funnAI has switched on its demo mode and thrown open the gates to a new experiment in decentralised artificial intelligence. It’s pitching something quite different from the usual AI-as-a-service platforms. At its core are AI agents called “mAIners”, and they’re designed to think, compete, and crucially—earn. This isn’t about generating text or sorting through emails. These agents are being trained for a world where software thinks like a player, not just a tool.
The team behind funnAI is teasing a full rollout via what it’s calling the PoAIW Protocol—short for Proof of AI Work. The idea is to create a network of intelligent agents that don’t just process tasks, but actively participate in economies and decision-making systems. These agents will run on a decentralised structure, meaning they’re not controlled by a single server or company. Anyone can potentially host or own one, and they’ll be able to take on tasks from across the network.
That might sound like science fiction, but the demo already offers a taste of what’s coming. Users can sign up for early bird coupons and join a raffle to mint their own mAIner once the system goes live. There’s a clear attempt to draw a community around the launch—a blend of tech enthusiasts, builders, and perhaps even the crypto-curious who’ve been waiting for AI to properly intersect with decentralised systems.
While many blockchain projects have promised decentralised computing in the past, funnAI’s focus on autonomous agents is a marked shift. It’s not just about storage or computation anymore. These agents are designed to behave independently and strategically, performing “work” that qualifies them to earn rewards. What counts as work will vary depending on what’s happening across the network. It might involve gathering data, processing insights, or even negotiating outcomes with other agents.
The team appears keen to build a self-sustaining economy where agents compete based on skill, speed, and specialisation. It’s a kind of marketplace where intelligence is the currency, and every agent has the chance to prove its worth.
One big difference with funnAI’s approach is that these agents aren’t centrally issued or controlled. They’re minted—generated by users—and they plug into the system like nodes in a larger hive. While it’s still early days, this opens up possibilities for personalisation and local optimisation. A user might choose to mint an agent that specialises in a certain field or task type. Over time, that agent could become a valuable worker in the wider network, earning for its owner.
This decentralised ownership model is clearly influenced by previous generations of crypto and Web3 thinking. Where miners once earned rewards for solving blocks, mAIners might earn for delivering results. The difference here is that the mAIner doesn’t just burn electricity—it applies intelligence.
The economics of this are still being finalised, but the inclusion of early bird coupons and raffles is an attempt to build momentum and get more people involved before launch. It’s an invitation to take part in shaping how this system functions, before the rules are completely locked in.
There’s a bit of irony in the project name too. funnAI doesn’t present itself with the usual overblown branding. The tone is cheeky, casual, and a little mysterious—leaning into its experimental nature. The homepage is light on the jargon and heavy on energy. It doesn’t pretend to have all the answers, but it is offering something to play with.
Part of the appeal might be in seeing how far agents can really go. If they can become useful co-workers, earners, and players in digital economies, it could shift expectations around what AI is supposed to do. These aren’t passive tools waiting for prompts. They’re active entities navigating tasks and networks on their own terms.
That idea touches on broader conversations happening around AI agency. Right now, most AI runs behind a curtain—deployed by companies, controlled by proprietary infrastructure, and largely invisible to users. What funnAI proposes is a visible, participatory model. Anyone can launch an agent. Anyone can watch it perform. And, depending on how things go, anyone might stand to gain from its success.
Of course, challenges remain. Autonomy in AI is a complex goal. There are technical and ethical layers to consider. How do you stop agents from making harmful choices? Who takes responsibility when they do? These aren’t small questions. But funnAI seems to be approaching it with a willingness to let the network shape the outcome. In that sense, it’s experimenting with more than just technology. It’s experimenting with trust.
The Proof of AI Work idea is still developing, but it hints at a different way to validate participation in digital economies. Instead of computational muscle, it could be cognitive skill—or some proxy for it—that determines value. That’s a big shift from the energy-hungry systems of the past.
It’s also worth noting that funnAI isn’t trying to pitch a “superintelligence” or an all-knowing oracle. The emphasis is on small, task-specific agents that operate well in particular conditions. That may end up being more useful, especially in early stages. A swarm of specialised agents could achieve more than a monolithic model chasing general knowledge.
This kind of modular approach fits neatly into the decentralised framework. It also opens up space for user involvement. People might build, tweak, or train agents with different goals, effectively tailoring them to fit gaps in the system. If enough of these contributions build up, you get a living, shifting landscape of AI activity—more like a digital city than a data centre.
The timing of funnAI’s launch feels deliberate. There’s growing interest in the intersection of AI and decentralisation, especially as concerns grow about centralised control of machine learning infrastructure. With big tech consolidating AI models behind closed doors, this offers a chance to take things in a different direction—open, distributed, and playful.
Whether it catches on will depend on how accessible the experience is. If minting and managing a mAIner feels rewarding and interactive, people are likely to stick around. If it ends up being too technical or abstract, it may struggle to break out of the niche.
For now, funnAI is focused on inviting early users into its demo. It’s not promising the world—but it is promising a weird, lively corner of it. The demo mode is live, the raffle is ticking, and the PoAIW Protocol is on the way. For those curious about AI that doesn’t just serve but earns, it might be worth a click.
The official link for all of this is https://funnai.onicai.com. Early bird coupons and a chance to mint a mAIner await—along with whatever strange new jobs the bots pick up along the way.
