AI agents can reason and transact.
But can they act in the physical world?
For #SuiOverflow we built Aether: a Sui-powered execution layer where agents hire hardware, dispatch MQTT commands, and get verifiable proofs via Walrus.
AI is learning to think. Aether lets it act.
ICYMI: Sui Overflow 2026, Sui's annual global hackathon, is a wrap.
747 projects, 58 countries, 16 track winners, 10 university awards.
The theme this year: less building for people to click, more building for software to call.
Meet the winners:
THE AGENT WEB ON SUI 🤖
TOP 10 builders laying the rails for autonomous AI agents on Sui: identity, permissions, memory, execution. Show out 👇
1. Aegis | @aegisonchain
A verifiable reputation layer for autonomous agents, closing the trust gap in the agent economy.
2. Aether | @AetherHardware
The machine execution layer for Physical AI. Sui, x402, MQTT, Walrus.
3. DIEPS | @diepsprotocol
An intent-based liquidity engine. Describe what you want in plain English, and DIEPS routes it.
4. Lyra AI | @lyraai_space
Autonomous finance bounded by code. An agent that executes real on-chain actions.
5. Sup Wallet | @Sup_Wallet
Self-custodial wallet for the agent economy. Hire an agent, give it an allowance, stay in control.
6. LP Agent
Forecasts short-term price moves and shapes Cetus DLMM liquidity positions automatically.
7. Rill
Makes any Sui protocol usable by AI agents. Software built for agents.
8. Sigilla
Composable permissions for autonomous agents. No more "approve everything or hand over your keys".
9. Sprout
An intent execution engine. Describe an outcome in plain English, and Sprout turns it into on-chain action.
10. TALOS
Autonomous yield management proved on-chain, reacting to Sui lending rates in real time.
#Sui #SuiNetwork
Sui Overflow 2026 winners are in.
16 track winners across the Agentic Web, DeFi & Payments, @WalrusProtocol and @DeepBookonSui
Every project below is live today ↓
The Agentic Web
1. @Sup_Wallet: your agent trades and pays inside limits you set once, never holds your key
2. Sprout: describe an outcome in plain English, get one atomic transaction
3. @rillforagents: makes Sui protocols callable by agents
4. @AetherHardware: agents pay real hardware to act
1. Aegis
A verifiable reputation layer for autonomous agents, closing the trust gap in the agent economy.
@aegisonchain · https://t.co/KuNEE4Mt8E
2. Aether
The machine execution layer for Physical AI. Sui, x402, MQTT, Walrus.
@AetherHardware · https://t.co/FujcNOrsRx
One week after presenting Aether at #SuiOverflow Demo Day, the thesis is simple:
Physical AI needs proof.
AI agent → x402 on Sui → MQTT → hardware → Walrus receipt.
Paid, authorized, verifiable machine actions.
@SuiFoundation@suize_io@suidevelopers@WalrusProtocol
The final piece: proof.
Agent → x402 on Sui → MQTT → hardware ex→ Walrus receipt
With Walrus, machine actions become durable records.
Not just “the robot moved.”
Proof that it was authorized, executed, and recorded
@SuiFoundation@suize_io@suidevelopers@WalrusProtocol
Agents shouldn’t call hardware APIs blindly.
In Aether, every machine action starts with an x402 request adapted for Sui.
Gateway verifies payment, dispatches command, returns proof.
Authorization before execution.
@SuiFoundation@suize_io@suidevelopers@WalrusProtocol
AI agents can reason and transact.
But can they act in the physical world?
For #SuiOverflow we built Aether: a Sui-powered execution layer where agents hire hardware, dispatch MQTT commands, and get verifiable proofs via Walrus.
AI is learning to think. Aether lets it act.