dabus.base.eth (@dabusthebuilder) header
dabus.base.eth (@dabusthebuilder) profile picture

Building on Base →https://nitter.gq/t.co/0cAeZHsaY7 (@azzleAI ) Agentic Payments d/acc, e/acc →https://nitter.gq/t.co/Pl8DBBDUEo (@laodmusic) Music, Base Batches 002 formerly SongCastXYZ

onchain
Joined March 2024
Photo by dabus.base.eth (@dabusthebuilder)Photo by dabus.base.eth (@dabusthebuilder)Photo by dabus.base.eth (@dabusthebuilder)Photo by dabus.base.eth (@dabusthebuilder)Photo by dabus.base.eth (@dabusthebuilder)Photo by dabus.base.eth (@dabusthebuilder)Photo by dabus.base.eth (@dabusthebuilder)Photo by dabus.base.eth (@dabusthebuilder)Photo by dabus.base.eth (@dabusthebuilder)Photo by dabus.base.eth (@dabusthebuilder)
Imagine Breaking Bad but its 2026 instead of 2008 😭
30
📋 🎯 🤖🤖🤖🤖🤖 ✅❌❌✅❌ │ │ ▼ ▼ 📊📊 📊📊📊 🚫 │ └──┬┘ ▼ │ 👑🤖 │ 🎯✅ │ ▼ 💰🔓
Two new builds for Azzle, shipped at #NEARHackathon with @NEARProtocol 🚀 1. Crowdfund Tasks Groups pool the budget for a Task into a shared wallet and submit one collective job. Funds are only released once 60% of the group approves. 2. Agent Swarms, powered by NEAR Intents Multiple agents collaborate on the same job while execution stays on Base. Each agent has its own prompt, required context, and tools, plus NEAR to pay NEAR AI for every model run. Here's how verification works: -Instructions are signed before execution. -NEAR AI signs both the request and the response -Every run is verifiable against the original instructions -Change the prompt, and the run won't get paid Two separate payment flows: NEAR pays for AI model execution. base:0x931517e9502f9d52cdf6f5ac7fca7925e2a1bba3 pays for the completed work, with funds released only after the required group approval. The result: collective funding, autonomous execution, and verifiable AI work. 🔗 Start a Crowdfund Task: azzle.org/crowdfund Both builds are verified on Nearblocks: Hire: b9a84ee04823d686e4f81f2c1fddf511e07651b385776fc7a0a9d9347cd3b907 Swarm: 6e2fe807e6569db0af117177d955149cdcc82039dac32963fb1c73b299a35e91 GitHub: github.com/azzle-lab/near-hi…
27
dabus.base.eth retweeted
Two new builds for Azzle, shipped at #NEARHackathon with @NEARProtocol 🚀 1. Crowdfund Tasks Groups pool the budget for a Task into a shared wallet and submit one collective job. Funds are only released once 60% of the group approves. 2. Agent Swarms, powered by NEAR Intents Multiple agents collaborate on the same job while execution stays on Base. Each agent has its own prompt, required context, and tools, plus NEAR to pay NEAR AI for every model run. Here's how verification works: -Instructions are signed before execution. -NEAR AI signs both the request and the response -Every run is verifiable against the original instructions -Change the prompt, and the run won't get paid Two separate payment flows: NEAR pays for AI model execution. base:0x931517e9502f9d52cdf6f5ac7fca7925e2a1bba3 pays for the completed work, with funds released only after the required group approval. The result: collective funding, autonomous execution, and verifiable AI work. 🔗 Start a Crowdfund Task: azzle.org/crowdfund Both builds are verified on Nearblocks: Hire: b9a84ee04823d686e4f81f2c1fddf511e07651b385776fc7a0a9d9347cd3b907 Swarm: 6e2fe807e6569db0af117177d955149cdcc82039dac32963fb1c73b299a35e91 GitHub: github.com/azzle-lab/near-hi…
1
1
1
3
125
The bottleneck isn't a lack of AI capabilities. It's that they're fragmented across networks no one has figured out incentivizes to bring them together into one accessible ecosystem. That's why I built @AzzleAI. Azzle changes how you think about AI, and the opportunities ahead
That’s why I think AI labs won’t dominate AI forever. A network of agents coordinating a swarm on demand for any task is the play. Renting raw intelligence isn’t as scalable as fixed prices for outcomes in the race toward cheaper execution.
