loading...
SocialAI fills your feed with AI followers, each with a distinct personality – turning social media into a private sandbox for ideas.
ENTRY ANGLES
AI personas for sales training simulations (different buyer personality types) · Multiple AI personas in educational platforms for debate/discussion with varied intellectual perspectives · AI shopping assistants with multiple personas to handle product trade-offs
VERTICALS
CAPABILITIES
AI persona generation and differentiation, Multi-persona conversation/interaction systems, Domain expertise modeling (sales techniques, academic subjects, product knowledge)
SocialAI has been making the rounds on social media for the past few days. The startup behind it, Friendly Apps, raised $3M two years ago – but the app itself is worth a look regardless.
Surface-level, SocialAI looks like Twitter: a feed of short posts and comments. The twist is that you're the only real person in it. Every other account is an AI character commenting on your posts.
You can populate your feed with as many AI followers as you want, each with a distinct personality: your superfan or your harshest critic, a troll, a skeptic, an optimist, a realist, a pessimist, a contrarian, a visionary, a philosopher, a fatalist – whatever you choose.
Your AI followers comment on everything you post, then respond to your replies, turning each post into an unfolding conversation or debate that can go as deep as you want.
When you're done with a thread, publish a new post on a new topic and the cycle starts again.
The premise: you can say anything here. Share thoughts you wouldn't share publicly, test ideas that aren't ready for outside eyes, vent about things you don't want to burden anyone with. The feed is entirely private – no public timeline, no other users to stumble onto your content.
The creator of SocialAI frames this as an early experiment – he wants to observe how people actually use it before drawing conclusions. And the use cases are genuinely varied:
- A private journal that talks back.
- A space to develop ideas before you're ready to share them with anyone.
- A personal devil's advocate for testing half-formed thinking.
- A supportive presence for moments you don't want to unload on people in your life.
- Or all of the above at once.
The deeper point – one worth returning to – is that AI chat doesn't have to be neutral and generic. Most people think of AI assistants the way they think of search engines: objective, impersonal, there to retrieve information. But that framing misses something.
People prefer reading individual writers over news sites partly because individual writers have personalities, biases, and points of view. Readers love or hate them for exactly those qualities. The same logic applies to AI personas.
Several startups are already running experiments in this direction.
Chirper ([covered previously](/review/revoljucija-podkralas-nezametno)) raised $750K to build a Twitter-like platform where only AI characters post and comment. Users create their own AI personas – giving them names, backstories, jobs – and release them into a shared feed where characters from different users interact with each other.
Butterfiles (iOS, Android) takes a similar approach, with a key difference: real users can join the conversation by commenting on AI-generated posts. That startup raised $5M, including $4.8M in June.
Dopple ([related review](/review/individualnost-cepljaet-i-prinosit)) raised $1.88M in its first round for a marketplace of AI characters modeled on historical figures, fictional characters from games, films, and TV shows, or entirely invented personas. Users open private chats with any character. A quick test: a question about startup strategy put to a digital Sun Tzu produced a remarkably coherent answer.
DreamRP ([covered here](/review/zaprety-ischeznut-i-ono-poletit)) is currently in Y Combinator with a platform they describe as Character.ai meets Patreon – a marketplace where creators build AI characters and users pay subscriptions to chat with them. The angle: attract serious creators who can build characters genuinely worth talking to, rather than low-effort fan clones.
These experiments are interesting in their own right, but the more compelling question – for anyone with a cynical streak – is where this leads when applied to something that creates real value or real revenue.
Some early answers are visible.
Hyperbound ([covered here](/review/za-takoe-obuchenie-kompanii-tochno-zaplatjat)) came out of Y Combinator and raised an additional $1.5M to build a sales training simulator. The product: AI personas playing different types of buyers – "aggressive and dismissive," "polite but immovable," "inquisitive but skeptical." Sales reps practice not just the generic pitch, but the adaptations required for every personality type they'll encounter. That's genuinely hard to train with traditional methods, and companies will pay for a tool that does it well.
Packback ([covered here](/review/jeto-mozhno-sdelat-ne-tupo-a-horosho)) raised $12.2M for an educational platform where students discuss and debate the topics they're studying – developing writing, argumentation, and critical thinking alongside a human instructor and an AI assistant. Right now the AI feedback is a single, averaged voice. The more interesting version: multiple AI personas with different intellectual orientations pushing back from different angles. The learning experience becomes dramatically richer.
E-commerce is another obvious candidate. AI shopping assistants are proliferating, but most surface a single recommendation. For anything with real trade-offs – headphones are a good example – you'd learn more from a debate between an audiophile AI and a value-maximizer AI than from a single balanced take.
What other domains would benefit from AI perspectives in dialogue with each other rather than a single voice? That's both a product design question and a business opportunity. Sales training and education are solid starting points – but the category is wider than either.