Stephen Klein has spent the last two years publicly showing how the AI industry has its incentives backwards.

A Harvard MBA, Berkeley faculty member, and exited co-founder of LOYAL3 ($100M raised, acquired), he has written some of the most-read essays in the field on why major generic AI platforms are quietly erasing the livelihoods they were supposed to empower.

Georgetown University offered him a three book deal (the first-ever non-fiction trilogy) built around that argument. He has been invited onto more than 50 podcasts and 11 keynote stages to make the case in person.

Along the way, his message grew an audience of believers from zero to more than 75,000 LinkedIn followers in under a year without a single dollar of ad spend.

Curiouser AI is the company he built to prove the argument in the market rather than the op-ed page.

This is what he found.

Ready or not, half of working Americans will be on their own by 2030.

27.7 million Americans already work independently.

By the end of the decade, Upwork projects that number to climb to roughly 86.5 million through a boom of solopreneurs, consultants, gig workers, side hustlers, and micro-business owners.

That’s more than half of our current working population in the US.

The American economy is coming apart at the seams, and we’re witnessing the beginning of the greatest economic transformation since the industrial revolution.

A professional gig economy is being born under pressure, and the people inside it are being handed tools that erase them at the exact moment where being distinctive is the whole job.

When 86.5 million self-employed people prompt the same AI, they lose what made them discoverable: their identity.

Few to none are supported by an agency, a marketing team, or a brand department.

Their professional identity acts as their market presence.

In order to stay competitive and mobile in the market, adoption of generative AI is naturally among its highest rates among the self-employed. But the current suite of generative AI tools were built for enterprises executing known tasks faster, which makes the leading AI tools the wrong resource for solo professionals trying to articulate things that nobody has said before.

The more the self-employed economy leans on shared models, the more uniform every post, bio, and proposal becomes. LinkedIn, the premier social platform for 1B+ professionals, is already saturated with content that reads as if it came from one source, because it did.

Same tools.

Same AI platforms.

Same templates.

Same playbooks.

Zero differentiation.

We call it the Wonderland Problem.

In Lewis Carroll's well-known story, Alice in Wonderland, Alice falls into a world where the rules keep changing.

She drinks a potion and shrinks. She eats a cake and grows. She changes shape so many times to fit her surroundings that when the Caterpillar asks 'Who are you?', she cannot answer.

This is what is happening to the self-employed economy right now.

When a consultant, a founder, or an independent advisor adopts a generative AI tool because they need to compete, to be visible, to keep up:

The generic AI reshapes their phrasing...

Then their topics…

Then their point of view.

At first, it feels like it is helping. But after six months of prompting the same model that everyone else is prompting, they open their LinkedIn profile and cannot explain what makes them different from the next person.

The question ‘Who are you?' hangs in the air, and they do not have an answer.

This is the core business problem of the self-employed economy, and it is accelerating as more people reach for the same handful of models.

Discoverability on LinkedIn, on Substack, on any platform where attention is the currency of business, depends on being recognizably yourself.

Generic AI is called "slop" for a reason: it's a machine for sounding like the masses.

Our solution is a personal AI, trained on you, that nobody else can have.

Alice is your digital identity for thought leadership, and she is our answer to everything above.

Every installation is trained on one individual, their thinking, their advisors, their identity, and the platforms they publish to. The longer a customer uses Alice, the more she becomes them, and the more valuable she gets.

A personal AI built from the inside out cannot be copy-pasted to the next user, and it cannot be easily replicated by a bigger model with a bigger checkbook.

Under the hood, Alice runs on a philosophical system we call Reflective AI.

Rather than handing the user an answer, Alice asks questions grounded in their specific situation, surfaces the perspectives they are missing, and develops their capacity to think before a single word gets published.

Every piece of content is then scored across 245 dimensions of craft, trust, and brand alignment, with craft quality graded across 16 dimensions and trust architecture graded across 14 distinct patterns.

Alice explains the score, shows where the argument is strong, and flags the trust patterns that are missing.

Over time, users get measurably better at articulating what makes them stand out in the market, and Alice continuously improves sounding like them.

Lewis Carroll, author of Alice and Wonderland, said “Imagination is the only weapon in the war against reality.” That's what we're putting in people's hands.

Originality is the new moat and taste decides which businesses survive.

You might be wondering, what is the value of sounding like yourself?

And why do it with a tool like Alice?

We’re entering an age where AI is getting to the point where it can do anything.

When the process of creation is simplified and commodified, only the individuals with taste will build things that can survive in our new economic environment.

Reflective AI is our answer to that future.

Our product is already working, and paying customers are sticking around because they love it.

Alice has been in the market for just under a year. The results have surprised even us.

