Most AI adoption in SDLC improves local productivity.
Enterprise delivery fails for a different reason.
It violates the 3C model: Context, Constraints, and Collaboration.
Consider a modernization example:
Context:
Eligibility and approval rules existed across legacy code, PDFs, and tribal knowledge.
AI-assisted discovery extracted them accurately.
Constraints:
Architects made a deliberate decision to externalize these rules for regulatory change.
That constraint existed only in documents.
Collaboration:
Developers and QA used AI independently to generate services and tests.
Six sprints later:
• Services re-embedded the rules
• Tests passed because behavior looked correct
• Audit questions failed because decision rationale was lost
The AI wasn’t wrong.
The SDLC system was incomplete.
What worked was enforcing 3C explicitly:
• Context persisted as shared, versioned SDLC memory
• Constraints became executable guardrails, not guidance
• Collaboration happened on a single execution surface shared by humans and agents
Enterprise AI succeeds only when all three Cs are present.
Miss one — and AI accelerates failure.
Top comments (0)