AI-assisted practice

More capacity.
Accountable judgment.

I connect specialist AI roles, evidence, and review to make complex work more manageable—and the decisions behind it more visible.

My contribution is the operating model: framing the problem, choosing the right perspectives, challenging the work, and deciding what earns acceptance.

How the practice works

Specialists contribute. I own the decisions.

  1. 01

    Frame the work

    Research + product

    Find the evidence, the uncertainty, and the decision worth making.

  2. 02

    Shape and build

    Design + content + engineering

    Develop alternatives and make an approved direction reviewable.

  3. 03

    Challenge the result

    Quality + specialist review

    Expose gaps, correct the work, and check it again.

  4. 04

    Decide and learn

    Human acceptance + measurement

    Accept, revise, or reject. Follow the effort and the outcome.

The program dashboard

Work, quality,
and value in view.

What needs attention? What has earned acceptance? Where is effort going? The dashboard connects those questions to the evidence.

UX practice

The review desk

Public demonstration

Three illustrative audit records. One transparent value scenario.

Measured ROI: pending
01 / Delivery3 awaiting a decision

See what is pending, who needs to decide, and what would make it ready.

Inspect the dashboard →
02 / Quality3 review examples

Follow a finding through its proposed correction, human disposition, and recheck.

Explore audit records →
03 / Economics5 modeled capacity hours

Compare a manual baseline with all human effort and the costs of using AI.

Explore the ROI model →

Example records and synthetic inputs. This demonstration explains the method; it does not report operational results.

ROI / Make the assumptions visible

Count the review.
Keep the losses.

Faster output is only part of the story. Useful value depends on comparable scope, accepted work, complete human effort, quality, and cost.

Change the assumptions

Illustrative scenario / One accepted work package

Manual baseline
10 hours
Production with AI
3 hours
Human setup, review, and rework
2 hours
Estimated capacity difference
5 hours

$725net estimated capacity value

At $150/hour, less $25 in tool costs and $0 other allocated costs. A modeled scenario, not cash savings or Rob’s measured return.

Seven specialist perspectives

The right role.
At the right moment.

Each role has a purpose, defined inputs, a reviewable output, and a human checkpoint. I select the perspectives the work needs.

Personified AI advisory roles, represented by AI-generated portraits. They are not employees; their work remains subject to human judgment.

Explore all seven profiles →

A worked example / Illustrative

Follow the work.
Find the decision.

A team wants to clarify a confusing service. Here is how an idea becomes a proposal that can be challenged and tested.

The brief

Two destinations appear to do the same job.

Which one should a person choose to manage a record? A plausible label change needs more than a confident explanation.

Sanitized demonstration
  1. 01 / Evidence

    Riley separates the finding from the hypothesis.

    The destinations overlap in a prototype inspection. That is a review observation; whether people are confused still needs research.

  2. 02 / Proposal

    Dana and Casey make an alternative concrete.

    Propose one task-focused destination, keep account settings distinct, and explain what each action changes.

  3. 03 / Challenge

    Quinn checks what the proposal could miss.

    Can someone acting for another person still find the right task? Are the access boundaries and keyboard path clear?

  4. 04 / Human decision

    I decide what is ready to test.

    With Piper’s criteria, Ethan can prototype the chosen direction. I review the result and decide whether to revise it or seek more evidence.

Acceptance needs evidence.

A proposed next check: ask relevant people to find the correct destination and explain what they can change. Record errors and assistance before claiming an improvement.

Explore the research-to-design method →

The connected case study

See how the pieces
work together.

Explore the audience models, role handoffs, governance, audit records, and value measurement behind the practice.

Explore the operating model
Evidence and current status

The role structure and measurement method are documented. The public dashboard and value scenario use demonstration data. Acceptance-based instrumentation and the measured ROI pilot remain pending. The worked example is a sanitized illustration, not a user-research finding. Agent review supports human judgment and does not replace research with people.

Rob’s AI Guide

Explore Rob’s work and ideas

AI guide · Introduction

Ask about Rob’s experience, approach, and ideas. Answers draw on his published work, with supporting pages and related reading or listening shown separately.

The AI guide is not connected yet. The suggested questions open portfolio searches.

Follow-up questions use the last two exchanges. The conversation clears when you leave or reload this page.