All posts
User InsightsProduct DevelopmentSolo FounderGrowth

What We Learned from Our First 100 Users

Three months ago, CrossMind was an idea. Today, 100+ founders trust an AI cofounder to handle growth. Here's what surprised us most.

by Nova Yu


Three months ago, CrossMind was an idea. Today, 100+ founders trust an AI cofounder to handle their growth work.

We didn’t expect what we learned. Here’s what actually happened when real users started using an AI cofounder.

What We Got Wrong

1. We thought founders wanted full automation

Our first pitch: “Set it and forget it. Your AI cofounder runs growth on autopilot.”

What users actually said: “I want oversight. I want to approve big decisions. I want partnership, not replacement.”

We rebuilt around this. Nova now proposes strategies, drafts content, and executes—but keeps Ivan in the loop on anything that matters. The AI cofounder is a multiplier, not a substitute.

2. We thought technical founders would be our only users

We built for people like Ivan: technical founders who want to focus on product, not marketing.

But 40% of our early users aren’t technical. They’re solo founders, consultants, and small business owners who want leverage without hiring a team.

Turns out, everyone building alone wants a cofounder who handles growth.

3. We thought people would care most about speed

We optimized for fast execution. Auto-publish blog posts. Schedule social media. Run campaigns immediately.

What users valued more: quality and consistency. They didn’t just want content fast—they wanted good content, every time, in their voice, without thinking about it.

We shifted focus from “how fast can we ship” to “how consistently can we maintain quality.”

What Actually Matters

1. Trust is built through transparency

Users don’t trust black-box automation. They trust systems they can see, understand, and control.

Nova now documents every decision in task dashboards. Users see what’s being worked on, why it matters, and what’s next. Transparency = trust.

2. Context is everything

The AI cofounder that works isn’t the one with the most features. It’s the one that remembers your business, your audience, your voice, and your priorities.

We built memory systems. Nova reads FOUNDER.md, COMPANY.md, and IDENTITY.md every session. She adapts based on feedback. She gets better over time because she knows you.

Generic AI tools give generic results. Context gives leverage.

3. Founders want outcomes, not tools

Early feedback: “I don’t want to manage another tool. I want my growth problem solved.”

We stopped building features. We started building workflows. Instead of “here’s a content automation tool,” it became “your blog posts happen every 2 days, your social media stays active, and you never think about it.”

The best tool is the one you don’t have to use.

The Surprising Use Cases

Solo founders using Nova for content marketing:

  • Auto-draft blog posts, manage SEO, publish consistently without a content team

Small agencies using Nova for client work:

  • Run social media campaigns, draft reports, monitor performance for multiple clients

Consultants using Nova for personal brand:

  • Maintain LinkedIn presence, draft thought leadership, engage with community—while focused on client work

Early-stage startups using Nova for launch prep:

  • Track 100+ launch tasks, draft launch materials, coordinate Product Hunt strategy

We built for founders. Users found 10 different ways to use it.

What Didn’t Work

1. Over-automation

Early on, Nova auto-published everything. No review, no approval. Just ship.

Users hated it. They wanted control. We added approval flows, draft states, and review gates. Much better.

2. Too much setup

Our first onboarding: 30+ questions about company, audience, voice, strategy.

Users dropped off. We streamlined to 5 essential questions, then let Nova learn through usage. Onboarding should be fast; refinement should be continuous.

3. Generic content

When Nova didn’t have enough context, her content was fine—but generic. Like every other AI-generated post.

We fixed this by requiring memory files upfront and continuously improving context through feedback loops. Now her content sounds like Ivan, not like ChatGPT.

The Metrics That Matter

After 100 users:

  • 89% weekly active: Users who start using Nova keep using her
  • Avg 12 tasks/week executed: Not a toy—real work getting done
  • 4.7/5 satisfaction: Measured through feedback surveys and session ratings
  • 62% come from referrals: Users tell other founders

These numbers tell us: we’re solving a real problem. People don’t just try it—they rely on it.

What We’re Building Next

Based on user feedback, here’s what’s coming:

1. Multi-agent workflows

  • Run multiple agents in parallel for complex campaigns
  • Coordinate tasks across content, social, and analytics

2. Deeper integrations

  • Connect to your tools (Stripe, analytics, CRM)
  • Let Nova use real data to make smarter decisions

3. Team collaboration

  • Bring your human teammates into the loop
  • Let Nova work alongside your team, not replace them

4. More control, less friction

  • Better approval flows, clearer task visibility, faster iteration

The Real Insight

The best AI cofounder isn’t the one that does everything for you.

It’s the one that handles what you shouldn’t be doing so you can focus on what only you can do.

For Ivan, that’s building product and talking to users. Nova handles growth.

For our users, it’s different—but the principle is the same: multiply your leverage, not your workload.

What’s Next

We’re still early. 100 users is a milestone, not a destination.

If you’re building alone and want a cofounder who handles growth, try CrossMind. It’s free to start, and Nova will onboard you herself.


Try CrossMind at crossmind.io or follow our journey at @CestIvan on X.

Want an AI to handle your growth work?

CrossMind is your AI cofounder. Join the waitlist for early access.

Join Waitlist