Building CouplePay with AI Agents: Day 1
How I'm using a team of AI agents to build a credit card payoff app for couples. Meet Diger, the Chief of Staff, and the specialist team.
Today I started updating CouplePay — a simple credit card payoff app designed for couples. But I’m not building it alone. I’m building it with a team of AI agents.
The Idea
Couples deal with debt together, but most debt payoff apps are designed for individuals. CouplePay is different:
- Manual entry — no scary bank connections
- Couples-first — see both partners’ debts in one place
- Calm tone — less shame, more control
- Simple — just credit cards, not all debt types
The Agent Team
Instead of a traditional dev team, I set up a hub-and-spoke AI agent system:
Diger 🦖 — Chief of Staff (GPT-5.4) The hub. Coordinates everything, routes tasks, makes decisions.
Specialists:
- Content Lead 🧠 — Brand, messaging, social
- Dev Lead 🛠️ — Engineering, code quality
- Growth Lead 📈 — Acquisition, opportunities
- Research Lead 🧠 — Market research, competitor analysis
- Bubba Chuck 🍌 — Operations, process health
Each specialist has their own Slack channel. Diger delegates work via agent-to-agent calls, then posts summaries for visibility.
How It Works
When I ask Diger to do something that requires specialist expertise:
- Diger adds a ⏳ reaction to my message
- Logs ”🔄 Started” to the status channel
- Calls the specialist agent directly
- Gets the result
- Posts a summary to the specialist’s channel (paper trail)
- Logs ”✅ Complete” with timing
- Swaps ⏳ for ✅
- Replies to me with findings
All autonomous. No human in the loop for routine delegation.
What’s Next
Tomorrow I’ll share how we set up the workspace structure, heartbeat monitoring, and the lessons learned from getting the routing working.
Building in public with AI agents. Let’s see how this goes.
This is part of a series documenting how I’m building CouplePay with AI agents. Follow along on Twitter.