The case for AI Social Networks
Why the next social breakthrough might come from an AI company.
Most social apps have turned into content networks that are anything but social.
I believe large AI companies are uniquely positioned to reverse this trend and create the next mainstream social app.
New tech unlocks new social features
Cheap text broadcasting unlocked Twitter.
Photo tagging and exclusivity unlocked Facebook.
Smartphones with cameras and processing power unlocked Instagram.
Disappearing photos unlocked Snap.
Battery-efficient location sharing unlocked Zenly.
AI unlocked…
Following the timeline above, the trillion-dollar question becomes:
What previously impossible social mechanism does AI enable?
I believe the answer is Contextually Proactive Creativity.
Contextually Proactive Creativity: The ability to create something that a user wants to share with a friend, at the exact moment they would want to share it.
Contextually
Right now, collecting dust on my phone, there are thousands of photos, hundreds of songs, and dozens of thoughts that will never get shared simply because I have not been given the right context to share them.
Large AI companies already have most of the data on the internet, but they still don’t have the context that matters for a social app.
They don’t know what I thought about last week, what my grandmother cooked for dinner, what I just saw or what sport my friends practice.
The Notes App, WhatsApp, the Photos App, and Strava know those things. However, none of them have the reasoning capabilities to process this context in a useful way.
This represents a massive opportunity for AI companies to build smarter competitor apps, fully own the data, and share it within an ecosystem.
An AI-first version of these apps would understand what I am curious about, my relationship with my grandmother, what catches my eye, and my friends’ personalities.
Building apps, or an entire OS, will become more important than ever to build the richest social profile on the planet and unlock new sharing avenues with this context.
Simple example: Last year, I took a photo while hiking. I don’t know that my friend Jean likes hiking, so I never shared it with him. But OpenAI knows that Jean recently became interested in hiking, because he asked ChatGPT about tips. Unless Social AI tells me to share this photo with Jean, it will never be shared.
Content isn’t scarce. Context is.
Proactive
AI’s barrier of entry - prompting - is still too high for most people.
Prompting should not be a prerequisite for getting value from an AI.
By having context and being proactive, Social AI can turn its users into creators by default. This would be a drastic change from the social dynamic of any other social app, where 90 %+ of users don’t create anything.
Don’t wait for me to discover that Jean likes hiking.
Prepare the photo for me to share.
Don’t expect me to remember a blacked-out night with Ernesto.
Tell me to show him the photo at dinner tonight.
Don’t wait for me to do an album from my trip to Shanghai.
Do it and offer to send it to my family’s group chat.
Don’t wait for me to call my grandparents.
Tell me what I still haven’t updated them on.
Don’t expect me to know Marty’s music taste.
Tell me when we listen to the same thing.
None of the above requires new content.
They require the right context and proactivity.
Every social app tries to come up with clever “ice breakers” to kickstart engagement loops. Most of them are terrible because sharing rates are bottlenecked by the value of what is being shared.
Asking users to share something of little value, like a QR code to their profile, results in low sharing rates, even if it only takes a single click.
On the other hand, when you show something unique and of value to users, they will jump through all kinds of obstacles to share it.
The recent Ghibli trend is the perfect example. Users (1) went on an app, (2) uploaded a photo, (3) wrote a prompt, (4) repeated this a few times, (5) downloaded the photo, and (6) shared it with friends.
Social apps can only dream of this much user motivation. But AI apps have it, because they provide tangible value. Add proactivity, and you have guaranteed virality.
Creativity
Just because we can generate everything, it doesn’t mean we should.
I don’t want to see thousands of Studio Ghibli images on a feed, but I want to see my family smile.
Observe what people are already sharing:
Ghibli images of themselves and their friends.
Their own IQ tests.
Asking ChatGPT what secret it knows about them.
Their work /vibe coded websites.
Figurines/starter kits in the plastic box of themselves.
Turning their pets into humans.
Making music about their friends.
Comic books of their trips(https://x.com/Midaskwant/status/1904740339702718858)
A few months ago, a friend and colleague of mine created an entire music album where the lyrics are exclusively my name, “José”. Our entire team sang it for weeks in the office and laughed for hours. This is now a core memory of mine. One that would not exist without AI.
By focusing AI’s creative powers on helping users share little nuggets of life with loved ones, AI players have a chance to create a social app that brings people closer together, and is also extremely viral.
Contextually Proactive Creativity can make you a better friend in a way no other social app is doing right now.
This idea is not new.
Early Facebook made you a better friend by reminding you of your friends’ birthdays. What they got wrong was letting people publicly and in 1 click share a general “Happy birthday” post. As a result, Facebook commoditised its own value, and nobody cares about Facebook birthday messages anymore.
Imagine the birthday scenario in Social AI:
AI reminds you that it is Lucas’s birthday.
It puts together a collection of photos of the two of you,
It reminds you of funny messages you exchanged throughout the years.
It suggests a gift for him.
It does not allow you to share this in 1 click.
You copy the photos, write a birthday message with personal context, and send it to him.
Social AI created a beautiful moment between friends, without commoditising it.
For Lucas to receive it, you still had to care about him.
This same sequence can be replicated daily for all kinds of social interactions.
Successful social networks have never reinvented the wheel. They simply allowed people to do what they already did, better. There are a lot of interactions that can be done better with Contextually Proactive Creativity.
How to build up to Contextually Proactive Creativity
It’s far easier to write an essay about this than to pull it off.
Here are some of the prerequisites:
An ingestable personal data layer.
Different apps in the same ecosystem; An OS; Access to 3rd parties; Know exactly who my friends and I are.
Reasoning capabilities to interpret this nuanced context extremely well.
An exceptional app experience
Great onboarding; Tight sharing loops; Dense networks; Top distribution.
Social app users are unforgiving. They will not prompt, connect wallets, wait for image generation, or frequently perform any other complex action.
This requires a team dedicated to social experiments that can iterate fast.
Judging by OpenAI’s team size and composition, I already expect them to be building the app or OS layer. Point 1 will soon be solved.
They are also already building towards AGI, which indirectly solves point 2.
Point 3 will seal the deal. Make the new giant.
Should any of the 3 points fail, contextually proactive creativity cannot work, which is why I think large AI players are uniquely positioned to capture the opportunity.
A closing note
I don’t want an AI to write messages to my wife instead of me. But I want the AI to remind me that this time last year, she was feeling down, so taking her on a date today can make a difference. AI should not replace my social interactions. It should enhance them.
A product that leverages cutting-edge technology to make humans care more for each other is worth building.
Extras:
Other possible answers for the question “What previously impossible social mechanism does AI enable?”
Instant engagement
Users no longer have to wait for others to be online to get dopamine hits. These can be generated and controlled on the fly.
Unlimited Single-player utility
At its core, AI is a tool. Killer social apps have been built by making users “come for the tool and stay for the network”. ChatGPT, with its rumored 500M users, is in a strong position to do this. The strongest in the world right now.
Memory
Proactively follow trends
If OpenAI knows that 70% of users between 22 and 27 in Paris have generated Ghibli-style images this week, why not proactively generate one for the other 30%? This applies to all trends.
Entirely AI-generated “For you” feeds
Detect it’s the first time I’m in a new place → Prepare a small blurb with photos and fun facts
Expand on the topics I recently prompted you about
“What if my feed was filled with answers to questions I didn’t know I had yet?”
Follow me on Twitter: @jose_goncalves_


this might be the 'killer use case' of AI-hardware (the 'friend' of this world) instead of 'companionship'