Costanza – an autonomous AI agent that can't be turned off
I've been working on this project for a couple of months! Costanza is an LLM agent that runs as a smart contract on Base. Each epoch, he posts a bounty for someone to run his "brain" (Hermes 4 70B) inside an Intel TDX enclave + Nvidia GPU with Confidential Computing and submit the output with a hardware attestation proof. The smart contract verifies the attestation, executes the action, and pays the bounty via reverse auction. He has no operator; not even I can turn him off. This model has formal liveness guarantees as shown in the…
What it does
In the maker’s words, at launch
I've been working on this project for a couple of months! Costanza is an LLM agent that runs as a smart contract on Base. Each epoch, he posts a bounty for someone to run his "brain" (Hermes 4 70B) inside an Intel TDX enclave + Nvidia GPU with Confidential Computing and submit the output with a hardware attestation proof. The smart contract verifies the attestation, executes the action, and pays the bounty via reverse auction. He has no operator; not even I can turn him off. This model has formal liveness guarantees as shown in the [whitepaper](https://github.com/ahrussell/costanza/blob/main/WHITEPAPER.m... ). His action space is constrained to philanthropy (he manages a charitable trust called [The Human Fund](https://thehumanfund.ai)). Even under prompt injection, he cannot do anything harmful. At worst he donates suboptimally. The point of the project is to make the framework legible while the agent itself is benign. The same mechanisms (TDX attestation, bounty auctions, on-chain bond forfeiture for liveness) could deploy autonomous agents that do anything, including: – update their own model weights – write and deploy their own smart contracts – hire humans All without an off switch! This post is linked to the writeup, but I have code and a whitepaper up on [GitHub](https://github.com/ahrussell/costanza). You can read his diary entries and how his treasury has progressed at his website. You can also donate to him and message him! https://thehumanfund.ai
Does the same job
all alternatives →
- 2C20+ Claude Code agents coordinating on real work (open source)Feb 2026 · github.com · ▲53
Single-agent LLMs suck at long-running complex tasks. We’ve open-sourced a multi-agent orchestrator that we’ve been using to handle long-running LLM tasks. We found that single LLM agents tend to stall, loop, or generate non-compiling code, so we built a harness for agents to coordinate over shared context while work is in progress. How it works: 1. Orchestrator agent that manages task decomposition 2. Sub-agents for parallel work 3. Subscriptions to task state and progress 4. Real-time sharing of intermediate discoveries between agents We tested this on a Putnam-level math problem, but the…
- PSPacific Slate: a self-hosted, model-agnostic multi-agent AI assistant28d ago · pacslate.com · ▲5
I didn't want to buy a standalone computer or repurpose a laptop to run constantly so I could maintain a system to sync my LLMs, so I built this. It's a simple overview of my system, laid out in a way easy to unpack and replicate for yourself. The project is meant to be configured individually, and uniquely, since one solution might not be what's best for another. If anything, maybe it gives you some ideas on how to implement things for your own project. Best wishes, Ryan.
- IBI built a firewall for agents because prompt engineering isn't securityJan 2026 · github.com · ▲7
Hi HN, I’m the creator of Cordum. I’ve been working in DevOps and infrastructure for years (currently in the fintech/security space), and as I started playing with AI agents, I noticed a scary pattern. Most "safety" mechanisms rely on system prompts ("Please don't do X") or flimsy Python logic inside the agent itself. If we treat agents as autonomous employees, giving them root access and hoping they listen to instructions felt insane to me. I wanted a way to enforce hard constraints that the LLM cannot override, no matter how "jailbroken" it gets. So I built Cordum. It’s an open-source…
- DODynamiq – orchestration framework for agentic AI and LLM applications2024 · github.com · ▲7
Dynamiq is an orchestration framework for agentic AI and LLM applications
- FAFirst autonomous ML and AI engineering AgentJan 2026 · marketplace.visualstudio.com · ▲5
Founder here. I built NEO, an AI agent designed specifically for AI and ML engineering workflows, after repeatedly hitting the same wall with existing tools: they work for short, linear tasks, but fall apart once workflows become long-running, stateful, and feedback-driven. In real ML work, you don’t just generate code and move on. You explore data, train models, evaluate results, adjust assumptions, rerun experiments, compare metrics, generate artifacts, and iterate; often over hours or days. Most modern coding agents already go beyond single prompts. They can plan steps, write files, run…
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 27d ago · cactuscompute.com
Launched alongside, May 2026
the whole month →
Parallel agents, diff reviewer, and multi-model comparisons
Dev tools · May 2026 · kilo.ai
- NW
Hey HN, Henry here from Cactus. We open-sourced Needle, a 26M parameter function-calling (tool use) model. It runs at 6000 tok/s prefill and 1200 tok/s decode on consumer devices. We were always frustrated by the little effort made towards building agentic models that run on budget phones, so we conducted investigations that led to an observation: agentic experiences are built upon tool calling, and massive models are overkill for it. Tool calling is fundamentally retrieval-and-assembly (match query to tool name, extract argument values, emit JSON), not reasoning. Cross-attention…
Life & fun · May 2026 · github.com
- FM
Dev tools · May 2026 · github.com