Jonas Kamber

✦ A new chapter ✦ Today, I officially start building Mono. I will build this business in public. I will post every (working) day. I will share my progress, ideas, struggles, financials, everything. And I would love to have you with me on this exciting journey. ⛵️

Today is a good day. Because I just released a big update. It makes the core interface on Mono (mono.chat) so much better. → The selected AI is now always visible → The Multi-Model-Research option is no longer hidden behind a menu → There is even a subtle color change whenever you’re using Claude, GPT, Gemini, Mistral, or Perplexity This is only one of several UX improvements I’m launching over the next couple of days. If you see something I should improve, let me know. Hope you like the new chat box design.

Poor performance on LinkedIn, exploding revenue 💥 This was week 18 of #buildinpublic. Let’s get into it. → I've spent 10+ hours crafting LinkedIn posts that provide real value. None of them resonated. I'm staying on it. → Referrals work incredibly well. I focused on different initiatives to fuel that fire. → In that process, I introduced a bug that sent emails multiple times to about 100 people. I'm sorry if you were affected. Fortunately, the complaint rate stayed at 0%. → I researched whether I could offer a European-only model option, so your data never leaves Europe. More on that soon. → Mono added another $7,716 ARR this week. The fastest-growing week yet. Next week, I'm improving what's already working before adding anything new.

Woohoo, that was fast! 🎉 Mono just hit $30,000 ARR, updating the milestone to $40,000. Word of mouth is the main driver. Customers recommending Mono to others. And it's snowballing. From 1,000 to 10,000 in 65 days From 10,000 to 20,000 in 15 days From 20,000 to 30,000 in 10 days Still building alone. Still 100% focused on building a better ChatGPT. It’s an incredible journey.

Day 88 of #buildinpublic While Mono approaches $30,000 in ARR, I'm in #UX design mode. A new chat box design puts all options in plain sight. No more hidden menus. A subtle color change that makes it clear which model is active. And more updates are coming. Better memory handling control. Smaller details that have been bugging me for weeks. This version will be the best version.

New milestone reached. 🙌 Yesterday, the 500th human signed up on Mono. All through my posts on LinkedIn, building in public, and happy users telling their network. Still $0 spent on marketing. And I'm getting a lot of positive feedback on the new Multi-Model-Research feature. This will likely drive the next 500 sign-ups. 🙌

Biggest feature. Fastest growth. This is week 17 of building Mono in public. Let's get into it. → LinkedIn impressions are still flat. I tried different angles and formats. Nothing clicked yet. I'll keep experimenting. → User growth was strong. Different customers posted about their experiences with Mono and motivated their network to make the switch. That was one of the biggest growth drivers. Mono almost hit 500 users. → The option to exclude OpenAI models was a big success. It brought in new users, and more people than I expected had opted out of OpenAI models. → Multi-model research is live. Users voted for this feature, and the interest was real, but the launch email and posts got much less engagement than I hoped. → Revenue is growing faster and faster. +$6,516 ARR added in 7 days (+40%), putting Mono to $22,620 ARR. I need to remind myself not to take this growth for granted. The first two channels are working like a charm. I hope the third one I start building next week will do the same.

Isn’t growth supposed to slow down 5 weeks after launch? It seems to accelerate. Just 15 days to go from $10'000 to $20'000 ARR. → Still building alone. → Still $0 spent on paid ads. → Still 100% product focus. What makes me especially happy: → People start posting about Mono → Everyone keeps their subscription (0% churn) Next milestone: $30'000 ARR

People on X like to say LinkedIn is cringe. It sometimes is. But you can also create momentum there. When Ioana Teleanu posted about canceling her ChatGPT subscription, 384 people liked it. I had a feature idea, jumped in, pitched it, and it resonated. Don't ignore LinkedIn, it's the main driver for Mono's growth ($18'828 ARR, 5 weeks after launch)

I just analyzed Mono's last 100 sign-ups (minimum 10 days ago). 22 people converted to paid subscribers. Average time to upgrade also improved from 5.4 days to only 3.3 days. For context: → Most SaaS get 1-2% conversion → Top products like ChatGPT reach ~5% → Mono is at 22% Never seen anything like it.

New growth record this week. 🙌 But LinkedIn impressions took a dip. I posted less general #AI content (which tends to reach more people) and more about building Mono in public. Finding the right balance between reach and conversion is something I'm working on. The upgrade rate hit 14% today. Out of 100 people trying Mono, 14 buy a subscription. This is incredibly high. Mono is getting a major update next week. A handful of early users already have access to Multi-Model Research. Mono lets Gemini, Claude, and GPT research a topic, cross-check through web searches, and combine findings into a summary paper. The goal: more thoughtful, well-balanced, and hallucination-free answers for complex questions. I plan to release this next week, depending on the feedback of early testers. Growth keeps accelerating. Mono added $4,092 this week, hitting $16k ARR for the first time.

