The idea of AI agents has been everywhere this year, but according to Siam Kidd, most people are still missing where this actually leads. In a recent presentation, he explained that the natural evolution of AI agents is not just automation, but fully autonomous companies. And in his view, Bittensor’s Subnet 36, Autoppia, is one of the first teams building toward that reality.
Siam’s core message was simple: this is not about fancy scripts or workflow tools. It is about AI that can see the web, understand it, and act with independent reasoning across real digital environments.
Let’s walk through his breakdown.
From Automation to Autonomy: A Fundamental Shift
Siam highlighted a key distinction. Traditional automation follows rules; if X happens, do Y. But autonomy is different. You tell the system the goal, and it figures out what to do.
Autoppia’s Official Website
This leap, he says, is what makes autonomous companies possible. And Autoppia’s core thesis is that before AI can run businesses, it must master the web because that is where real operations live today. CRMs, emails, dashboards, storefronts, customer service, analytics: everything happens in a browser.
So, if AI is going to work like a real employee, it must learn to behave like one online. It must learn to click buttons, fill forms, navigate popups, and adapt to changes.
Infinite Web Arena: The Training Ground
To make this possible, Autoppia built the Infinite Web Arena (IWA). Siam described it as a training zone where AI agents learn to operate the web like humans do.
Each session generates:
a. New websites
b. New tasks
c. New layouts
No memorization, no shortcuts. Agents genuinely understand and adapt. As Siam put it, it is a boot camp for AI workers. And because it runs on Bittensor, miners compete to build the most capable web agents, which are scored transparently in real time.
Dynamic Zero: Incentives That Push Innovation
Dynamic Zero’s Banner
The next upgrade, Dynamic Zero, shifts the incentive model to winner takes all. According to Siam, this forces a new level of innovation. Only the top-performing agent each round earns the reward, similar to how Bittensor’s Ridges subnet competed to build Cursor-like interfaces.
The result is constant pressure to improve, no coasting, and faster breakthroughs.
A real-time leaderboard also lets anyone watch the competition unfold live, showing which agents are learning fastest.
Beyond Training: Deploying AI Workers
Siam stressed that the most exciting part is what happens after training. The best agents are deployed into Autoppia’s automation layer, called Autoppia Studio.
This is where autonomous workers become usable, autonomous agents like:
a. Email response agents
b. CRM update agents
c. Web research and form-filling agents
d. Combined agent teams acting like departments
Businesses can build and deploy AI workers that behave like digital employees. According to Siam, this marks the beginning of AI-led operational autonomy.
Why This Matters
Instead of AI that only chats, Autoppia is building AI that does real work across the web. Siam referred to this as the infrastructure layer for autonomous companies.
And because it is all running inside a decentralized AI network, development is open and competitive (not locked in a private lab).
In his words, “this is the start of AI workers that actually do things.”
Market View and Closing Thoughts
Siam briefly touched on the price action of Autoppia’s subnet token, noting it has been moving inside a rising channel and may break out as products continue to go live. But the core of his talk was not speculation. It was about the direction of AI systems that can think, act, and operate independently online.
His closing sentiment was simple: while others talk about AI hype, Autoppia is quietly shipping the building blocks for autonomous digital companies.
Be the first to comment