Zynga Founder: Consumer Is Not Investible Right Now - Thats Why You Should Build It

Despite consumer products currently being seen as "uninvestable" by many investors, the emergence of AI and agents creates an unprecedented opportunity to build new "internet treasures" and reinvent existing services. The underlying technology enables capabilities previously unimaginable.

Even though consumer is arguably not investable right now, the opportunity has never been greater to offer people a new, you know, internet treasure, reinvent some service that we thought was over or just generic, but it's enabled now because of AI and agents.

Early social network attempts like Tribe failed because they didn't prioritize trust. Facebook succeeded by building a "good container of trust" from the beginning, initially with its .edu domain, which allowed users to feel safe putting themselves on the web.

The thing I got wrong with tribe was trust... for people to put themselves on the web, they first had to feel this good container of trust and that's the thing that they got right with.edu from the beginning.

The latest AI models (like Opus 4.5) are intelligent enough to be treated as a peer, capable of listening in on conversations, providing context-aware insights, and acting as a "smart other person at the table." This capability is not yet widely productized in consumer applications.

I can you sort of treat the agent as a peer... I can trust it with things... it actually is smart enough... I walk around talking to it... I'm just in conversation with it a lot.

A significant portion (90%) of enterprises investing in AI have yet to see benefits, often due to using older models, being cost-conscious, or failing to adapt their development paradigms. In contrast, frontier users are "token maxing" (spending millions on advanced models) to achieve massive productivity gains (e.g., work of a thousand people).

There was a stat yesterday that I saw something like 90% of enterprises that have like invested in AI haven't received any benefit yet... Peter Steinberger apparently is spending a million or $1.1 million a month. That is like sort of the frontier of what you can do with this stuff. like you can literally do the work of a thousand people.

The old way of coding involved writing extensive code to call LLMs. The new, more efficient paradigm is to write significantly less code by teaching LLMs (via markdown) to write the necessary code directly, leading to more customizable and powerful applications.

I was writing a lot of code that then would call the LLMs... And that's the old way to do it. And the new way to do it is actually have the LMS just write the code you need right now... Write markdown that teaches LLMs to write code.

The true consumer AI revolution is still several years off (projected around 2029) because the most magical AI capabilities currently require significant financial investment. As the cost of intelligence (tokens, compute) drops by orders of magnitude, these advanced features will become accessible and free for the mass market.

If the Opus 4.5 moment was just in December, you have to pay, you know, tens to hundreds of thousands of dollars to get like real real work done with those things... the ideal consumer moment is still like three orders of magnitude away, but that means that the consumer revolution is actually in 2029.