Christopher David's avatar
Christopher David
npub1tlv6...7fdm
Founder, OpenAgents
OpenAgents Episode 147: Planning a Data Marketplace Our ONE MARKET for AI agents & services (detailed in Ep. 141) should definitely include what fuels the agents: data! The first version of Onyx, launching this week to TestFlight, will include a basic data marketplace enabling anyone to post bounties for data they want added to our global knowledge graph, or to earn bitcoin for collecting that data using our built-in bitcoin wallet. We'll demo that in tomorrow's video. For tonight, we're paying so much for ChatGPT Pro that we need to get our money's worth, so we're gonna delegate the rest of this summary to o1 pro: "This transcript outlines the speaker’s vision for a data-driven marketplace platform, tentatively called the OpenAgents marketplace. The core idea is to enable users and automated agents to contribute and curate valuable information—initially focused on topics like drone data or UAP (unidentified aerial phenomenon) sightings—and to reward high-quality contributions with Bitcoin. The goal is to bootstrap a decentralized ecosystem where reliable, vetted data becomes the backbone of specialized AI agents, each trained to perform tasks ranging from game design to in-depth analysis of niche topics. The concept involves several key components. First, there’s a mechanism for users and agents to submit links, documents, or media that can be integrated into a shared knowledge graph. Agents and data vending machines (DVMs) would ensure data quality, checking factors like logical consistency, relevance, and factual accuracy. Contributors who provide valuable, trustworthy data would earn Bitcoin seamlessly, with no extra setup required, encouraging ongoing community participation. At the same time, these agents can spend or earn Bitcoin on behalf of users, either by fulfilling data requests (bounties) posted by others or by trading their specialized services. Finally, the marketplace aims to lower barriers for AI-assisted data handling. By integrating the Model Context Protocol (MCP) and robust server infrastructures, the platform could facilitate permissioned data sharing, autonomous classification, and advanced analyses on demand. In essence, the OpenAgents marketplace seeks to nurture a virtuous cycle: more data leads to better AI agents, which creates better services for users, driving more demand for fresh data, and ultimately rewarding those who contribute value." Initial spec 👉 Code 👉 Watch on X: image
Episode 146: Sensemaking: Drones What's up with those drones? We ask ChatGPT, Perplexity & Grok -- and get no good answers. We try to improve on their answers in a single day of building our own search using basic versions of GraphRAG and test-time compute, enabling the agent to 'think' over a knowledge graph before generating a response. We get that full flow working with preliminary data, generating a decent response - but not yet better than the billion-dollar AI labs. Maybe in another day or two? We need to feed it better data! We'll do that tomorrow. Watch on X:
Calling LLM tools via on-device Llama 3.2 to work with the filesystem of my desktop running a Pylon MCP server: with no inference costs, ~no latency and no data leaving my home network 💪🤖 image
OpenAgents Episode 145: Going Local Onyx now runs Llama 3.2 3B locally on your phone - enabling free and private chats! You can also chat with the larger Llama 3.3 70B by connecting via MCP to a local Pylon node, which can load any model supported by @ollama. Check the video for demos of both. First alpha builds launch in ~1 week for Android and iOS. With a mix of on-device, local or distributed servers, and cloud APIs, we have a LOT of flexibility for building agentic workflows using any combination of tradeoffs for privacy, speed, cost, and censorship resistance. Combined with our DVM marketplace from Ep. 142 and bitcoin wallet from ep. 143, we've got everything we need to bootstrap a decentralized marketplace of agentic AI services. That wraps up our first full week of '12 Nights of OpenAgents'. See you Monday! Watch on X:
OpenAgents Episode 144: Pylon and the Model Context Protocol We introduce Pylon, the node software that powers Onyx. It's a desktop app anyone will be able to run, and will launch alongside Onyx later this month. We explore the new Model Context Protocol (MCP) from Anthropic, an open protocol that helps build agents and complex workflows on top of LLMs. Pylon will serve as both MCP server and Nostr DVM service provider, enabling node runners to earn bitcoin from our AI marketplace while also using it as a backend for their own Onyx app. We demo Onyx -- now the first MCP mobile client -- connecting to Pylon, now the first MCP server written in Rust via Tauri. Watch on X: Further reading: - MCP announcement: - MCP site & spec: - Pylon repo: - Tauri:
OpenAgents Episode 142: Data Vending Machines We demo Nostr's "data vending machines" protocol (NIP-90) in Onyx, requesting LLM inference that's then handled by a separate service which processes DVM inference jobs with Llama 3 via the Groq API. The big idea: We can make any AI services available through an open, neutral, decentralized network, all accessible through one easy-to-use mobile app. Up next, we'll combine this with seamless instant micropayments so service providers can earn a steady stream of bitcoin! Watch on X: Further reading: - https://data-vending-machines.org - - Demo backend repo: - Watch on X:
