Workflow · Production

Ackgent: Rapid AI agent development on GCP with Google ADK and declarative Agent Config

The problem

Building AI agents from concept to production is too slow and overly complex, with developers bogged down in boilerplate code, infrastructure wrangling, and tool integration mechanics rather than focusing on agent logic and value.

Workflow diagram · grounded in source
1
User request received
trigger
“When a request is received (e.g., via the Cloud Run endpoint), the runtime identifies the target agent.”
2
Root agent routes request
routing
“It analyzes the user's intent and intelligently delegates the task to the most appropriate specialized agent using the sub_agents configuration.”
3
LLM decides and ADK executes
ai_action
“When the LLM decides to use a tool or delegate to another agent, ADK handles the execution via the implementation references provided in the configuration.”
4
Sub-agent executes with tools
ai_action
“The datetime agent demonstrates how to extend an agent with external tools tools. The agent can access the current date and time defined in the tools.py of the repository”
5
Results returned to root agent
output
“Return the prime number results to the root agent.”
Reported outcome

Ackgent, built on Google ADK Agent Config, streamlines the agent development lifecycle, reduces boilerplate, and makes testing and deployment significantly faster, offering an immediate boost in productivity for developers building and deploying AI agents.

Reported metrics
Developer productivityimmediate boost in productivity
Testing and deployment speedsignificantly faster
Reported stack
Google ADKAckgentAgent Configgemini-2.5-flashGoogle Cloud RunMCPmarkitdown-mcpuv
Source
https://mlops.community/blog/ackgent-rapid-agent-development-on-gcp-with-adk-and-agent-config
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Ackgent, built on Google ADK Agent Config, streamlines the agent development lifecycle, reduces boilerplate, and makes testing and deployment significantly faster, offering an immediate boost in productivity for devel…

What tools did this team use?

Google ADK, Ackgent, Agent Config, gemini-2.5-flash, Google Cloud Run, MCP, markitdown-mcp, uv.

What results were reported?

Developer productivity: immediate boost in productivity; Testing and deployment speed: significantly faster (source-reported, not independently verified).

How is this workflow AI workflow structured?

User request received → Root agent routes request → LLM decides and ADK executes → Sub-agent executes with tools → Results returned to root agent.