July 2025

Development recap for July 2025, highlighting new features, integrations, improvements, and fixes across GL AIP.

1. Basic Agent Interface (LangChain)

   Introduced a basic agent interface built on LangChain to simplify agent creation, execution, and integration.

2. Multi-agent Interface (LangGraph, LangChain, and Google ADK)

   Added support for multi-agent orchestration using LangGraph, LangChain, and Google ADK, enabling complex agent collaboration patterns.

3. MCP Interface (Google ADK and LangChain) with Podman Support

   Introduced an MCP interface compatible with Google ADK and LangChain, with Podman support for improved containerized execution.

4. A2A Interface (Google ADK and LangChain) with Podman Support

   Added an Agent-to-Agent (A2A) interface supporting Google ADK and LangChain, enabling direct agent communication with Podman-based execution.

5. BOSA Support

   Enabled BOSA support across the platform to expand tool interoperability and integration capabilities.

   1. BOSA Core – Logger (Anonymized Logging)

      Introduced anonymized logging in BOSA Core to improve observability while protecting sensitive data.

   2. BOSA Connector

      Added a BOSA Connector to simplify integration between agents and BOSA-powered services.

   3. BOSA Observability Tools for GLLM Agents (FastAPI)

      Introduced FastAPI-based observability tools for GLLM agents to enable monitoring, debugging, and performance analysis.

   4. BOSA Tools Interface for GLLM Agents

      Added a standardized tools interface to allow GLLM agents to seamlessly discover and invoke BOSA tools.

6. Simplified MCP Hello World and New GitBook Publication

   Simplified the MCP “Hello World” example and published updated documentation in a new GitBook covering LangChain, LangGraph, and Google ADK.

7. Demo AIP

   Released a demo version of AIP to showcase core capabilities, integrations, and agent workflows.

8. Overtime Agent Improvements Due to Organization Changes

   Continuously improved agent behavior and logic to adapt to evolving organizational structures and workflows.

9. Stress Test Agent Improvements

   Improved agent stability and performance under high-load scenarios through enhanced stress testing and optimization.

10. Benchmarking Open-source LLMs Using Agent Runner

    Added benchmarking for open-source LLMs using the Agent Runner, with initial recommendations such as Qwen / Qwen3-32B-AWQ.

11. Weekly Report Agent Integration with BOSA Connector (Google Drive)

    Integrated the Weekly Report Agent with the BOSA Connector for Google Drive to enable automated report generation and synchronization.

12. Enable Model Switching with LM Invoker on LangChain Agent

    Added an LM Invoker to LangChain agents to allow dynamic switching between language models at runtime.

13. Integration with GLChat Admin Dashboard via GL AIP Endpoints

    Integrated agents, tools, MCP, and runner management into the GLChat Admin Dashboard using GL AIP endpoints.