# Hyperspace More Nodes Is All You Need: Nicolas Schlaepfer AI Engineer · 2024-09-04 Nicolas Schlaepfer introduces Hyperspace's decentralized AI network and new product, a node editor for power users that generates agentic plans via a DAG orchestration model (HyperEngine v3). The network, with over 15,000 nodes, leverages consumer devices rather than GPUs, running inference through Llama.cpp. The product combines prompt engineering, visual React flow, Python execution, and RAG-like web browsing, using Qwen 2 instruct for reasoning and LLaMA 3 70B for summarization. Schlaepfer emphasizes diverse open-source models and a virtual file system for agentic primitives. Availability via waitlist is announced later this week. ## Questions this episode answers ### How does Hyperspace's agentic planning model generate task plans from a user query? Nicolas Schlaepfer explains that Hyperspace fine-tuned a model to output agentic planning experiences. When a user submits a query, the model returns a directed acyclic graph (DAG) in JSON format, representing a methodical step-by-step plan. This DAG then populates a node editor, giving users a visual overview and the ability to edit each step before execution. [1:47](https://aie.addtry.com/0a973a3d-4767-48aa-8b87-126cdb7be558?t=107000) ### How does Hyperspace provide AI inference without owning GPUs? Nicolas Schlaepfer describes Hyperspace as a decentralized AI network where community members contribute computing resources from their personal devices. Their AIOS app, available for Windows and Mac, uses Llama.cpp for inference. A user can download AIOS and get chat inference from another user's computer anywhere in the world, such as from someone in Belgium, eliminating the need for centralized GPU clusters. [0:32](https://aie.addtry.com/0a973a3d-4767-48aa-8b87-126cdb7be558?t=32000) ### What method does Hyperspace use to scrape and prepare web content for LLMs? Nicolas Schlaepfer states that Hyperspace employs an in-house web scraping system using Puppeteer and Beautiful Soup. It scrapes websites and then converts the retrieved HTML into Markdown format, making it easier for large language models to digest and use as context. This provides a RAG-like browsing experience integrated into their node-based workflow tool. [2:01](https://aie.addtry.com/0a973a3d-4767-48aa-8b87-126cdb7be558?t=121000) ### What are the key features of Hyperspace's node editor for power users? Nicolas Schlaepfer presents a node editor with a visual React flow where each node represents a task from the DAG plan. Users can edit titles, task descriptions, and expected outputs, and add new nodes. Execution uses a reasoning model (Qwen 2 instruct) and a summarization model (LLaMA 3 70B). It includes a terminal for running Python code and a virtual file system, building primitives for agentic behavior like memory and code generation. [2:29](https://aie.addtry.com/0a973a3d-4767-48aa-8b87-126cdb7be558?t=149000) ## Key moments - **[0:00] Intro** - [0:32] "We have no GPUs. We're building a community who takes the resources from their personal computer and contributes to our decentralized network." - **[1:17] Hyperspace** - **[1:47] Planning** - [1:47] Hyperspace's fine-tuned model outputs agentic planning experiences as a DAG in JSON format. - **[2:29] Node Editor** - [2:29] Hyperspace's node editor and terminal provides a power tool for power users, allowing modification of each node in a React flow and Python execution. - **[2:44] Demo** - [4:14] Hyperspace uses Qwen 2 instruct for reasoning and LLaMA 3 70B for summarization during execution to produce diverse synthesized answers. - **[5:08] Outro** - [5:19] "Happy to announce that it'll be available later this week via waitlist, so go ahead and pay attention to our Twitter: @HyperspaceAI." ## Speakers - **Nicolas Schlaepfer** (guest) ## Topics Agent Skill Design ## Mentioned AIOS (product), Beautiful Soup (product), HyperEngine (product), Hyperspace (product), Llama (product), Llama.cpp (product), Puppeteer (product), Qwen (product), React (product) ## Transcript ### Intro **Nicolas Schlaepfer** [0:14] Hello everybody, my name is Nicolas Schlaepfer. I'm an AI engineer. I think I'm at theright place. Today I'm going to be pronouncing a new product we've been working on at Hyperspace. A little bit about us: we are a decentralized AI network. We have no GPUs. We're building a community who takes the resources from their personal computer and contributes to our decentralized network. We have a product currently out called AIOS. You can download it for Windows and Mac. It uses Llama.cpp and does inference. So technically, you can download it, get inference from somebody in Belgium, and you can have a chat experience like that. So we're hoping to—oh, actually, sorry. We really believe in diverse models. We think not having just one big closed-source model is the answer for the best AI experience. ### Hyperspace **Nicolas Schlaepfer** [1:17] Having a mixture of a bunch of great experts will provide the best AI experience. So that gets us to our self-titled product, Hyperspace, which will be built on our network. This is a really interesting product. It's a mix of prompt engineering, visual React flow, Python execution, and RAG-like web browsing. So for the first thing with this product, we wanted to build a fine-tuned model that outputted agentic planning experiences. So you'd put in a query, you get in a DAG, or you get out a DAG in a JSON format. ### Planning **Nicolas Schlaepfer** [2:01] And this is a methodical plan from that query. We also wanted to have a primitive of having an in-house web scraping experience for LLMs. So we're using Puppeteer and Beautiful Soup to scrape websites, convert that HTML into a Markdown, something easier to read for LLMs to digest. Kind of the product we're going to talk about is we have a node editor and a terminal, and this is going to be a power tool for the power users. This node editor will allow you to change each node in the React flow to fit your needs. ### Demo **Nicolas Schlaepfer** [2:44] And this is coming from the DAG orchestration model we have. Here's a little video demo of our product. I recorded it yesterday in the hotel, so I'm sorry for the audio. And here we go. **Guest** [3:08] Welcome to a very early look at Hyperspace. Let's begin with a sample query. Once you submit your query, you're brought into our node editor view, where each node is streamed in from our DAG orchestration model, HyperEngine v3. We want to emphasize that the user is still in full control. They can edit the title, task description, and expected output. They also have the freedom to add as many nodes as they like. Let's execute. Now that our outputs are done, we can talk a little bit about the outputs. Each output is coming from each node, and each node is creating a query based on that task. It's combining the overall goal, the local goal, and whatever has happened in the previous node. We're using a reasoning model and a summarization model. For reasoning, we're using Qwen 2 instruct, and then for summarization, we're using LLaMA 3 70B. This helps provide a diverse set of synthesized answers. For the outputs that do have Python, we can go ahead and run them and see them in our terminal. The terminal will automatically open it up with the output. We want to provide the groundwork for agentic behavior in the future by providing these core primitives. Over here on ourright, we have our virtual file system that changes our directory. We're trying to build out all the primitives to what an agent would need: memory, Python execution, planning, and code generation. ### Outro **Nicolas Schlaepfer** [5:08] Thank you so much for watching. We're very excited to get this out in the coming weeks. Happy to announce that it'll be available later this week via waitlist, so go ahead and pay attention to our Twitter:@HyperspaceAI. Allright, thank you. That's my time. --- This library is powered by PodHood (https://podhood.com), the podcast website platform.