AI transformation: from vision to execution

Welcome to this fireside chat. We're gonna be focused on agentic adoption, both upward and downward. I'm joined by Giovanni Faucette, who is the chief of staff to the CPO and VP of product and tech ops at Skydance. Yes. We, sitting in your role, AI transformation really goes one of two ways. C suite vision and really tactical day to day.

Can you describe your role for someone who has not sat in it, and what does that mean, sitting at the the front of AI transformation?

Sure. So I have a interestingly double barrel role. So I... My chief of staff side is pretty much like Dash, the AI bot for Asana. And then I also am the vice president of technical program management where I lead a pretty awesome team of special operations folks, business operation, and program and project management people.

So I technically bridge the gap between vision and strategy all the way down to execution, so I have a pretty unique connective tissue between both sides.

Yeah, awesome. And I think not just execution, but really translating it for for all your teams, which also is just so hard as teams are siloed in different locations, different time zones, working in different tools.

Before you started changing things at Skydance, what did fragmentation actually look like? And, yeah, talk to us about that.

We have several places at which information is flowing. So there's Slack, you have Asana itself, Docs, Google Drive, Word documents, PowerPoints, slide decks. So there wasn't a particularly centralized space at which information was going back and forth or in a streamlined way going up or down.

So everything was kind of across different silos and different teams, and each team had their different method of doing things. Status reporting all the way down to team meetings and ways at which they let their leadership or their peers know what's going on. So very much fragmented all the way from top to bottom.

Yeah, and what if, like, for kind of giving that information up to the C suite, how dependable was the reporting that you were delivering to them? A little rough.

You know, we have in the silos people doing duplicative work, so you have different groups doing the same things but not knowing. Information is stale on different cadences too, so you have a Friday readout versus a Monday readout, and status reporting on Wednesday, you know, you're missing time in those, in those statuses. So, you couldn't really trust where the information was coming from or the time at which you were getting it if it was the most current information available to you.

Yeah, and what's interesting too is like I think with AI supporting to make those reporting more efficient, if they're not accurate, it's gonna cause C Suite to want more reporting, more oversight, be throwing more meetings on the calendar, which really defeats Yeah. The Efficiency. So how did you identify C Suite reporting as one of your main major use cases, and what did you build?

So it was our first practical use case out of two organizations coming together that have never worked together in their entirety at once. We had a lot of changes, lots of teams coming together, and new leadership that needed visibility.

So our chief product officer has two instances of reporting every week, so one of the things that I noticed was, hey, how do we get this guy status on a Monday and on a Thursday? Two of which, one is for him and one is going upwards further.

Right now, or right now it's fixed, but before it was very much fragmented across the board. So we started with a very small team that was very ingrained in Asana. So end to end work management, status reporting, all the information was captured there. And we were able to identify, you know, hey, can we start taking some of this work and some of these statuses and pulling them into a pretty quick three bullet executive report.

So through the portfolio, we identified what was our chief products officer's priorities for the week. Sometimes those are monthly or they change.

So we created a custom field that had priority status p one through p three. P ones are things that he cares about. So we could tag those in a portfolio, and then each week or each every two weeks that program manager statuses on those, and we were able to spit out that big status report that way. And given the format at which he requires status to be, it was more of a can we get the outputs ready to just be copied and pasted into a spreadsheet at the time or a document, so that's kinda where we landed end to end. Yeah. So not all the way there, I wish we would be able to put it into a dashboard, but that's target state.

Yeah. And you mentioned just kinda like competing priorities, or just how priorities shift.

Obviously, kind of like cascading those down, but under like c suite priorities, they change week over So how do you build something that sustained priorities that were changing week over week or month I over needed something that wouldn't break the foundational framework.

So for us, was that custom field where we could toggle on and off, like, hey, this is priority one this week, but maybe in two weeks it's now dropped down to a priority two.

So it doesn't change the framework at which you get the data out. But it's a click of a button for us when we're just like, okay, like this week he really is caring about this, let's put this on p one, and then it'll spit out the next week for us and and his status reporting.

Yeah. Awesome. So we talked about upward. I wanna talk about, downward, managing adoption. And what I found really interesting is you didn't mandate the team to use AI, But, you know, c level reporting is one challenge, but getting users to actually change their daily habits and their day to day is just a really different kind of hard.

And you advocate for an accessibility and lived in experience, and really understanding the tools that your teams were using. Talk to me through that. Like, how did you get your team to use AI without requiring them to to use it? Sure.

So I'm very much a try it.

I don't wanna say never. You can't really break anything. You know, take a task that you're doing that's super repetitive or something super small that you're just like, I don't know, maybe this could be agentic, maybe it something that can come off my plate, save me a couple minutes every week, every day, and try it, see what happens. Take, you know, your Friday status reporting and turn it into, let me click this button, have it generate, and then see what it spits out and then let me edit it if it's not something I'm looking for.

And with that, you're fine tuning yourself but also your expectations on what you're expecting to get out of AI and what it's generating for you. So I'm very much a try it, see how it fits within your routine. You don't need to go out and say like, hey, I'm gonna try to jam this process in to make it an agentic process. There's...

That's not really the way to do it. And I found much more success in adoption and having people breathe and live in it in their day to day than, you know, saying, hey, you have to use this today, tomorrow, or the next week.

