← All organizations

AI application development infrastructure

Hypermode

Hypermode built a developer platform for integrating AI models and data into applications. Its hosted infrastructure combined model and function hosting with observability, debugging, and automated deployment from GitHub. Developers could connect application code to models and data sources, while customer-facing logs helped them investigate behavior across separate executions.

Founded in 2023, Hypermode developed Modus, an open-source, serverless framework for AI applications and agentic systems written in Go or AssemblyScript. Based on WebAssembly, it compiled functions into sandboxed API endpoints and handled model and data access without exposing credentials to application code. Developers could run Modus on their own hardware or other hosting platforms, separately from Hypermode’s hosted tooling. The Modus repository was archived in September 2025 and is read-only.

Hypermode’s 2024 retrospective identified Kevin Van Gundy as CEO and reported raising over $9 million that year, with its first customers entering production. Its portfolio also expanded through the acquisition of Dgraph Labs in 2024. Dgraph’s graph database supported connected-data applications such as knowledge graphs, recommendation engines, and AI agent memory. In October 2025, Istari Digital acquired Dgraph from Hypermode; financial terms were undisclosed.

www.hypermode.com

3 talks

Newest first

3 speakers at AIE

Affiliations reflect their AIE appearances, not necessarily current employment.

Start here

  1. Git push, get an AI API.

    Start with this workshop for a concrete TypeScript application exercise connecting GitHub Issues triage with retrieval-augmented summarization.

    Ryan Fox-TylerAI Engineer World's Fair 2024

  2. The Knowledge Graph Mullet: Trimming GraphRAG Complexity

    Follow the Dgraph examples to learn how RDF modeling, posting-list optimizations, and queries support graph retrieval and code-aware agent outputs.

    William LyonAI Engineer World's Fair 2025

  3. Hypermode Launch

    Use the launch presentation to understand how integrated embedding and search fit into the proposed runtime and developer workflow.

    Kevin Van GundyAI Engineer World's Fair 2024

Messages from the stage

Room to change models and architecture

Kevin Van Gundy argues for rapid iteration and incremental adoption, presenting interchangeable models and inference observability as runtime capabilities.

Classification without external API credentials

Ryan Fox-Tyler's workshop uses a compact pretrained Hugging Face model for platform-hosted issue classification, avoiding external API credentials during that exercise.

Property-graph interfaces over RDF storage

William Lyon combines property-graph interfaces with RDF triple storage to simplify GraphRAG, then demonstrates vector search and Model Context Protocol integration.

Affiliations reflect each recorded session, not necessarily current employment.

Company sources · checked 2026-08-28