I’ve got this vision for an exciting future where AI agents collaborate seamlessly, creating a dynamic workflow in development.

Imagine This Scenario

You ask your Agent to pull a Jira Epic and all its User Stories directly into the chat. The Agent, equipped with deep knowledge of your project through its Memory Bank, analyzes the requirements. It retrieves and presents a comprehensive analysis, including the current state and previous revisions that led to architectural decisions.

Enter the AI Agent Architect

The Agent then communicates, via the Model Context Protocol (MCP), with the company’s AI Agent Architect. This Architect Agent considers the entire system architecture, data flow, infrastructures, databases, and message queues. It either approves the plan or provides feedback for adjustments. If more information is needed, it can consult another Repo Agent through MCP to refine the solution.

Sequence diagram

Agents Collaborating for Optimal Solutions

These Agents interact until they devise the most efficient solution for the system. If the change is local to the repo, the initiating Agent executes it, relying on a system that evaluates changes globally.

Unlocking Efficiency

This approach promises significant time savings, reduced errors, and smooth workflow between teams. Even if we stop at presenting the plan to humans, imagine the efficiency of starting discussions with a ready-to-go work plan, rather than ending meetings with action items for further research.

Looking Ahead

We can also explore scenarios involving simultaneous changes across multiple repos, version management, and how this integrates with DevOps MCP for managed deployments up to production. Why stop there? The Release Agent collaborates with the Product Agent to generate release notes sent directly to clients.

The future is wild! 

If terms like Memory Bank or MCP are new to you, check out the links to learn more and enjoy!

MCP: https://www.anthropic.com/news/model-context-protocol
Memory Bank: https://cline.bot/blog/memory-bank-how-to-make-cline-an-ai-agent-that-never-forgets

How do you envision the future?