Neoncoral Is Building an AI Social Network and Being Very Quiet About It

The Macro: Consumer Social Is Due for a Weird One

Consumer social is the category that venture capitalists claim to have given up on and then keep funding anyway. For good reason. The returns when it works are absurd. The odds of it working are terrible. And the definition of “working” keeps shifting.

The last real wave of new social products came from the short-form video era that TikTok kicked off. Before that, it was ephemeral messaging with Snapchat. Before that, the photo-sharing wave that Instagram rode. Each generation of consumer social has been built on a new interaction primitive. Something about how people create and share changes, and a new platform emerges around that change.

The current thesis floating around is that AI will be the next interaction primitive. Not AI as a feature bolted onto an existing social network (that is what every incumbent is doing), but AI as the fundamental thing you interact with. Your feed is generated. Your conversations include synthetic participants. Your content is co-created with models. The social graph itself might be partially artificial.

This is either the future of how people connect online or it is deeply dystopian. Probably both. But the market is enormous, incumbents are struggling to figure out what AI-native social actually means, and there is room for a startup to define the category before the big platforms figure it out.

BeReal tried to redefine social around authenticity. Locket tried to make it about home screen widgets. Dispo tried to bring back delayed gratification in photos. None of those stuck at massive scale, but they proved that consumer appetite for new social experiences is still there. The question is whether AI changes the interaction model enough to create a durable new platform.

The Micro: Stealth Mode With a Loud Investor List

Neoncoral is a Spring 2025 Y Combinator company building what they describe as “AI consumer social.” That is almost the entirety of what is publicly known. The website at neoncoral.inc is a single stealth page with a contact email and a list of angel investors.

The founder is Thomas Liao. His background is the most interesting thing about the company right now, because it is unusually strong for a consumer social play. Liao was the ML lead at Scale AI, which means he spent time thinking about data quality, model training, and the infrastructure that makes AI applications work at scale. He also worked on evaluations and post-training at Anthropic, which means he has hands-on experience with the frontier models that any AI-native social product would need to build on.

That combination is rare. Most consumer social founders come from product or design backgrounds. Liao comes from the AI infrastructure side, which suggests that whatever neoncoral is building, the AI is not a feature. It is the foundation.

The angel investor list on the landing page is deliberately prominent: people from Cruise, Figma, Tesla, OpenAI, Anthropic, Pinterest, Perplexity, Scale, and Addepar. That is a heavy lineup for a company that has not publicly shown a product. It tells you that people who understand both AI and consumer products looked at whatever Liao pitched and wrote checks.

Beyond that, there is almost nothing to evaluate. No screenshots. No beta waitlist. No blog posts. No tweets explaining the vision. The stealth email address is literally [email protected]. Liao is being intentional about keeping this under wraps, which in consumer social can be smart. You only get one launch. First impressions matter more in consumer than almost any other category.

The company is currently a team of one, which is not unusual for a consumer social play in stealth. You do not need a big team to build a prototype. You need a big team to scale one.

The Verdict

I am going to be honest: this is a bet on the founder, not the product, because there is no product to evaluate. And as founder bets go, this one is interesting. Liao has a technical background that would let him build AI-native social mechanics that less technical founders would have to outsource or fake. He has relationships at the companies making the models that would power such a product. And he chose to go consumer when his background would easily support a B2B AI infrastructure play, which suggests genuine conviction rather than category opportunism.

The stealth approach is a double-edged sword. On one hand, consumer social launches benefit from surprise and controlled rollout. On the other hand, the longer you stay silent, the more you risk building something in isolation that does not match what people actually want.

What I am watching for: the reveal. When neoncoral finally shows what it is building, the product will either feel obvious in retrospect (the way Instagram felt obvious once you saw it) or it will be so weird that it takes time to understand. History says the weird ones are the ones that win. The ones that feel obvious on day one tend to be features, not platforms.

If Liao has built something that makes AI feel like a natural part of how you interact with other people online, not a gimmick and not a crutch, this could be one of the more important consumer launches of the year. If not, it joins the long list of well-funded consumer social products that never found their audience. I genuinely do not know which it will be, and that is what makes it worth watching.