Selected research
Notes on
Enterprise AI
in Practice
Enterprise AI
in Practice

Enterprise AI Research

A framework to test in real work
Four ideas drawn from Andrew David W.Y.’s research on enterprise AI operating capability. Open a research note below for the article and its source. Card imagery is illustrative.

Framework at a Glance

  • Value defines the result needed. That result determines what the process must do.
  • At each handoff, check separately whether the upstream deliverable is acceptable and whether the next step can begin.
  • Record execution, acceptance, downstream use, and progress against the objective separately.
  • Ask someone else to repeat the work to test whether the method can be used independently.
  • The four scales describe the same result at different levels and constrain one another.
  • People and AI share the same acceptance standard. Accountability stays with the responsible person.

Research notes

01 · Research series · 01/05

How the four scales keep each other aligned

Operating outcomes, business workflows, job responsibilities, and specific actions are not four independent checklists. They are the same work unfolded at different levels.

Read the full article →
02 · Research series · 02/05

Why every handoff needs two judgments

At every handoff, ask two questions separately:

Read the full article →
03 · Research series · 03/05

Four kinds of evidence show whether capability has formed

Evidence is not a record added after the work. It is the basis for judging whether the work can be done well again next time.

Read the full article →
04 · Research series · 04/05

Why people and AI need one acceptance standard

Acceptance standards judge whether the result is acceptable, not who made it.

Read the full article →
05 · Research series · 05/05

How a value-stream map clarifies a business workflow

A value-stream map draws a business workflow from beginning to end: the final outcome to deliver, the processes along the way, what each step hands over, and the two judgments made at each handoff.

Read the full article →
06 · Research notes

Enterprise Governance: From Authority Controls to Verifiable Accountability

Governance becomes verifiable when responsibility is tied to observable delivery and handoffs, not merely completed approvals. Clear acceptance standards and separate handoff decisions make it possible to locate problems through evidence.

Read the full article →
07 · Research notes

Outcome Accountability and Lasting Capability in AI Marketing

Faster generation does not, by itself, create reusable marketing capability. Clear outcome standards, evidence of downstream use and independent repeat execution are needed to connect AI-assisted work with observable business results.

Read the full article →
08 · Research notes

Verifiable Outcomes in Blockchain Applications

Tamper-resistant records do not automatically produce useful business results. Blockchain traceability becomes operationally meaningful when the recorded states support clear acceptance and handoff decisions.

Read the full article →
09 · Research notes

Choosing a Pilot's Scope: Why Smaller Is Often More Effective

A bounded workflow makes it easier to establish a baseline, locate problems and compare results. Starting with the smallest testable unit provides a sounder basis for deciding whether to expand.

Read the full article →
10 · Research notes

From Tool Efficiency to Operating Capability: What Is Usually Missing?

Faster tasks do not necessarily lead to stable improvements in business performance. Clear outcomes, separate handoff decisions, common standards and independent repeat execution connect local efficiency gains with reusable organizational capability.

Read the full article →
11 · Research notes

Intelligent Organizational Capability: From Individual Proficiency to Transferable Capability

Reliable organizational capability requires effective practices to work beyond the individuals who first mastered them. Aligned process definitions, separate handoff judgments and independent repeat execution provide evidence that a method can be transferred and retained.

Read the full article →
12 · Research notes

Organizational Knowledge Management Systems: From Accumulated Documents to Reusable Capability

A knowledge management system should help people deliver acceptable results, not merely store and retrieve documents. Anchoring knowledge in shared process definitions and testing its reuse makes it possible to maintain transferable operating capability.

Read the full article →
13 · Comparative analysis

Comparative Analysis: Efficiency Gains and Lasting Capability

Faster execution in one stage is not the same as a method that others can reliably reuse. Acceptance results, downstream uptake and independent repeat execution help distinguish temporary efficiency gains from lasting organizational capability.

Read the full article →
14 · Field observations

Observations from Practice: Review Takes Longer After Tools Go Live

Faster generation can coexist with longer review when acceptance standards and handoff decisions remain unclear. A bounded comparison of review time and rework can reveal whether the bottleneck lies in definitions and handoffs rather than tool speed.

Read the full article →
15 · Educational example · No investment advice

Evaluating Evidence Under Uncertainty: A Digital Asset Example

An educational Bitcoin example shows how to define a deliverable, separate testable conditions and check whether someone else can reproduce a judgment. The scenarios are hypothetical and do not constitute investment advice or a price forecast.

Read the full article →

Four Research Themes

The V2.0 framework is an interim research summary. Its practical effects need testing in business pilots.

Stakeholder Value

Identify who needs the result and what would make it useful to them.

Four Scales

Trace the same result across operating outcomes, business processes, functional responsibilities, and individual actions.

Two Handoff Checks

Check separately whether the upstream deliverable is acceptable and whether the next step is ready to start.

Evidence and Reuse

Record the four kinds of evidence separately, then ask someone else to repeat the work independently.

AI SKILL

Turn methods into reusable rules

Explore how AI Skills relate to manuals and shared acceptance standards. Read trial examples for defining results, checking handoffs and reviewing repeatability.

Explore AI Skills →
A useful workflow has a result standard that people can apply whether a person, AI or a combination did the work.
Workflow design principle