Research

Original research into how people put AI to work.

Eldris studies why access to AI so often stops short of use, in organizations and in people's own lives. Commission a study designed around your question, or license reports for use inside your organization.

Research programs

Two programs, one question.

Why does access to AI so often stop short of use? Each program holds a series of bounded studies, not one study of everything.

Enterprise AI Adoption Gap

People inside organizations, deciding whether to use AI for their work. Their choices are shaped by roles, incentives, accountability, and the need to coordinate with others, and by what the organization rewards or punishes.

Illustrative question, not a finalized study

What keeps an employee who has access to AI from using it for a particular work task?

Consumer AI Adoption Gap

Individuals deciding whether AI is worth using for their own tasks. Their choices depend on personal circumstances and on the task itself, which differ from one person, and one task, to the next.

Illustrative question, not a finalized study

When does a consumer try AI once but decline to use it again for the same task?

How a study is framed

Every study starts with a human decision.

In both programs, someone decides whether to use AI. What differs is the context around that decision, so each study pins down four things before any fieldwork begins.

01

The decision

The specific choice being studied: to try AI, to keep using it, to trust its output, or to hand it a task.

02

The population

Who is making that choice, described precisely enough that the findings say who they apply to and who they do not.

03

The task

The work in question. People use AI for one task and refuse it for the next, so a study about "AI in general" answers very little.

04

The boundary

What the study covers and what it leaves out, stated up front, so nobody stretches the result further than the evidence goes.

Commissioned research

A study designed around your question.

An organization that needs evidence about how its people, or its customers, come to use AI can commission a study. The work is scoped together and agreed in writing before it begins.

  • Your question Framed with you as a decision, a population, a task, and a boundary.
  • An agreed design and analysis How the study will gather evidence and how that evidence will be analyzed, settled before fieldwork.
  • A report Findings, methods, and limitations, written for the people who will act on them.
  • A readout or workshop An executive session to walk through what the study found and what it means for the decision.
  • An optional public report Only if both sides agree to publish.

Rights, exclusivity, publication, and how research independence is protected are scoped case by case for each commission. Nothing on this page sets standard terms.

Reports

Free to read, or licensed for internal use.

Each Eldris report is either free to download or licensed for use inside your organization, for teams that want the evidence without commissioning a study of their own. The scope and terms of a license are agreed before the report is delivered.

If you need evidence on a topic we have not covered, tell us what it is and who would read it. It helps decide what we study next.

Free report · By Dr. Jeffrey Allan · October 3, 2026

Consumer AI Trust Is Task Specific

An executive guide to deciding what consumers will let an assistant do across assistance, transaction approval, and standing authority.

Free report · By Dr. Jeffrey Allan · October 3, 2026

From AI Time Savings to Business Value

An executive guide to deciding what an AI productivity gain is worth across task speed, operational integration, and verified economic return.

Free report · By Dr. Jeffrey Allan · October 3, 2026

The Employee Side of AI Adoption

An executive guide to choosing the support, rules and workflow changes that make AI useful at work.

Free report · By Dr. Jeffrey Allan · October 3, 2026

What Enterprise AI Adoption Numbers Actually Mean

An executive guide to choosing the measures that matter across access, use, operational integration, sustained reliance, and realized business value.

Free report · By Dr. Jeffrey Allan · October 3, 2026

Why AI Adoption Benchmarks Travel Badly

An executive guide to deciding which comparisons belong in a market plan, investment case or rollout target.

Methods and quality

Stated for each study, not promised in advance.

Each study's design will specify how participants are sampled, what is measured, how the analysis is done, what the limitations are, and what review the work receives. Studies use original fieldwork with recruited participants, sampled to match the population each study defines.

Whether a particular study is preregistered, externally reviewed, or submitted for peer-reviewed publication will be stated for that study, as it actually happens. We will not describe a study as validated, published, or peer reviewed until it is.

A finding is only as useful as the honesty about its limits.

Research and commentary

Commentary is not research.

Eldris also publishes commentary: decision briefs, policy notes, and observations from building software. That writing is argument and experience. It is not research evidence, and it is never sold as a report.

Have a question worth studying?

Tell us the decision you need to understand and who makes it. A short description is enough to start.