Investor, CEO & founder Grace Gong on why AI is a multiplayer game, how to tell real AI from hype, and who owns the handoff when it breaks.
Grace Gong’s calendar is a live read on what the market is actually talking about. She interviews founders, executives, and investors weekly, and hosts the dinners and summit stages where they compare notes. What she keeps hearing from that seat is that AI didn’t collapse work down to one person and one model — it multiplied the number of parties who have to stay in sync. Your AI, your teammate’s AI, your team’s AI, and the humans still accountable for the outcome. That’s the coordination tax, and it’s why every guest on this show answers questions while playing Snake. The snake grows with every pellet until it runs out of room. So does unmanaged coordination.
She’s the founder and CEO of Smart Venture Media, host of the Smart Venture Podcast and Venture with Grace, and the organizer behind the Smart AI Summit. She’s an angel investor and venture scout backing AI startups, a LinkedIn Top Voice with 59,000-plus followers, and the author of “How to be a VC,” a No. 1 Amazon bestseller in the VC category. She joins Dan O’Connell, CEO of Front, on Untangled Ops to talk about why she’s bullish on the multiplayer AI era instead of the one-person billion-dollar company, how she tells real AI infrastructure from a wrapper, and who she thinks is accountable when a handoff breaks 18 months from now.
Key takeaways
AI turned work into a multiplayer game. As more companies deploy AI agents, the coordination is about to become even more complicated between AI agents, and AI agents and humans.
Shared context is the thing worth wiring up. Grace points to Slack’s MCP: the value is pulling context out, not adding another tool.
Being 10% better isn’t a moat for AI success. The companies that hold up are reinventing how people work with AI and fulfilling meaningful use cases.
When the handoff breaks, a human still owns it. AI won’t take accountability alone. Humans will remain in the loop.
Trust is the part AI can’t deliver. People still prefer humans when trust matters most. Judgment, empathy, and context are why the hardest moments stay human.
Full transcript
Transcript has been edited for brevity and clarity.
Dan O’Connell (host): Hey everyone, and welcome to Untangled Ops, the show where operators, founders, CEOs, and investors talk through the work between the work — the clunky handoffs and lost context, the pileup between people, and now between people and AI. That’s the coordination tax: the overhead nearly every business pays and almost nobody tracks. I’m Dan O’Connell, the CEO of Front, and your host for this episode.
My guest — I’m thrilled to have you here. I feel so honored to have Grace Gong. Grace is the founder and CEO of Smart Venture Media, host of the Smart Venture Podcast and Venture with Grace, and the brains behind the Smart AI Summit. We were fortunate to share that stage. Thank you for having me. I hope to be back next year.
She’s also an angel investor and venture scout backing AI startups. Her portfolio includes companies like Hyperbolic and Espa Labs. Welcome, Grace. Great to see you.
Grace Gong (guest): Thank you so much, Dan. I’m so thrilled to be here. First of all, your office is amazing, and this is such an amazing spot, and you’re so friendly, and I love your marketing team — and we’re about to play some games.
Dan: So let’s get to it. Before we close out with a rapid fire round, where you’ll give Snake your best shot, I’ll say this: I’m not only intrigued by the answers you’re going to give, I’m also very curious to see how you play Snake. That’s the game we make every guest play, because it’s the 8-bit version of the unmanaged coordination tax.
My first question for you. Our coordination tax research found something that should bother every AI optimist. AI didn’t lower the coordination tax. It inherited it. Then it scaled it. You talk to founders, execs, and operators every week. Is that what you’re seeing too?
Grace: Yeah, for sure. I feel like AI has been making a huge shift from a single-player game to a multiplayer game. I was very fortunate to have dinner with Slack’s CMO last week. Slack just launched their own MCP, and essentially what people are doing is pulling different data out of their Slack workspace. That really helps a lot with having different context for people.
But essentially, we’re entering the major movement of people doing work collaboratively with both AI — the AI agent itself — as well as the human side of things. So there’s going to be so much communication between your AI, your colleagues’ AI, your team’s AI, and the people. I definitely see this being a huge trend of people trying to hold it together during this time.
Dan: At one point there seemed to be a trend of the one-person company, having your whole business run by AI. You seem to be betting a little bit on the opposite. You actually think enterprise AI is swinging back to multiplayer. How come?
Grace: Yes. I was just on a podcast, literally maybe twenty minutes ago, with a partner from Upfront, and we were both agreeing on this. He was talking about the founder running the one-person company selling Ozempic — and now that founder is actually feeling somewhat lonely. As a billion-dollar company founder, you feel lonely.
I mean, I can understand that, but the thing is, people were missing the part of actually interacting with other humans. I personally feel like we live in a society where community, as well as human interaction, is still going to be a huge thing. So I’m personally very bullish about multiplayer AI.
Dan: Yeah, I love that. I love hearing that. I agree wholeheartedly. It’s hard to build things alone. People won’t have these shared journeys together. I think that’s part of the process.
Grace: Exactly.
Dan: And so, as much as I believe there will probably be the billion-dollar single-person startup, I very much know that’s a very lonely journey. I’m curious to see if it plays out that way. But again, I believe in the multiplayer universe that we’re in, and I think that continues to play out.