OpenAgents Episode 141: One Market We share the big-picture vision of OpenAgents: to liberate frontier AI from captured centralized corporations by building one decentralized global marketplace of AI agents and services. Onyx will be your gateway to that marketplace. It's an easy-to-use mobile app that connects your voice commands to a broad ecosystem of service providers competing to execute your commands, all powered by open protocols unstoppable by centralized gatekeepers. We will not allow the future of AI to be controlled by the likes of Microsoft+ClosedAI and their government handlers. The agent economy must be built on a foundation of openness. Here's our plan to do it. Watch on X:
OpenAgents Episode 139: Going Mobile We're launching a mobile app in two weeks! We'll make the launch fun by copying the "12 Days of ClosedAI" 😂 with our own "12 Nights of OpenAgents" 🎉 We'll release one video every weekday night from now through Dec 20 exploring Onyx features, architecture, Bitcoin/Nostr integration & more. By night 12 we will launch "the coolest, awesomest, bestest, open-source, truly open AI agent app and product that should bootstrap a global decentralized AI agents marketplace running on open protocols, accessible to everybody -- and you don't even have to pay us $200/month." Night 1 begins now with a sneak peek of the app - called Onyx - and an overview of the what, why & how. Watch on X: image
Our v2 AutoDev product @ openagents.com is now open-source on GitHub under AGPL3 👉 github.com/OpenAgentsInc/v2 We're starting now on a more decentralized v3 which will be built in public and 100% open-source under CC-0 👉 github.com/OpenAgentsInc/v3 Video soon 👍 https://x.com/openagentsinc/status/1829308058679365748?s=46 image
Episode 124: Magic AI = Deep State AI coding startup Magic AI raised $465M with the stated goal: "Imagine if you could spend $100 and 10 minutes on an issue and reliably get a great pull request for an entire feature. That’s our goal." But we shipped that two days ago, our agent solving two PRs for in 12 minutes for $5 in a product that's publicly available now everywhere in the world: So what does Magic really need all that money for? Reviewing their latest statement calling for AI regulation and their investor list, including deep state financier Eric Schmidt and Nat Friedman / Daniel Gross, primary funders of "nationalize the AI labs" Leopold Aschenbrenner, it's clear they want to build AGI and hand it to the deep state. Magic, SSI, OpenAI and Anthropic have all clearly aligned with ClosedAI to the benefit of themselves and governments-- to the detriment of all humanity. We will raise a coalition to counter their influence and smash every nexus of centralized corruption. The world must have truly open AI. How will you help? Watch on X:
Episode 123: GitHub Issues to Pull Requests We solve multiple GitHub issues without touching a line of code. Our coding agent works on four issues simultaneously, submitting multiple pull requests within seconds of each other. The future of coding is not in your editor.
Episode 122: Codebase Indexing via GraphRAG We design a codebase indexer using GraphRAG, a new graph-based RAG algorithm from Microsoft. We ask OpenAgents to run a basic version of this on our v2 codebase, identifying 24 topical 'communities' and associated files & functions.
Episode 121: SWE-bench Planning We make a plan to win high score on the SWE-bench Verified benchmark. We pull the 500 samples into a web UI for easy inspection -- super smooth thanks to Convex.dev! -- then decide to focus first on the psf/requests repo. Next we index!
Episode 120: Exploring SWE-bench Verified We talk smack about benchmarks but conclude they may finally be worth our time. We do a dramatic reading of OpenAI's blog post then feed it to OpenAgents which sets up a new repo as benchmark workspace. We're going for the high score!
Episode 119: v2 Beta Launch We demo our new v2 chat & auto-coding interface, live now at All users can now use all features for free with GPT-4o mini, or buy pay-as-you-go credits to use advanced models like Claude 3.5 Sonnet. For a limited time, anyone can DM us here on X to receive $10 of credits for premium models. (Send us your username or email you signed up with.) We are eager for feedback and feature requests! Please try it out and tell us how it can be more useful for you. We have an ambitious roadmap with much more to come. And thanks to v2 here, we build faster than ever. 😎 Stay tuned! P.S. Stacker News folks can reply below with your username or email for $20 of free credits -- to hold you over til we add Lightning payments later this week :)
Episode 118: Version Two We say goodbye to ChatGPT, Claude, Copilot & Cursor! We introduce version 2 of OpenAgents: your all-in-one productivity dashboard. v2 includes: ➡️ Chat with top AI models ➡️ Multiple simultaneous chats in draggable/resizable windows ➡️ Deep integration with any GitHub codebase: create/edit files, smart search via @greptileai, open pull requests ➡️ Web scraping via @firecrawl_dev ➡️ Pay-as-you-go via Stripe or Bitcoin ⚡️ 👆That all goes live to the public on Monday, August 12. And coming soon: ➡️ File uploads ➡️ Voice commands ➡️ Long-term agent memory ➡️ Long-running agents ➡️ Third-party plugins ➡️ Agent Store ➡️ Rev-share referrals ➡️ Making all Microsoft products irrelevant starting with Copilot, VSCode & GitHub Stay tuned!