The rare moments you realize the power of an idea in product development. I just had such a moment. Let me explain. Mono lets you use all major LLMs side by side. You prompt, Mono selects the best model. Not happy with an answer? Select an alternative. But what if they worked together as a team? We know diverse teams are more successful. So I started building a Multi-Model Research feature for Mono. It’s like ChatGPT's Deep Research, but with checks and balances. Gemini Pro, GPT 5.2, and Claude Opus each conduct research, orchestrated by Mistral. So I asked it how much screen time is ok for my daughter. The results made me pause. The combined answer is more balanced and thorough than any individual response. But I wanted to see the differences. So I added a section highlighting each model's weaknesses and blind spots. This section is more powerful than I expected. It reveals what Claude, Gemini, and GPT each missed. Showing clearly how much more powerful this multi-model approach is. Should Grok be included as well?

Do we live in a bubble? I definitely do. AI is everywhere. On my feed, in our UX Meet-up talks, and in every new startup idea. I built my whole life around it, building Mono. But let’s zoom out for a second. On this visual, heavily inspired by @damianplayer recent post, each dot represents 3.2 million people. Only 0.3% of people actually pay for an AI subscription. 84% have never touched an AI chat interface (knowingly). We're not in the middle of the #AI era. We're at the very beginning of it. The opportunity in front of us is still almost entirely untouched. Let's build.

I've built multiple businesses over the last 10+ years. Nothing really hit. There was the design blog. The iPhone socks. The graphic design course. The revenue tracking tool. The casual mobile game. The photo-box rental business. The building material marketplace. The user testing tool. Every business found some fans. Most are still running, but nothing went through the roof. A year ago, I started Meme Studio. It showed the least success. It made it easy for me to pivot. But I had to decide what's next. I'm carrying around an ever-growing list of 100+ business ideas for years, but I had to decide which one to work on. So, I challenged myself: validate 10 business ideas in 10 weeks. Week 1: CV improvement tool Week 2: Domain brainstorm tool Week 3: Job ad improvement tool Week 4: Self-updating help center Week 5: Chat with major AI models in one tool I had to stop the challenge in week 5. I knew this was the one. So I started building. I could reuse a lot of the code from the previous project and got to market fast. 4 weeks ago, I launched Mono. It's still very early, but the graph shows that pivoting was the right call. Most of us tinker with a business idea. If you have one, just start building. It's never been easier to create a realistic prototype. It might take a few attempts to find something that works. And it's probably not the first idea. But I wouldn't have found the right thing if I hadn't tried so many times. Just get started.

#QuitGPT claims that over 1.2 million people have joined the boycott of ChatGPT. The movement centers on OpenAI president Greg Brockman's $25M donation to Trump's MAGA Inc., ICE's use of GPT models for screening, and concerns about AI psychosis and emotional dependence. It mostly found traction on Instagram, but I also saw posts on LinkedIn. I'll be honest: I couldn't fully verify every claim on the QuitGPT site (link below). Some figures seem solid (the Brockman donation appears in public filings, the video of Scott Galloway is real), while others (like the AI psychosis estimates) need more context. The underlying sentiment is real, though. ChatGPT is no longer a love brand like it was for many people. I often hear "I'd rather support you (Mono) than give this money to OpenAI". But it's one thing to be angry. It's another to actually switch tools.

Only +34% revenue growth this week. Only, haha. It's insane that I even had that thought. My weeks on Mono feel like months. Things move so fast. This week started with Mono hitting the leaderboard of fastest-growing startups on unicor .ne (by TrustMMR, where I share all revenue data). For a brief moment, Mono even reached number one. Then came the punch in the face: a 15-minute outage, because a service provider had a major system failure. Not fun. But every setback is a setup for something better. I moved Mono away from that provider and connected directly to OpenAI, Gemini, and Claude. Result: Mono is now more stable and even faster than before. I was still able to ship new features: → Link validation: Mono now shows when a reference link is unreachable, clearly exposing hallucinations. → Smart model recommendations: If a model can't do something, like create images, it suggests another model that can. → Search through chats: something I wanted to add for weeks now. Mono grew to $11,760 in annual recurring revenue. Just 4 weeks after launch. And all that building Mono alone. What a roller coaster. What a week. What a journey.

I added 3 new testimonials to the Mono website. Thank you, Boris, Arinda, and Klinton. Testimonials matter because they reduce uncertainty for people who are curious but not sure yet. Nielsen (Global Trust in Advertising) found that 83% of people trust recommendations from friends and family, and 66% trust online consumer opinions. The Medill Spiegel Research Center also reported that showing reviews can increase conversions by 190%. Since I hashtag#buildinpublic, here is how I got these testimonials: I kept an eye out for comments here on LinkedIn. If one stood out, I simply asked if I could use it as a testimonial on the website. Pretty straightforward.

Mono keeps #3 spot among fastest-growing startups on Unicorne today. But shipping doesn't stop. Latest update: AI models now understand their own limits. Ask Claude Opus 4.6 to generate an image? It used to claim it did, but nothing appeared. Confusing if you're not deep into #AI. Most people don't know Opus can't create images. And the models themselves don't know either. Now, Mono "tells" Claude, GPT, Gemini, and others what they can and can't do. They'll even suggest better models for specific tasks. Imagine #ChatGPT telling you to use Nano Banana for image generation. OpenAI would never do that, Mono does. Using multiple models in one tool has never been this easy.

Wow 😳 Mono is officially one of the fastest-growing startups right now. It's bouncing between second and third place. Still can't quite believe it.