Methodology

Honest, defensible, provable

The point of r.Potential is to remove guesswork from AI investment. That only works if the method underneath is worth trusting.

01

Evidence over adoption

Shipping an AI tool is not the same as creating value with it. Every configuration is measured against outcomes — value created, work changed, confidence earned — not seat counts or usage curves.

02

Independence over sponsorship

Recommendations are governed independently. No vendor can pay for placement, and nothing rises in a portfolio except on evidence. When the board asks why, the reasoning is defensible and auditable.

03

Humans as primary actors

Configurations are combinations of humans and agents — designed to automate work without dehumanizing workers. Workforce impact is stated up front, never discovered after the fact.

04

Living rankings

Business conditions change and deployments succeed or fail. Rankings re-score continuously as deployment data flows back in, so what you see is what is true now.

What feeds the graph

Three classes of data, one coherent picture

Proprietary employment data

Decades of workforce data — how work is actually structured, staffed, and paid across industries.

Public market signals

Deployment announcements, vendor performance, labor market movement, and financial outcomes across 7M+ companies.

Your company data

Org structure, workflows, and systems — placing your organization precisely within the global picture.

From data to decision

01

Observe

The graph watches how AI meets work across the market.

02

Configure

Candidate human + agent configurations are generated for your org.

03

Score

Each gets value, workforce impact, and a confidence rating.

04

Re-rank

Deployment evidence flows back and the portfolio re-scores.

Adoption tells you what shipped.
r.Potential tells you what worked.

Request access