mini intro
===

​[00:00:00] 

Welcome to Launch Pod, the show from Log Rocket, where we sit down with top product and digital [00:00:15] leaders. Today we're doing something a little different and bringing you a mini episode with Cory Lieb. Got VP of Product and Optimizely we're going to be covering her framework for using AI in both customer facing solutions and Optimizely's own product [00:00:30] workflows. In this episode, Cory shares how her team uses Copilot and Claude for everything from PRD writing and meeting recaps to translating technical language into accessible insights. She also talks about developing Opti GPT, an [00:00:45] internal AI assistant that surfaces customer data, product knowledge, . HR policies in seconds. And finally, how she built a customer facing AI query tool that also became a surprise hit internally for measuring customer health and [00:01:00] engagement.

So here's Corey ligo in this AI mini episode.

Jeff: but There's one thing you touch on that I wanna make sure we don't overlook before we go. that is how the team at Optimizely is using AI in their own workflows.

As much as people wanna know, you know, my board comes to [00:01:15] me and says, Hey, sprinkle a little, you know, AI on that thing. The big thing that almost everyone really wants to talk about is. How are you getting value in data? How are you being more productive from using some of these tools on your own end?

What does that look like on your team at Optimizely and maybe [00:01:30] across product and even across the org.

Cory: Optimizing as an AI first company. True and true. So again, not just in the solutions we're building for our customers, but in how we are using AI in our day-to-day, which I really like that. And in fact, I got my team [00:01:45] together the other day because as a product org we're chatting constantly and sharing new AI tools be it Bolt or Claude and how are you using it and what did you think of the, you know, the prototyping capabilities of this.

And I tried this one for vibe coding and here's what I thought. And so. We [00:02:00] are in the thick of, we are trying all kinds of different tools. We have some that we have enterprise licenses for we use copilot a lot in our day-to-day, for example, through Microsoft, and I'll talk about some of our use cases there.

Our engineers work with Claude and [00:02:15] we're exploring other solutions. So it's very much evolving and growing in terms of the AI tools. That we're incorporating in our day-to-day work. But I got my team together the other day because I was like, Hey we're chatting a lot about this. Let's hop on a live call, carve out a little time, and each [00:02:30] person come with an AI tool that you've been playing around with, and demo it, share it, show it so we can all get more ideas from each other and sort of help to hone in on which ones might be the best that we wanna get.

An enterprise grade license for and use consistently across the team. So we [00:02:45] did that and there were so many great examples. I think some of the common things that, that we're using AI tools for and more obvious use cases are like PRD writing and refinement. To enterprise licenses through apps like copilot, have [00:03:00] integrations to Jira, for example.

So help with writing your PRD and then get it, get the acceptance criteria and get it translated into stories that you can share with your engineering team, like making that very seamless. What was a very time consuming task before um, also, you [00:03:15] know, less technical PMs in working with engineers or other technical folks and trying to translate.

Whether it be snippets of code or just technical language that is not as natural to them using AI tools to say, [00:03:30] can you break this down for me in layman's terms, or Can you make this into something visual that I can better understand? So team, the team's been doing that. We use the heck out of copilot for recapping meeting notes.

Action items. Summarizing things. If we're out [00:03:45] for a day or a week, the work doesn't stop. But copilot can summarize, here's what you missed in your chats. Here's what you missed in your emails and help you with what you need to focus on. I also loved one of the PMs on my team. She's like, I love the microphone feature on copilot and the [00:04:00] AI tools 'cause I'm a talker versus a writer, so I wanna just talk and sort of brain dump.

What I'm thinking about, be it debriefing from a meeting or to start to articulate around A PRD, I just wanna like voice it and then copilot, [00:04:15] cleans it up into this beautiful document that looks like I spent hours on it. So those use cases I really liked. We also would, Optimizely have developed an internal Opti GPT tool.

Jeff: this one was really cool when I heard about this a little bit earlier, so.

Cory: I, I [00:04:30] absolutely love it and we continue to enhance it. But this is everything from, if I want a query to say, tell me about this customer. Which products do they have with us? How much a RR are they spending with us? What have their recent support tickets been about? Like if [00:04:45] I get an escalation or an issue or I'm just checking in with a customer in A QBR, what a quick way to get a sense of and sort of get an executive brief of what's going on with this customer.

Similarly. To, if I'm a new employee or I work in one product area and I wanna know about another [00:05:00] product area, I can query about our products, get information about that. We have an hr capability, so if I wanna know about policies or, you know, time off or US holidays, I can query Opti GPT for that.

So that's been really cool. And then we also use external [00:05:15] tools where we're not uploading sensitive information, but we can do. Market research for what's publicly available to, you know, when we're considering pricing or things like that. So we'll use chat, GPT perplexity, for example, to do market research, and then we're tinkering with tools [00:05:30] like cursor, lovable rep to do vibe, coding, prototyping and so on.

So those are some of the things that we're doing. But again, every day we're actively sharing hey, we came across this, or I tried that one. And just sharing those ideas across the team as we all learn.

Jeff: I think that's one of the biggest [00:05:45] things is how do you kind of enable people to try things. How do you set kind of Mission parameters, not to use the rocket metaphor

. But how do you set the mission parameters there of here's, you know, the big picture, what we're trying to do? But go forth and try things. And it's okay if, you know, sometimes it's not gonna work, [00:06:00] but when you get something that works, share it. You know, we try and post a lot about these kinda wins of process.

How are you using tools to enhance and how could others potentially use it? One thing we found on that, I love the kind of opted GPT to ask questions [00:06:15] internally. one really interesting thing about AI, I feel like has been emergent capabilities. You build something to do X, but because you did it right to do that, it has the ability to do, you know, Y or Z or Omega or some other fantastical thing.

We launched into our product, not [00:06:30] for us, mind you, but for our customers. We launched what we call Ask Lea, which is the ability to kind of query. In a chat function not like build charts, but those same kind of questions, you know, how is this customer doing?

Or how are people using this new feature? But one of our CS team actually realized she could ask [00:06:45] it, how is this customer using the product lately? And, the insights it gave. She basically came back and said, this is better than a lot of CS tools we have for account health right now because it's going in en able to actually talk about engagement.

And this has been a huge [00:07:00] kind of new internal tool that we'd even build for, but now is used across our CS team and some of our SDR teams that focus on the inbound around. Understanding customer and user health. So what we built for external became an internal tool that has super wide adoption now.

Because that but because someone [00:07:15] took the idea and shared it, I feel like is the biggest thing and talked about that.

But it all still comes from the human element of what are the key things that are really interesting and having those human conversations up front. 

You can enhance em, but you can't replace those. 

Cory: Absolutely. And I'm a big advocate for share information, share [00:07:30] learnings, just like we've been doing today, even when it's fumbles and you don't get it right, but what did you learn and how did you pivot? To me, that's where I learn the most from others as well. And so I do regularly have these sharing sessions with with the PMs on my team.

And then we also [00:07:45] set OKRs around. Making sure we're incorporating AI into our day-to-day work in this way and so on. So yeah, very top of mind for me.

Jeff: Yeah. And then I think just the last thing I wanna share 'cause what you said there sparked my mind real quick is, you know, it's sharing and kind of trying [00:08:00] and everything like that. One of the biggest ways I've figured out to learn is actually try to do something and get it like halfway there and kind of fumble my way through it.

Just try until I can't do it, and then show an engineer and be like, Hey, look at what I did. It's pretty cool, huh? And like, oh my God. And you can see their, you know, [00:08:15] smoke come outta their ears and, but then they take it and kinda work on it a little bit. And like two days later I have a great finished product.

​ [00:08:30]