Yeah, and it almost makes it easier though, also having AI in the tools that they're already using, rather than introducing new tools to get them to adopt it. But I love that too because I think you can really share in the pain and the challenges your team is facing and you're in it with them, but also celebrating the wins and recognizing best practices and sharing those as well.

And you recently tried an AI teammate yourself. Walk us through what that looked like and how that kinda changed how you Yeah. Spoke about it with your team.

So I tried one of the out of the box ones, the PM, or project manager, program manager one. I wanted to try it myself, so I'm living and breathing by my own philosophy here. So I was like, I'm, you know, unfamiliar, pretty much uncomfortable, you know, with this. I don't really know what what I'm doing or what it can do, so instead of reading a bunch of articles and watching some videos, I was just like, you know, let me let me add a teammate to a project just to see what it could do.

So I, you know, use a project that is ongoing for me, I started up the the PM teammate and added it to it, it just instantly had context. And in seconds, it's creating tasks, sub tasks, it's, adding comments to these tasks, and then I can go in and be that human intervention to either, you know, stop it in its tracks or redirect it in commentary and things like that. So it's quite incredible, like in a click of a button I have something that had context immediately in a project. The one learning that I had is your projects have to be a little bit more refined, So your tasks, your your descriptions, your comments in there, you need to be a bit more detailed.

It's not gonna know enough context if you don't give it to it. And that's the one thing I say about all AI is like, it only knows how much you tell it.

So the quality of your projects and the quality of information you're putting in there is super important.

But it was quite quick how up to speed it got, or onboarded it became to my project, whereas like, me as a person, it's gonna take me, you know, a day or a week to learn about a program or a small project to just read all the context, meet all the people and do all that and by using a teammate, it's just adding it there and it knows who's assigned, it knows your tasks, it understands context and how you set your projects up and things like that.

Yeah, awesome. I'm gonna go off the cusp a little bit because I think what I'm hearing too is I think sometimes with new technology too, if the technology's not giving you exactly what you want, it's quick to say, this isn't working or I'm not getting value out of this, but with AI you really have to train it and you have to give lots of feedback.

So what would you, like, kind of for... And I don't know if you have any people on your team that are kind of, like, resistant or kind of giving that push back, but are there kind of, like, tips or tools or best practices you found for people that have been more kind of resistant to adoption?

Yeah. It's one of those things where, yeah, if it's not giving you the results you're looking for immediately, it's just like, oh, why am I using this, I could do it myself.

Starting small, so those small examples really kind of break that chain there, so if I'm like, okay, I have this really repetitive task where I have to approve a task in Asana every week, and it sends me a summary, all those things, I can have a teammate do that now instead of me reviewing all the bits and pieces if it's pretty standardized.

So taking a very small task to you and building that into your routine is best advice I have.

Often times, I don't get the results I'm even looking for, so I've been met with frustration where I'm just like, I thought this was gonna be much faster, much easier, and it wasn't. And it's just like, okay, is it something that I need to automate?

Sometimes no.

Part of what makes things agentic is routine process, things that are foundational baseline, it can't take away your decision making, you still need to be very much involved and see what's going on for those decisions.

But yeah, starting small in those routine things that you're just like, you know what, today I'm gonna see if this is more helpful than it was yesterday with me doing it.

Yeah, it's really interesting because I think a lot of people who, in AI leaders, are really trying to tackle the most challenging problems, like the hardest problems, those take time rather than kind of finding the quick wins, the efficiencies. And I think you mentioned that too within the c the c suite reporting of the data's already clean, you have it there, how do you, you know, Exactly. Operate Yes. Yeah.

So as humans and agent teams evolve in how you change, how your organization operates, what are some use cases and strategic objectives that you are guiding to in this new hybrid world?

So first is just to get people more comfortable with using agentic pieces or AI as companions, so I always call them like your your productivity companion.

Everything is always human led. I stand by that. You know, you have to be able to, you know, intervene, have that, you know, human choice in decision making, but really, we're starting to get into more of a hybrid situation, so it's keeping an eye, you know, keeping things visible across the board, having your team work with their agentic partner and share those those learnings across the board too has been super helpful as we kind of move forward in this hybrid space.

But it's always human first and like your agentic is, your agentic companion is your partner, so it's more of a productivity improvement than anything in my opinion, but kinda pushing that forward and seeing where we can get through learnings across teams to then scale across the board, because it is very new to to folks, so we're just trying to get off the ground a little bit.

Yeah. Awesome. I'm going to ask some quick hitters, short questions.

So what's one change management role that enterprise AI officially breaks?

Top down mandates. I think, you know, kind of getting in there and trying is the best form of adoption you can get across the board.

Yeah, we said quick, so I don't Yeah, no, no.

I won't go lower. Yeah, go go for it. Like, What does the human agent workforce look like at Skydance in two years?

Always human led, so I stand by that.

With your productivity companion by your side in many many situations, so it could be status reporting, workflow improvement, process improvement, those types of things. Awesome.

And then what's one thing for the leaders in this room that they should go and do on Monday morning?

Try something. Try something you haven't.

Pick a small task, always small, to start. See what you could do. Again, can't break anything, so just try it out, see what works. That's the best way to get into it, and then expand from there. Do medium problems, large problems, and then you'll see how powerful, you know, Asana AI teammates they can be.

Yeah. Awesome. That is it, and I want to thank you for joining us here. Sure. Yeah.

So lovely to to speak with you and learn more.

Cool. Thank you. Yeah. Thank you for having me. Thank you.