You said being 10% better isn’t a moat right now. When you evaluate an AI company for your own portfolio, what’s the tell that there’s real infrastructure versus a wrapper that breaks the first time it meets a real org chart or scale?
Grace: The best companies in any of the categories are literally the people who are reinventing how people think, as well as how people function with AI. I was talking with a company on Monday, and essentially what they’re doing is an AI-native version of Figma or Canva. Nowadays, people are betting on AI agents interacting with other AI agents, so a lot of the AI-native companies are designing for the agents. I think that’s a really interesting concept, and I see it playing out in different categories — whether it’s design, work productivity, CRM, among other spaces.
On the other hand, you really have to find the real-life use cases to make the AI really work. You really need to find the industry that has the problem you can solve with AI and thinking about things fundamentally. That’s how I think about it.
Dan: Eighteen months from now, who’s accountable when the handoff breaks? Is it still a human question, or has AI taken over and solved all the problems?
Grace: Maybe a little bit of both. Like AI, but with human in the loop.
Dan: Final answer?
Grace: Final answer. What do you think?
Dan: I agree wholeheartedly. I very much believe that automation eats the world over time, on the longest time scale. And I think we very much see that people seek human experiences, and that as humans we deliver judgment and empathy that I think is impossible to replicate. I think people will probably debate me on that. But eighteen months from now, we’ll continue to see AI deliver the ability to do more complex things. I don’t think it replaces the humans at all. And definitely not as fast as I think some of the market probably expects.
I think we sit in one of these really, really interesting, fast-moving markets around CX and customer support and customer service, which obviously has a lot of pressure around the economics and the beliefs of transformation. But what we are seeing and hearing from customers is very much that they want the best of both worlds. They want to drive efficiency. They want to drive productivity. They want to get rid of the easy stuff. And they want to deliver these exceptional experiences. And I think those exceptional experiences are really delivered by people — especially in the hardest moments, the ones that are frustrating, or when something is broken. That probably requires, again, going back to a stronger sense of judgment and empathy.
Grace: I would add, too — I really like what you said about judgment and empathy, because essentially what AI can never deliver is the trust, and also the interaction between humans. Why do people still go into a bank, when everyone knows an online transaction is fine? People still want the human touch point, where you can feel you can count on someone, you can explain something with better context, and then a human decision maker can really see through the different cases instead of giving the one-size-fits-all answer. So I really feel like humans will still be in the loop.
Dan: Yeah. I’ve been fortunate to buy a home, and when I went through to do wire transfers, I definitely wanted the person to be there — even though, as a technologist, I wholeheartedly believe I could probably do this online. I definitely wanted the trust factor. So I think that continues to play out in the world. But the very real opportunity is for automation to continue to help us do better things and take on more tasks.
Well, that is a wrap on the major questions. Now the fun part. We’re going to transition into a game of Snake. While you’re doing that, I’m going to give you a bunch of rapid fire questions. The only thing I need you to do is focus on the game and don’t overthink your answers. Rapid fire. You ready?
Grace: I’m ready.
Dan: Great. Let’s go.
All right, Grace, here we go. Snake game. The score to beat is 1,555. In the meantime: one AI product you can’t live without at the moment?
Grace: Claude. Claude Cowork.
Dan: A founder or company that’s being slept on right now?
Grace: Actually, I think there’s a company called Xspot, which I’m an investor in too. They make an amazing automated email product.
Dan: Honest opinion on AI and the job market. Is it cutting or creating more roles?
Grace: Creating more roles. I think it’s creating more fun roles, because you don’t have to do the hard work of going into Excel and really building a financial model.
Dan: What’s your contrarian belief compared to other investors when you decide to invest in a company?
Grace: I think you don’t have to know about the product at the very beginning.
Dan: What’s one thing people should stop using AI for right now?
Grace: Writing.
Dan: What stage in a startup is most interesting to invest in, and why?
Grace: Seed. Because I think there are so many changes. You’re just betting on the people, and everything is unpredictable.
Dan: What’s the first site you open on the internet?
Grace: The first site I open on the internet, in the morning? Actually, LinkedIn. I love LinkedIn. I just love content creation as well. I love the business news, and I love to follow Dan and other thought leaders in the space to see what they’re up to.
Dan: Most overhyped AI capability right now?
Grace: Can I say coding? I don’t know, maybe that’s underhyped or overhyped. I think it’s overhyped in Silicon Valley. I mean, AI is amazing at coding, but yeah.
Dan: The most dangerous place to put an AI agent in business today?
Grace: Uh, maybe agentic payments. It’s really scary.
Dan: What’s one task AI should have figured out a long time ago but still hasn’t?
Grace: Folding laundry.
Dan: That’s a good one.
Dan: All right, that’s a wrap on this round of Untangled Ops. Grace, I think you set a new high low score, so kudos. Thanks for being here and giving us an honest take on everything to do with AI and the market. We’ll pull up the leaderboard and see where you stack ranked with the other guests.
For more conversations on wrangling unwieldy customer ops, subscribe to Front’s YouTube channel @FrontHQ. Thanks again for watching.