Skip to content

Releases: MobileGraph/mobilegraph

0.4.0-alpha Release

0.4.0-alpha Release Pre-release
Pre-release

Choose a tag to compare

@bharatjyoti bharatjyoti released this 13 Jul 14:39
a5aaa1c

Multi Model support added - OpenAI, Gemini, OpenRouter, Claude

v0.0.3-alpha

v0.0.3-alpha Pre-release
Pre-release

Choose a tag to compare

@bharatjyoti bharatjyoti released this 07 Jul 18:50
c3f4d7f

v0.0.3-alpha introduces the Agent Runtime, transforming MobileGraph from a collection of AI utilities into a complete AI application framework capable of orchestrating stateful, multi-step workflows.

This release establishes the architectural foundation for building intelligent applications using reusable agents, graph-based execution, tools, and memory while remaining fully Kotlin Multiplatform compatible.

MobileGraph v0.2.0-alpha

Pre-release

Choose a tag to compare

@bharatjyoti bharatjyoti released this 30 Jun 06:49
731ce58

MobileGraph v0.2.0-alpha

We're excited to announce the second alpha release of MobileGraph, a Kotlin-first framework for building AI applications, Retrieval-Augmented Generation (RAG), and future agentic workflows on Android and Kotlin Multiplatform.

This release completes the Knowledge Layer (Phase 2) and lays the foundation for the upcoming Agent Runtime.

✨ Highlights

Core Runtime

  • Provider abstraction
  • Session management
  • Middleware
  • Event system
  • Chat memory
  • Streaming support

Knowledge Layer

  • Document abstraction
  • Document loaders
  • Text splitters
  • Embedding models
  • Embedding stores
  • Vector stores
  • Retrievers
  • RAG pipeline
  • Context builders

Developer Experience

  • Kotlin DSL
  • Provider registry
  • Retrieval registry
  • Builder APIs
  • Passive observability
  • Kotlin Multiplatform architecture

Roadmap

Completed

  • ✅ Phase 1 – Core Runtime
  • ✅ Phase 2 – Knowledge Layer

Next

  • 🚧 Phase 3 – Durable Agent Runtime
  • 🚧 State Graph Engine
  • 🚧 Checkpointing
  • 🚧 Human-in-the-loop
  • 🚧 Multi-agent orchestration

Thank you to everyone following the project. Community feedback and contributions are always welcome.