OpenAI shutting down Sora as a standalone product says something important about consumer AI:
Most people do not want creative superpowers. They want less work.
That is the mistake a lot of founders are still making.
They confuse technical possibility with product demand.
Just because a model can generate cinematic video does not mean millions of people have a recurring need for “cinematic video generation.”
That is not a market.
That is a demo.
And this is exactly why Jobs To Be Done (JTBD) is still the ultimate truth behind consumer products.
Consumers do not buy capability. They buy progress.
The consumer internet does not reward what is impressive.
It rewards what becomes a habit.
That habit only happens when a product solves a real recurring job.
Nobody wakes up wanting “AI video.”
They wake up wanting to:
- make an ad faster
- send something funny to a friend
- make content without thinking too hard
- explain an idea visually
- kill boredom instantly
Those are jobs.
The AI is irrelevant unless it makes one of those jobs meaningfully easier, faster, or better.
This is where a lot of AI products die.
They are built around what the model can do, not what the user is actually trying to get done.
That creates a lot of first-week excitement and very little long-term retention.
In consumer, novelty is easy.
Habit is everything.
JTBD is not some PM framework. It is just how reality works.
Users are not “using” your product.
They are hiring it.
They hire it to make progress in their life.
That progress might be:
- functional
- emotional
- social
Instagram was not hired to “share photos.”
It was hired to project taste and identity.
TikTok was not hired to “watch short videos.”
It was hired to destroy boredom with machine-like precision.
ChatGPT is not hired because people “want AI.”
It is hired because it removes friction from dozens of annoying tasks.
That is the actual wedge in consumer AI: not intelligence, but effort compression.
The winning consumer AI products will not say: “Look what this model can do.”
They will say: “Here. I already did the hard part.”
That is what people pay for.
People don’t know what they want. They do know what they hate.
Henry Ford’s line is cliché, but still directionally right:
“If I had asked people what they wanted, they would have said faster horses.”
Consumers are terrible at describing the solution. They are excellent at revealing the friction.
They will not ask for:“an AI-native multimodal creative environment.”
They will say:
- “I need this by tonight.”
- “I don’t know how to make this look good.”
- “Can this just do it for me?”
That is the opportunity.
The best founders do not take user requests literally.
They decode them.
Because people may not know the product they want.
But they are constantly revealing the job they want done.
That is where almost all great consumer products come from.
The big lie in AI consumer right now: everyone is a creator
No, they are not.
This is where a lot of the market is delusional.
Founders keep betting that if AI makes creation easier, millions of people will suddenly become active creators.
That is mostly fantasy.
Most people do not want to create.
They want:
- to consume
- to scroll
- to react
- to lightly remix
- to get a good enough output with almost no effort
That was true before AI.
It is probably still true after AI.
AI lowers the cost of creation.
It does not magically create creator desire.
That distinction matters a lot.
Because if your startup thesis is:
“Now everyone can become a creator”
…you are probably building on a very shaky foundation.
Why companies like Wabi may not work
This is why I’m skeptical of products like Wabi (started by Replika founder, raised $20mn in seed round, received a whole lot of love from YC and a16z)
They may look exciting.
They may attract creators.
They may even get strong early engagement.
That still does not mean they become enduring consumer businesses.
Because AI may not actually change the underlying structure of the market.
It may just flood it with more supply.
And more supply does not automatically create more demand.
What probably happens instead is:
- a few creators become much more leveraged
- a lot of users become casual remixers
- everyone else keeps consuming
That is not a creator revolution.
That is just a more efficient content machine.
And that is a much worse business than many founders think.
Why?
Because creation tools are not enough.
You need either:
1. Utility
Something that helps people complete a job they already care about
2. Distribution
A place where people actually come to consume
The second is incredibly hard.
Because AI lowers creation cost.
It does not lower distribution cost.
And distribution is still the bottleneck. Always has been.
This is the real opportunity in consumer AI
Not more blank canvases.
Not more “make anything” apps.
Not more infinite creative possibility.
The real opportunity is in compressing work.
The winners will be products that eliminate:
- hesitation
- complexity
- setup
- uncertainty
- taste anxiety
- execution overhead
The future consumer AI winners will not feel like toys.
They will feel like relief.
That is a much better business.
Final take
Sora’s shutdown is not really about Sora.
It is about a lesson the market keeps teaching over and over:
Consumers do not buy magic. They buy momentum.
They do not want AI.
They want progress.
They do not want infinite creativity.
They want less friction.
And they definitely do not want another tool that asks them to do more work.
That is why JTBD is still the most important lens in consumer products.
It forces you to stop asking:
“What can the technology do?”
And start asking:
“What painful, recurring job can I make disappear?”
That is where the real companies get built.
Everything else is mostly a demo.
What’s your take?
