Introducing Hyperspace: A Decentralized AI Network for Personal Computers

Join Nicholas Schlapfer, AI engineer at Hyperspace, as he introduces their new decentralized AI network and showcases their innovative product, Hyperspace, which empowers users to build customized workflows and leverage diverse models for AI-driven experiences.

  • 1. Speaker: Nicholas Schlapfer, an AI engineer at Hyperspace.
  • 2. Hyperspace is a decentralized AI network with no GPUs.
  • 3. It's building a community of users who contribute resources from their personal computers to the network.
  • 4. The network utilizes a product called AIOs, available for Windows and Mac.
  • 5. AIOs uses Llama CPP and performs inference.
  • 6. Hyperspace aims to provide diverse AI models, believing that multiple great experts will offer a better user experience than relying on one large, closed-source model.
  • 7. The company is developing a new product called "Hyperspace."
  • 8. Hyperspace combines prompt engineering, visual React Flow, Python execution, and web browsing.
  • 9. One goal of the new product is to create fine-tuned models that output genetic planning experiences.
  • 10. Users input queries and receive a directed acyclic graph (DAG) or JSON format, representing a methodical plan from the query.
  • 11. The system includes an in-house web scraping experience using Puppeteer and Beautiful Soup to convert HTML into markdown for LLMs to digest.
  • 12. Hyperspace features a node editor terminal for power users, allowing customization of each node.
  • 13. Users can change titles, task descriptions, expected outputs, and add as many nodes as desired.
  • 14. After execution, each output comes from individual nodes, creating queries based on tasks, goals, and previous node information.
  • 15. The system utilizes reasoning models (Quen 2 Instruct) and summarization models (Llama 370b).
  • 16. These models provide diverse synthesized answers for outputs containing Python code.
  • 17. Hyperspace aims to build out primitives that an agent would need, including memory, Python execution, planning, and code generation.
  • 18. The new product will be available later in the week via GitHub.
  • 19. Attendees are encouraged to follow Hyperspace AI on Twitter for updates.
  • 20. Hyperspace looks forward to releasing the product and getting feedback from users.
  • 21. The company believes that providing core primitives will help create agentic behavior in the future.
  • 22. The new product aims to provide a diverse set of synthesized answers using advanced models for reasoning and summarization.
  • 23. Hyperspace emphasizes user control, allowing users to edit titles, task descriptions, expected outputs, and add nodes as desired.
  • 24. Users can execute tasks and see the output in a terminal that automatically opens with the desired result.

Source: AI Engineer via YouTube

❓ What do you think? What do you think is the most significant implication of decentralized AI networks like Hyperspace, and how might they shape the future of artificial intelligence? Feel free to share your thoughts in the comments!