Companies Track Employee AI Use with Dashboards

Business Insider reported that companies including JPMorgan, Meta, and KPMG have built internal dashboards to track how employees use generative-AI tools, and independent coverage corroborates specific cases. KPMG has rolled out a dashboard that lets employees monitor their own usage, benchmark against peers, and track progress toward a 75% AI-usage target across its US advisory business, per People Matters and HRKatha. JPMorgan monitors how developers use tools such as GitHub Copilot and Claude and ranks engineers by usage, eFinancialCareers reported. Reporting indicates other firms, including Disney and Amazon, track AI activity as well.

eFinancialCareers documented "token-maxxing" (also rendered "tokenmaxxing"), where employees run up large AI-token bills, sometimes exceeding their own pay, to demonstrate adoption before cost controls are in place. This is a textbook case of an adoption metric, once tied to rewards or visibility, creating incentives to optimize the metric rather than the underlying productivity.

Adoption tracking typically captures telemetry such as API call counts, token usage, prompt lengths, and tool-embedding events. That telemetry becomes useful for analytics or model-training pipelines only after teams instrument logging and map records to downstream systems. Instrumentation that retains identifiable prompt text or user metadata creates materially larger privacy, IP, and compliance obligations than aggregated counters, especially where prompts contain client or employee data.

For data and platform teams, dashboard design encodes governance choices: what is logged, how long it is retained, whether it is anonymized, and whether it may train internal models. Those choices shape what telemetry can later support, from A/B testing to incident investigation, and they intersect with works-council and labor-policy constraints in some jurisdictions.

• •Whether firms move from raw usage counts to outcome measures such as error reduction, cycle-time impact, or quality sampling.
• •Published data-retention and anonymization policies as early indicators of governance maturity.
• •Regulatory or works-council pressure on prompt-level logging.