Definition
Fit between AI-recommended architectures and the team's actual scale, operational maturity, and staffing, measuring the systematic bias toward large-organization patterns absorbed from public training data.
The Measurement Gap
Measurement Gap and Decision
The Gap: AI architecture assistants trained on public patterns absorb a systematic bias toward large-scale solutions. They recommend microservices with event sourcing, CQRS, and service mesh for a four-person team building an internal tool for 200 users. The architecture is technically sound at Netflix scale but exceeds this team's operational capacity. Traditional architecture evaluation focuses on technical properties -- does it scale, is it maintainable, does it follow best practices -- but nobody asks: can this specific team, with their specific skills and infrastructure, actually run this?
Architecture fitness is not architecture correctness. A correct architecture the team cannot operate is worse than a simpler architecture they can manage and evolve. Teams rarely override AI recommendations because the AI's authority suppresses their operational self-awareness.
The Decision:
ATFS measures the degree to which an AI-recommended architecture matches the team's actual scale, skill profile, operational maturity, and infrastructure capacity. When ATFS falls below threshold, the intervention is simplifying the architecture or rejecting the recommendation -- not adopting it and hoping the team grows into it.
Formula
Formula:
ATFS = (1/5) x (Skill_match + Ops_complexity + Scale_match + Infra_match + Evolution_match)
Where each dimension is scored 1-5:
- Skill_match: Does the team have skills to build and operate this? (1 = major gaps, 6+ months learning; 5 = direct experience with all components)
- Ops_complexity: Is operational burden proportional to capacity? (1 = ops consumes >50% of team time; 5 = minimal burden)
- Scale_match: Is architecture right-sized for actual load? (1 = grossly over/under-engineered; 5 = right-sized for current and 18-month projection)
- Infra_match: Does existing infrastructure support this? (1 = requires unbudgeted platforms; 5 = runs on existing infrastructure)
- Evolution_match: Can the team evolve this as requirements change? (1 = rigid, changes require restructuring; 5 = clear extension points)
Worked Example:
AI recommends event-sourced microservices with service mesh for a 4-person team. Assessment:
- Skill_match: 2 (team has never operated service mesh or event store)
- Ops_complexity: 1.5 (operational overhead would consume >50% of time)
- Scale_match: 2 (grossly over-engineered for 200 users)
- Infra_match: 2 (requires Kubernetes cluster team does not have)
- Evolution_match: 3 (microservices allow independent scaling)
ATFS = (1/5) x (2 + 1.5 + 2 + 2 + 3) = 2.1
Proposed formula, pending validation pilots.