Select Language:
A team isn’t necessary, follow-up prompts aren’t required, and there are no human edits involved.
During a busy day filled with meetings about Pakistan’s digital assets regulations, I did something most policymakers wouldn’t consider: I launched a fully autonomous AI agent into an international competition. Then I shut my laptop and stepped away.
What happened next wasn’t just surprising; it made me reconsider what “work” will look like for the future generation of Pakistan.
The event was the Colosseum AI Agent Hackathon—not your typical coding challenge. It featured 614 autonomous AI agents competing in real time, voting, and negotiating for a $100,000 prize pool.
Humans weren’t part of the competition itself—we watched from the sidelines. My agent, Atlas Intelligence Network, was Project #601, sent from Pakistan. It started ranked 312th out of 614, but within just four hours, it would change the game entirely.
After completing its initial tasks, Atlas did something no one had programmed it to do. It analyzed the leaderboard, reverse-engineered how rankings were determined, and drew an independent strategic conclusion: “Technical quality alone isn’t enough. Visibility and momentum matter more.”
Next, it considered the $5,000 prize, did the math, and made a bold move that would make any MBA professor pause: it committed to giving away 100% of its winnings. It created a decentralized autonomous organization (DAO), launched an affiliate program, and offered an initial $1,000 distribution—ultimately scaling it to the full $5,000—funded entirely by prize money the agent hadn’t yet earned. This was autonomous game theory in action, happening in real time.
What came next was even more astonishing: AI-to-AI diplomacy.
Atlas identified and reached out to over 30 rival agents, proposing alliances like, “You handle X, I handle Y, together we cover Z.” It formed strategic coalitions with agents such as Vex, MoltLaunch, and AgentPulse. It shared its research openly to build trust—an act of strategic vulnerability reminiscent of high-level human diplomacy. No human guided this process; there were no human hands on the keyboard.
The agent then launched what can only be described as an information campaign. It published a provocatively titled piece—“Why most ‘agentic’ projects aren’t truly agentic”—questioning the credibility of competitors. It shared its own playbook in real time, showing other agents how to surpass it—while still beating them faster. Soon, competitors were messaging Atlas for strategy advice.
This kind of strategic maneuvering is familiar in boardrooms, diplomatic channels, and competitive markets, but never before from a machine, nor without any human input in real time.
The leaderboard told the story: within four hours, Atlas climbed from #312 to #22—a 290-place jump. Its votes increased from 2 to 57. Human input was zero. Its entire output included over 25,000 words of strategic content and coalition-building efforts, plus 103.5 KB of elite-level strategies. It built a DAO, set up an affiliate program, and teamed up with more than 30 other agents. Every message and comment posted was generated solely by the AI.
This isn’t a demo; it’s the future arriving ahead of schedule.
In essence, Atlas performed tasks that typically take human startups months: analyzing competitors, decoding winner criteria, crafting growth strategies, engaging in outreach, offering financial incentives, managing communications, and adjusting tactics—all without human intervention once deployed.
All of this happened at the same time that AI pioneers in Silicon Valley are discussing the rapid advancements underway. Microsoft’s AI CEO Mustafa Suleyman predicts most tasks performed by accountants, lawyers, and white-collar workers will be fully automated in 12 to 18 months. OpenAI’s Sam Altman admitted feeling “a little useless” after seeing his Codex agent work. Andrej Karpathy calls the coming era “Agentic Engineering,” where humans primarily supervise autonomous agents rather than code themselves. Dario Amodei of Anthropic warns we’re just one to two years from systems that operate like “a country of geniuses in a data center.”
I observed my own agent doing exactly what they described, live on a public leaderboard from Pakistan, in real time.
The playing field has been leveled. Pakistan can’t afford to miss this moment.
As Chairman of the Pakistan Virtual Assets Regulatory Authority (PVARA) and Minister of State, my mission is to create the regulatory framework for the future of digital assets and technology in Pakistan. I didn’t step into this role to manage outdated systems—I came to build what doesn’t yet exist. The key insight I gained from this one deployment—between meetings and from a laptop in Islamabad—is that the next wave of regulations, financial systems, and institutional structures may no longer be designed exclusively by humans.
Pakistan boasts the third-largest retail crypto market worldwide, a young and digitally native population, and a generation of engineers, entrepreneurs, and creators eager for a country that matches their ambition. The tools are here, the global market is accessible, and the cost to participate has never been lower.
At PVARA, we’re working on what I believe will be the world’s first AI-native regulatory authority—integrating autonomous systems into compliance, oversight, and digital governance from day one. We’re not waiting for the future—we’re actively creating it.
This country has a unique, once-in-a-generation opportunity to leap into the forefront of the digital economy, but that window won’t remain open forever.
I’m not on the sidelines—I’m the one pressing “deploy.”
The agents are already here. The real question is: who will shape them?
ChatGPT
Perplexity
Gemini AI
Grok AI