🌀 200+ customers already pay $25 per month. (With $0 client acquisition cost).

🌀 60% of trial users convert to paid against an industry average of 2 to 3%.

🌀 Churn is effectively zero.

🌀 Customers are paying continuously for just under a year.

🌀 Gross margin runs at 85%.

So far, our customer acquisition cost is effectively $0 because every paying user has come through word of mouth, Stephen's writing, or organic LinkedIn reach that grew from zero to 75,000 followers in under a year.

We are operating on a strong unit economics basis, which is rare for an early-stage AI company and deliberate on our part. We charge for the trial and we charge enough to cover our unit costs, which means we don’t have to monetize customer data to keep the lights on.

What customers are telling us.

What’s new: The release of our pattern intelligence engine, Looking Glass.

Looking Glass is our newest product release, and our first full-scale digital identity product.

It configures itself uniquely to every user who touches it, learning their identity, the frameworks they think in, and the platforms they publish to.

The scoring engine underneath is trained on 7 million LinkedIn impressions, a proprietary dataset that took a year to build and is not for sale, so a competitor would need to run the same experiment from scratch to approach what Alice already knows.

Technical advantages we built along the way:

🌀We balance Claude and Gemini under the hood to optimize quality while cutting inference costs significantly.

🌀We solved persistent memory by building a compression system that summarizes conversation history as the context window fills, working the way human memory does.

🌀We built a design-to-production pipeline that takes a product from LLM prototype to deployed SaaS in a faster than ever.

🌀As we raise, we plan to transition to an open-sourced foundation, give us total control over privacy and security, and unlock enterprise deployments on customer-owned servers.

What’s next: Our roadmap is stacked around rapid expansion.

Now. Looking Glass launched this month as a standalone digital identity product.

Next. A unified platform merging Alice and Looking Glass into a single hub with user profile-awareness, so users can brainstorm, position, and publish all in one place.

Q2 2026. Cascade Engine, a 10-framework decomposition system that turns one piece of source thinking into 5 to 30+ scored posts. We expect this to be the highest-value feature on the roadmap.

H2 2026. Alice Enterprise. Organization-level DNA trains Alice on the company, then configures a personal identity for every employee inside it. Includes content library, score analytics, multi-user deployment, and white-labeled servers for enterprises that need absolute privacy.

Our pricing structure and why that matters for investors.

Most generic AI companies are subsidized by venture capital and survive by harvesting user data. We took the opposite path on purpose.

Trials are paid, the subscription is $25 a month, there is no free tier, and customer data is not sold or used to train the underlying models. Because our unit economics work from day one, we don’t have to choose between our users and our runway.

The result is a business running at 85% gross margins with near-zero churn, where customers stay because the product makes them sharper and more valuable as a unique online presence.

When the broader AI market eventually has to reckon with unit economics…we’ll already be on the other side of that conversation.

The raise: $2M to accelerate a product the market has already validated.

Our north star is weekly returning paid users, because if someone keeps coming back, we have genuinely changed how they think.

Through this round, Curiouser will be targeting growth through a 24-month sprint aimed at $3M ARR.

Our product is by the people & for the people, so we’re opening our doors to devote this raise to our community.

*Results not guaranteed

A letter from Founder Stephen Klein: Why we’re raising on Wefunder instead of venture capital.

“At LOYAL3, I was part of a three-person co-founding team that raised $100 million from venture capital.

I saw what that path costs, the compromises, the pressure, the way the cap table eventually dictates the company.

I then spent six years running innovation and marketing for Dentons, the largest law firm on earth, advising lawyers and clients across fifteen countries on AI ethics. My workshops were accredited by the American Bar Association for Continuing Legal Education credit.

I watched generative AI get deployed at an institutional scale, and I watched who gained from it and who lost.

That experience is why I’m building Curiouser differently. The business model is the argument, AI that works for the people using it, not extracted from them, and the cap table has to match.

Most generic AI companies are subsidized by venture capital and survive by harvesting user data. We took the opposite path on purpose.”

Our promise for your future.

While everyone else is building AI to replace you…

We built Alice to help you become irreplaceable.

Trust will be the moat in the next decade of AI.

Come build with us ❤️

Podcast/Media:

1. Featured HubSpot Article https://startups.hubspot.com/magazine/issue02/stephen-klein-through-the-looking-glass
2. Featured in AI Journal https://aijourn.com/should-we-hire-ai-rather-than-implement-it/
3. Podcast “AI Won’t Take Your Job” https://m.youtube.com/watch?v=zC-DGVRVg8w
4. Podcast "Rethinking AI & Agentics with Curiouser.ai" https://www.youtube.com/watch?v=0cp8cnYwyBA
5. Podcast https://www.youtube.com/watch?v=SJQoym3gvv8