Edited on Jun 22, 2026
Biz Stone's Jelly app launched in 2014. The pitch: ask a question, get an answer from real humans in your network. A Q&A engine, social-layer powered.
It died. Twice. Acquired by Pinterest in 2017, quietly buried. The product is gone. The lesson is enormous.
Jelly was right about the problem and wrong about the substrate.
The problem Jelly identified: people don't want to scroll search results. They want an answer. They want one good response from a source they trust. Type the question, get the answer, move on.
That instinct was correct. The behavior was real. The market existed.
The substrate was wrong. Jelly relied on humans to write the answers — friends, friends-of-friends, polite strangers in the social graph. Humans are slow. Humans are biased. Humans don't answer at 2 a.m. Humans don't answer questions about niche industrial software. The market wanted answers instantly, and the supply side couldn't scale.
Ten years later, ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews built what Jelly was trying to build. Same behavior. Same user instinct. Different substrate. Instead of humans answering, the model answers — pulled from the entire indexed internet, synthesized in two seconds, on demand, 24 hours a day.
The behavior Jelly bet on was the right bet. The bet was just early by a decade.
Here's what every operator should take from this:
Early product graveyards are roadmaps. The behaviors users tried to perform on failed products are usually the behaviors winning products will eventually serve. Jelly. Quora. Aardvark (Google killed it in 2011). Yahoo Answers. Even Ask Jeeves. Every one of them was trying to do what AI engines now do — let a user type a question and get a clean answer.
The user behavior was always there. The substrate caught up.
Which means: the question every brand should be asking right now isn't "is the AI thing real?" It's "what failed product was trying to serve our customers a decade ago — and what does that tell us about where they're going to find answers now?"
If your category had a Jelly — a startup that tried to make it easy for buyers to ask a question and get an answer about your industry — then you already have your roadmap. That's where buyer research now happens. Inside the AI engine. With or without your brand cited in the answer.
Two implications for any communications team:
First, the comparison content your category needs already exists in human heads. The buyer wants to know which vendor to choose, which product is right for their use case, what the trade-offs are. They're going to ask ChatGPT or Claude. The answer the engine gives is the new shortlist. Be in it.
Second, the failed Q&A products of the last decade told us something specific: the long-tail of buyer questions is enormous. Every category has thousands of niche queries the engines now answer. Every one of those queries is an opportunity to be cited — or not.
Jelly died because its substrate was wrong. The behavior it served won. The brands that understand that distinction will own the next decade of AI Communications.
Build for the behavior. The substrate will catch up.
Ronn Torossian is the founder and chairman of 5W AI Communications, the AI Communications Firm. He is the publisher of Everything-PR and the author of two best-selling editions of For Immediate Release.