Research agenda

There is no shortage of evidence that AI can make tasks faster. The more difficult question is how organizations turn those gains into durable changes in value, accountability and the way work gets done.

This page describes a proposed research agenda. It does not present results of a completed original study, interviews or a published dataset.

Questions worth examining

1. Where does AI-enabled individual productivity convert into end-to-end operating improvement, and where does the benefit stall?
2. Which ownership and financial-governance arrangements make outcomes easier to verify?
3. How do enterprises assign authority, access and accountability when agents act across functional boundaries?
4. What mechanisms support human trust, auditability and safe scaling?

Proposed method

The intended approach combines a structured review of public research, practitioner interviews conducted with informed consent and a transparent synthesis of themes. Where possible, examples will be checked against documented measures rather than relying on headline performance claims.

Interviews would be anonymized unless contributors explicitly agree to be named. Any future report would distinguish participant views from measured outcomes, explain sample limitations and separate correlation from causation. No employer-confidential operational details or client data will be used.

Open invitation to critique

I've published a short briefing for independent reviewers of the proposed four-contract operating framework for AI agents. It explains the relationship to prior research, the model's current limits and five focused questions where criticism would be especially useful. No external endorsement or validation is claimed.

How to contribute

If you lead enterprise AI, transformation, governance or value realization and would like to exchange views on these research questions, please get in touch.

Background reading

• McKinsey: The state of AI in 2026.
• NIST AI Risk Management Framework.