🌍 Follow-the-Sun Development Experiment: Building Decision Copilot MCP with AI Agents

[mrdavidlaing]

What if we could develop software 24/7 using human mentors guiding AI agents across time zones?

I'm launching an experimental development approach where 2-3 human mentors in different time zones each supervise AI agents to implement the a medium complexity software project - Decision Copilot MCP server - in continuous shifts that follow the sun around the globe.

The Experiment

Each mentor works a ~8 hour shift in their local timezone, acting as the decision maker and guide while AI agents serve as implementers. Here's how it works:

Shift Cycle

1. Start of shift: Receive handoff from previous mentor, review overnight progress
2. During shift: Set goals for AI agents, check in every 1-2 hours, help agents get unstuck
3. End of shift: Commit work, document progress, brief next mentor in 1-2 hour overlap window
4. Repeat: Next mentor takes the baton for their timezone

Example: A Day in the Life

EMEA mentor completes stakeholder management UI components and hands off to US West Coast mentor. US West Coast mentor reviews the code, identifies integration issues with the backend API, guides AI agents to implement the missing Firebase queries, and passes clean working code to Asia-Pacific mentor who focuses guiding the AI agents to implement the next feature in the plan. At the end of their shift they handover to the EMEA mentor... and so the cycle continues

What You'd Do

• Guide AI agents through feature implementation (not coding yourself)
• Make architectural decisions and provide direction
• Troubleshoot when agents get blocked
• Maintain code quality and consistency
• Participate in brief handoff meetings with other mentors

AI Tools We'll Use

• An Agentic coding tool - eg, Claude Code, Cursor, Windsurf etc
• Github
• GCP Firebase

Time Commitment

• Core shift: ~6 hours of periodic check-ins (not continuous work)
• Handoff meetings: 30-60 minutes at shift boundaries (perhaps just recording and watching a video?)
• Flexibility: Work around your existing schedule within your timezone

Ideal Mentor Profile

• Time zones: Looking for coverage across Americas, Europe/Middle East, Asia-Pacific
• Experience: Architectural experience building medium complexity applications
• Availability: Can commit to regular check-ins during your local work hours
• Curiosity: Excited about AI-assisted development and novel workflows

Success Metrics

We're measuring this experiment against traditional development:

• Velocity: Can we complete the MCP server faster than a co-located team?
• Quality: Does the code maintain consistency across shifts?
• Experience: Is this more engaging than traditional development?
• Learning: How effectively can humans direct AI agents on complex tasks?

Results and findings will be shared publicly as a case study.

Why This Could Be Groundbreaking

• Continuous development: Code evolving 24/7 instead of 8/5
• AI collaboration: Learn how to effectively direct AI agents on complex tasks
• Global teamwork: Experience seamless international collaboration
• Research value: Pioneer a potentially transformative development methodology

Ready to Join?

Comment below with:

1. Your general timezone (e.g., "US West Coast", "Central Europe", "East Asia")
2. Your preferred daily availability window
3. Brief note about your development background
4. What excites you most about this experiment

Let's see if we can create something amazing while pioneering the future of software development! 🚀

[jschomay]

Cool! I am interested.

1. US East
2. AM
3. I've done a thing or two with AI
4. Exploring techniques to move AI beyond call/response interactions and towards continuity, also more natural interactions and the required context for those affordances