Riding the AI Storm: My Startup’s Wild 2-Year Journey Through Innovation and Chaos! (I/IV)
MindOS: A Year of Turbulence in the Eye of the AI Storm
In late 2021, armed with the above immature and geeky demo video, we approached investors to talk about AGI, AI agents, LUI, and the future of human-machine interaction. We pitched an AI-native operating system. The investors, with their typical curiosity, asked, ‘What’s AGI? How long until it’s a reality?’
The Most Common Questions from Investors:
- What exactly is AGI? When will it materialize?
- How can AGI solve any specific problems better than current AI?
- Big companies haven’t cracked it, so what makes us think we can?
- If the big players succeed, how can we compete?
- Why name a virtual character platform an OS?
- What’s our moat? Our unique strategy?
- What’s our business model? How much money are we aiming to make this year?
We optimistically thought AGI might take 20 years, but we slyly suggested to investors it could be around 5 years. A bit of entrepreneurial optimism, you know?
But reality had a different plan. Less than a year into our journey, ChatGPT burst onto the scene, making AGI a household concept overnight. It was a game-changer for the world and for us.
Before ChatGPT, we thought AI was then alike Level 1 of autonomous driving. The key question was how to monetize immature technology and sustain cash flow until maturity. We chose B2B, focusing on marketing scenarios because they had the funds and the appetite for innovation.
After ChatGPT’s debut, our biggest challenge became how to rapidly build barriers in a fiercely competitive landscape. We made two major pivots: 1. Shifting from B2B to B2C to speed up decision-making and scale-building, and 2. switching from using open-source fine-tuning to directly integrating GPT’s API. After all, users care about efficiency and cost-effectiveness, not the nitty-gritty of our tech.
Character.ai was also making waves back then. Amidst a declining economy, the AI sector was thriving like never before. For us, first-time entrepreneurs, experiencing such a monumental wave felt exhilarating yet overwhelming — Whilst the AI gold rush, the competition intensified, especially with OpenAI, who seemed to be everywhere, doing everything, rapidly and efficiently.
Our initial thought of being years ahead in understanding AI turned out to be just a few months. By the time we developed our product, the tech giants were already on our heels. In today’s era, startups either need to look further ahead, focusing on products with a perceived five-year window (which might actually be just two), or concentrate on niche markets overlooked by the giants. We chose to aim big still — it’s a once-in-a-lifetime opportunity.
After a reflective holiday season, I realized AI’s transformative journey through four stages:
- AI changing distribution — already a reality, birthing mega-corporations.
- Natural language altering UI — as seen in recent developments like copilot.
- Agents as a new software paradigm — exemplified by the recent GPTStore.
- The ultimate AI-native interaction — natural language chats and a flexible, immersive whiteboard interface. This is the future.
We were playing in the third stage with AI agents. Now, we’re setting our sights on the fourth stage, the Canvas. In June, we developed the earliest version of Canvas. By early July, we launched MindOS on ProductHunt, featuring Agent creation, a marketplace, and Canvas interaction capabilities. It was our first step onto the international stage and quickly became one of the most acclaimed AI products.
Seeing ByteDance’s coze.com, with many features similar to ours, was incredibly validating for our small team.
However, being a small company with big ambitions is like David versus Goliath. In the mobile internet era, startups could leverage advanced foresight. Now, AI has lowered the barrier to innovation, making data and design less distinctive. The remaining barriers are scale, capital, and supply chain.
Before we could fully leverage our scale advantage, OpenAI launched GPTStore. Some of our team members were understandably anxious, but we had anticipated such moves, just not at OpenAI’s breakneck pace. We held an emergency strategy session and reaffirmed our commitment to a bigger, longer, and more ambitious path. Our goal: to be LLM agnostic and the all-in-one interface.
Before we could fully process our anxiety, Sam Altman was ousted from OpenAI, only to be rumored to join Microsoft. Then, in a stunning reversal, he returned to OpenAI and restructured the board. This also revealed a bombshell: AGI surpassing human intelligence might be closer than we thought.
As a startup still finding our footing, this AI whirlwind has been a rollercoaster. One moment we’re soaring high, the next we’re plummeting. It’s an unpredictable and exhilarating journey.
So, what’s OpenAI’s ultimate goal? What’s the endgame for AI? I have my theories and invite everyone to join the conversation and explore this fascinating future together.
Please read through part II — IV here:
More about me:
- Product homepage: https://mindos.com
- YouTube: https://www.youtube.com/@Mind-OS
- Twitter: @KissonL
- LinkedIn: https://www.linkedin.com/in/kisson-songqi-lin-49309516/
- Email: k@mindverse.ai
