set of people.

Another example might be having a doctor on demand, or at least somebody that you can ask questions to all the time, and there are many companies that are trying to do this in a way that is both safe and complies with regulatory guidelines, et cetera. But again, that was not something that was possible earlier because of cost and resourcing, et cetera. Fascinating things there around mental health and a lot of areas.

They definitely come with some risk and one has to be careful about that, but it certainly opens up access, affordability, et cetera, in a way that wasn't possible five years ago.

And yet when you step back, you look at the technology adoption curve and it's standard, and I think no different in this case, where you're gonna have some already adopters and then the early majority and the late majority and the laggards. And it requires us as leaders to run our change management process.

Some of those things that may encourage this are experimentation, encouraging those early adopters to be the scouts and show the rest of the organization, or they show the rest of the team what's possible to invite the skeptics in and make them part of the team that is evaluating some of these solutions. None of these tools are perfect, but the pace of change and improvement that we are seeing is something that we have not seen before.

So for every criticism or concern that we have today, those quickly get reduced significantly, if not raised within a few months or a few quarters. So it is incumbent on all of us to keep trying and may not be deployed ready just yet, but at least to keep playing with it and see what the current capabilities are and be aware of both the pros and the cons.

And not look for solutions to jam this technology in, but really see what kind of problems or opportunities would benefit from this amazing technology. And it is a tool after all. It's not a solution that's looking for a problem. It should be a powerful tool that is deployed to the right problems or opportunities. There are also some companies that I've seen that also use the stick.

Which is as an example, they might say that if your team does not have a hundred percent adoption of the AI tools that we provide, we are not entertaining any request for additional headcount for your team. So that's certainly an interesting technique. I haven't seen many companies adopt that, but it's certainly one example of the stick. To really push that adoption because we're not seeing the level of adoption that maybe we want.

So it's interesting to hear that folks are going the other way.

Is the impact and risk, and it just allows you to list all these different ideas that come up in brainstorms on this matrix. And you can use ROI, et cetera for the impact piece. And then there are elements of risk that come up. So we certainly don't want to spend too much time in things that are low impact, regardless of the risk. So low impact, low risk, still not impactful. So we ignore that. Low impact, high risk, definitely we wanna stay away from that.

And so the easy place to start is things that are high impact and low risk. And typically those are things that are around summarization, internal tooling, those kind of things. And there are lots of projects that. Companies find where this kind of work can be super impactful and is lower risk. Great way to, to get the ball moving and get people playing with this kind of technology and see the real business impact of it internally.

But usually the horizon two kind of things are in the high impact, but maybe higher risk. There are ways to try and mitigate the risk. You do assumption testing, you can step your way into it, but at least being open to the fact that's really where the prize is and having things in place to move towards that. One of the questions that could be helpful in trying to explore what those horizon two are, high impact, what higher risk items might be, is what if we started our company today?

So we didn't have the legacy of the past. We start our company today with all the tools that are available, AI and non-AI tools that are available today. What would we do and how would we do it? What would the company look like? What would the offering look like? What would the pricing look like?

So it just gives people space to think about what could be, because you can be sure that there are some new companies who are starting with exactly that mindset and without even having to do it as a thought exercise, it is their reality that they do not have the baggage of the past.

What advantages do organizations have over AI first startups and what unique challenges do they face in contrast?

Similarly, the other side of that coin is because of those, they've been successful doing what they've done in the past and how they've done it, they can also be risk averse to changing the formula, messing with the machine. And there can also be inertia about Here's what we've done, here's how we've done it. These are the customers, here's what they like, et cetera. So the same reasons that they are successful and so many of the benefits that they have can also hold them back.

And of course, we explore ways to get past those barriers or friction points at least.

If you really lean into the carrot, you know, what are like the actual leadership tactics that you can implement?

Being able to find and celebrate some of these early wins, especially if they're impactful, can help generate some internal momentum and enthusiasm for the usual cautious skepticism for anything that is hyped. And we've seen this in the past before of, oh, we've seen this hype cycle before and there isn't anything real there, so let's just wait for the cycle to pass.

Instead, we are able to see that, sure, there is a lot of hype, but there may be something that is useful and impactful for us and let's see what we can get out of it. So being able to celebrate those early wins can be helpful. The third one is co-development, which I think is super, super important because rarely are we doing things just for ourselves. So we might be building things for internal teams.

We might be building things for members, but making sure that we are developing these solutions, especially 'cause they're so new in partnership with the customer, whether that customer is internal or external. And that is what drives not just the success of the feature being shipped, but more importantly the adoption and the impact of that feature and the value of that co-development cannot be understated.

So yes, it may make things feel like they're going a little bit slower during the development process. Usually the result and the adoption is way worth it.

Making sure that you're developing solutions that are within the bounds of the regulation, even as companies are maybe trying to shape regulation to accommodate this new technology, but making sure that the solutions that we're deploying are very much in line with the existing regulations. There's obviously in healthcare, the do no harm, and making sure that we really stick by that, and that can mean lots of different ways to mitigate some of the risks that may come with this.

So making sure that's. Outlined upfront and center, many companies will create before they start creating solutions, will create a set of AI principles. How do we wanna make sure that we are developing solutions? So examples might be to be very clear and transparent to customers and to members on when they're dealing with an agent and a virtual agent versus a human agent. Maybe one example, if there's anything that an AI agent is unsure of, that they will triage to a human person.

So there are lots of ways to mitigate that, but I think starting with those principles for your company is super important and helpful before going down specific areas. Being mindful of bias that may exist with some of the foundational models and ensuring that you're doing things to plug those gaps and make sure it's as unbiased as possible for your population can be helpful. And ultimately, no matter what we do, we have to test a lot.

And it's not just testing before deployment, but it's continuously monitoring and course correcting based on user feedback. Your own quality control process that is continuing post-production is important, so making sure that the guardrails that you set are working the way that you intend, and there's no model drift or other things that are happening, even post-production.

So I think all of those can help, but being aware and naming those challenges upfront and then trying to mitigate them, whether you use techniques like pre-mortems or other ones, can be helpful to name all the risks and concerns and how likely they are and what the mitigations are for, especially those that are high impact. And even if they're lower probability, they can still be important enough to know how we're gonna mitigate them.

Okay, if we take off our strictly CPO hat and we look into our crystal ball here, just yourself, obviously you've done a lot of thinking about what's possible, where we're at right now. What do you see happening in the next three to five years? What are you excited about in terms of AI developments? And from your perspective, your, in your professional experience, what do you think existing companies should be doing to prepare now to take advantage of that horizon two?

Just not the quality of the voice, but the inflections, the tonality, the size. It's just, it's amazing what's happening and we'll continue to see that evolution, but not just voice, but this kind of technology. So I think the ways for me to think about how we can take advantage of it is to continue to keep an eye on what problems are we solving, what opportunities are we trying to solve for?

And invest in people in training, in creating space for these prototypes and proof of concepts to be able to see what the current technology can unlock in terms of the problems or opportunities that we are interested in. I think those can be helpful. What constraints could this eliminate? So asking those questions when you see some new capability or new technology come up. Is it cost? Is it time? Is it resources?

And we talked about the AI interviews as just one example, asking those kind of questions that are helpful in that creative thinking, in the diverge phase of what might we do with this capability? What would we do if we didn't have this constraint? Can help unlock some of those opportunities so that no matter what the specific capability is, that may be available in three years or five years.

That we have people who are ready, available and excited about bringing that and plugging it into that solution.