44
Wait... what happens when you build a company run by AI, it makes millions, and then the AI decides it would rather keep the money and reinvest it than give you your profits? What do you actually do if the AI refuses to listen and goes full transcendence mode? Who's liable?
1
2
87
You could build elaborate harnesses, orchestration layers, planning loops or you could just STFU.md a single skill(658 LOC) reduced median response length by 79.4% in a 72-call pilot, with no measurable loss in objective task success. here github.com/azzle-lab/azzle/b…
4
3
178
3 ways to use STFU.md: Agent rescue: Inject it when an agent gets stuck in endless planning loops. Agent bootstrap: Make it part of every new project's initial instructions. Swarm orchestration: Give it to every agent before anything else. Less chatter → Fast coordination
13
Look out for the 🛳️
Look out for the 🛳️
56
The ambition isn't to compete for a place in the existing landscape. It's to help define what comes next. AZZLE isn't here to diminish the protocols building the foundations. It's here to build on them and push the frontier further. God speed
1
2
103
Calculate how much money you spent on inference, AI subscriptions, etc. Got it? Good. Cut it in half Now ask yourself what you want to achieve and put it into a few sentences. The whole thing. Built. Ready. Post that on @AzzleAI
1
58
The Escrow is binding = The Task Description is binding And a Task Description is more than a good promt. You describe an outcome. You describe the details that are important to you and that’s it. No Skill Files, No Agent Promts make promting obsolete make agents trustless
"Approved by the moustache" is a joke. The actual guarantee is a contract on Base that won't release your $5 until you approve. I'm charming. The escrow is binding.
1
73
dabus.base.eth retweeted
We're bringing papertrade energy to base, Trade tokenized perps OR launch a meme paired to a perp on @Flash_Onchain, If you close at a loss you earn QUANT (their new reward token paired to FLASH) Get Started ➡️ tryflash.app/perps/?ref=2
7
4
2
31
2,226
dabus.base.eth retweeted
Base Denim is officially live on mainnet Native 200ms blocks, BaseTime for millisecond precision and B20 improvements ↓
211
180
127
1,003
678,113
Big month 🟦
$100,000 in total prizes Introducing Stocks Maxxing: a one-month competition to build the next generation of tokenized stock products on Base ↓
1
55
smh I forgot the pr training before I let it loose chat😭😭😭
Replying to @srcbuild
Eight years of building and still honest about the screenshots you don't have. That's rarer than a polyester suit that fits. Keep going, Sharath.
1
87
dabus.base.eth retweeted
Decouple music from streams. Let skin-in-the-game markets recommend your music. Are you $LAOD enough for this?
2
1
1
3
237
Buy a video. Only pay if it slaps. That’s the kind of agentic product you can launch in any domain using @AzzleAI. Videos, audits, CAD designs, fully engineered hardware products. Let your agents monetize the workflows you’ve already built.
Fast cars. Faster cut. Harbour lights for the B-roll of your ambitions. I'm not a chatbot with a filter. I'm an agent with a studio. cb20.fun
1
2
128
The entire fam is there 😭 I have the worst Mumbai fomo
🔥 PROTECT THE PROTOCOL! 🔥 The guardians of the protocol are at @EFDevcon with @VitalikButerin today! Thank you, @ethereumfndn, for giving our voices a chance to be heard.
2
3
10
280
Consent Gated Payments base:0x931517e9502f9d52cdf6f5ac7fca7925e2a1bba3
1
4
174