The hidden cost of moving fast with AI tools without resolving upstream ambiguity.
Technical debt is well understood. Decision debt is not — and it's the dominant failure mode in AI-assisted product development.
Decision debt accumulates when teams defer upstream decisions (what to build, for whom, and why) while accelerating downstream execution (writing code, shipping features). AI coding tools make this worse by removing the natural friction that used to force decision-making.
Traditional workflow: Decide → Struggle to build → Ship
AI-assisted workflow: Vaguely decide → Instantly build → Ship the wrong thing faster
^^^^^^^^^^^
Decision debt accrues here
The core thesis: Execution speed is now commoditized. The bottleneck has shifted from "can we build it?" to "should we build it?" — and most teams haven't updated their processes for this shift.
| Category | Description | Example |
|---|---|---|
| User Debt | Building without validated user need | "Users probably want this" → ship → no adoption |
| Scope Debt | Undefined boundaries compound with AI speed | "Make it smart" → AI generates 10x scope → paralysis |
| Priority Debt | Everything feels buildable, so nothing gets ranked | 50 features in backlog, all "possible this sprint" |
| Measurement Debt | No success criteria defined before building | "We'll know it when we see it" → endless iteration |
| Integration Debt | Components built independently without system design | 5 AI-generated microservices that don't talk to each other |
A simple diagnostic for teams to assess their current decision debt load:
Score = (Ambiguity Level × Execution Speed) / Decision Clarity
Where:
- Ambiguity Level (1-5): How well-defined are the requirements?
- Execution Speed (1-5): How fast are you shipping?
- Decision Clarity (1-5): How explicit are your what/who/why decisions?
Interpretation:
- Score < 2: Low debt — you're deciding before building
- Score 2-5: Moderate debt — slow down on execution, speed up on decisions
- Score > 5: High debt — stop building, start deciding
# Run the interactive assessment
python tool/assess.py
# Run with a specific project context
python tool/assess.py --project "Our new AI-powered onboarding flow"tool/assess.py is a CLI tool that walks you through the Decision Debt diagnostic. It asks structured questions about your current project and produces:
- A Decision Debt Score with category breakdown
- Specific debt items identified in your project
- A paydown plan — concrete decisions to make before writing more code
This framework emerged from building AI products at Tesla, Amazon, Udemy, and now at my current startup. The pattern was consistent: the teams that shipped the best products weren't the fastest builders — they were the clearest thinkers.
I write about this in my LinkedIn series on decision debt and vibe coding failure modes.
- Agent Patterns — Constrained multi-agent systems for better AI decision-making
- The "decision debt" concept connects to Ward Cunningham's original technical debt metaphor — but applied to product decisions rather than code quality
MIT