Where are you on the measurement curve?

Four questions, one per level of the measurement ladder. Answer honestly — the point is the gap, not the grade.

01 Usage Do you know what AI actually costs you? No consolidated view of tokens or spend Aggregate spend only — not who or what drives it Tokens and spend attributed per engineer, monthly

02 Quality Do you know whether agent-written code holds up? Agent and human code aren't separated at all Agent PRs get reviewed, but nothing is tracked Acceptance, revert and regression rates, agent vs. human

03 Productivity Can you weight output by how hard it was? We count PRs, commits or story points, unweighted Work is tiered informally; nothing is measured against it Complexity-weighted throughput or one-shot rate per tier

04 Benchmarking Do you know where you stand against peers? No external comparison Anecdotes, conference talks and vendor claims Benchmarked on AI-attributed share, effective FTEs, spend

Your measurement level

Level —

—

Discuss your result →

Levels follow the four-level measurement framework in The Agentic Awakening (Bessemer Venture Partners, 2026).

AVC Fund AI VALUE CREATION