The model is powerful enough. What comes next?
Announcing the AIA Ecosystem Fund
Who decides what it does?
We've been sitting with that question at DeAgentAI for two years now. The past two years have been a blur of model releases and benchmark breaks. But we've always cared more about something else: when an AI Agent makes a decision on-chain, how do you know that decision is trustworthy? Who guarantees it won't contradict itself? When something goes wrong, who's responsible?
No easy answers. But these are the questions we've been working on since day one.
The AIA Ecosystem Fund is our next answer.
What we're seeing
AI Agents are no longer a concept. From automated workflows to on-chain autonomous trading, Agents are moving from demos into real running systems. But a healthy Agent ecosystem can't run on infrastructure alone. It needs an application layer, compute support, and real-world scenarios to validate against.
The compute problem is more urgent than most people realize. Inference demand is exploding, but existing GPU solutions are hitting their ceiling on cost and energy. Our view: if the cost per inference unit can't drop significantly, high-frequency autonomous decision-making by on-chain Agents is theoretically possible but practically out of reach. That's not a problem software can fix.
So the AIA Ecosystem Fund bets on two directions: AI Agent application layer, and Physical AI hardware infrastructure.
Who we backed
AliceAI
The first time we sat down with the AliceAI team, they asked us a question: in an age of information overload, what's actually scarce?
Not data. There's more data than anyone can process.
What's scarce is judgment, the ability to take a pile of contradictory information and arrive at one clear, executable conclusion. That's what AliceAI does. Across crypto markets, sports events, political outcomes, and public affairs, it compresses complex, fragmented information into a single signal: direction, confidence level, validity window. Every call goes one way. Results are recorded on-chain and can't be revised after the fact. Users receive signals via Telegram Bot, they can act on them manually, or authorize automatic execution.
What we're backing isn't just the product. It's the direction — verifiable judgment as a primitive. Prediction markets are becoming one of the most important information-pricing mechanisms on-chain. Alice has a shot at becoming the core entry point for that category.
ASIC AI Chip Project
The first thing this team said to us: GPU is not the destination. It's a bridge.
We agree. An ASIC chip purpose-built for Transformer architecture can do things GPU solutions simply can't, on inference speed, cost per unit, and energy efficiency. This project is targeting a 10x reduction in cost per token processed versus current GPU setups. Not incremental improvement. An order-of-magnitude gap.
For us, there's a more direct implication: lower inference costs mean on-chain Agents can actually run at high frequency. We came in at the seed round because this window won't stay open.
What this means for the DA community
The AIA Ecosystem Fund isn't a standalone financial vehicle. Every project we back will have a real connection to DeAgentAI's infrastructure — Alice's judgment signals can feed into AlphaX's execution layer, and the ASIC chip directly lowers inference costs across the entire network.
For AIA holders, ecosystem expansion means more real use cases, more developers and users entering the network. We're not doing portfolio allocation. We're building an ecosystem.
We're still looking
The AIA Ecosystem Fund is early. We're fully open to strong projects.
Leading the fund is our Co-founder Selwyn, who has transitioned from DeAgentAI to take full charge of AIA Ecosystem Fund operations. Selwyn brings years of experience at top-tier USD funds, with multiple successful exits and a total return across all investments exceeding $4 billion.
AI Agent application layer and Physical AI hardware infrastructure are our current focus areas, but they're not our boundaries. If what you're building has anything to do with the future of AI, we want to talk.
No deck required. No polished metrics needed. We just want to hear your most fundamental take on where this is going.
Website: deagent.ai
X: @DeAgentAI
Discord: discord.gg/officialdeagentai