The 40-Point Perception Gap #
- Developer Productivity Benchmarks 2026 (Larridin) — METR randomised controlled trial with experienced open-source developers: 19% slower with AI tools despite feeling 20% faster. A 40-point gap between perceived and actual productivity. The finding is particularly significant for legacy modernisation contexts, where developers are most likely to feel confident using AI on unfamiliar code — precisely when the gap would be largest and most costly.
- Enterprise AI Coding Tools ROI: 2026 Case Studies & Metrics (Exceeds.ai) — Healthy ROI is 2.5-3.5× (average) and 4-6× (top quartile), but only when the cost denominator includes actual token and usage-based costs. Safe AI code acceptance rate: 25-45%; teams above 45% need to audit whether accepted suggestions survive code review and production deployment without disproportionate rework.
Code Abundance: Enterprise Governance Gap #
- The 2026 State of Code Abundance Report (CloudBees) — 81% of enterprise leaders report production issues tied to AI-generated code; AI now generates or assists 61% of the average enterprise codebase; most organisations still lack visibility, governance, or attribution tools. The “code abundance” framing names what the governance gap actually looks like: not a shortage of code but an excess of code no one can properly account for.
Legacy Modernisation: Senior Oversight as the Pattern #
- Vibe coding goes enterprise: What you need to know about AI-driven legacy modernization (CIO) — AI systems can now read legacy code, extract business rules, and generate verified modern replacements at enterprise scale. The integration pattern that works: senior engineers define architecture + delivery framework + guardrails before AI acceleration begins. Vibe coding tools are fast but ungoverned; enterprise operations need agentic AI with staging, risk assessment, and IT approval built in.
- Vibe Coding for Enterprise: Why Governance Matters (2026) (Opsima) — Governance framework for enterprise vibe coding: vibe coding without governance = shadow IT at scale. Required elements: architecture review gates, AI-output acceptance criteria, comprehension verification checkpoints.
Cross-links #
- Vibe Coding Approaches The 40-point perception gap (METR) applies directly here — enterprise legacy projects are where overconfident AI use is most dangerous.
- Team & Org Use of Claude Code Abundance Report (81% production issues) is the team-governance story for this same finding.
- [trust-overextension-quest] The perception gap + code abundance combination is the most direct enterprise evidence yet for the trust-overextension thesis.
- Emerging pattern: The METR 40-point perception gap and the CloudBees code abundance report arrived in the same cycle and tell the same story from different angles — developers feel faster, organisations are generating more code, and neither group is measuring understanding or production quality accurately.
- Quality signal: METR randomised controlled trial is the most methodologically rigorous finding in the enterprise vibe coding space — not a survey, an actual experiment.
- Keyword suggestion:
"code abundance" enterprise AI 2026 governance — CloudBees coined this; will propagate.