E.I.

From A.I. to E.I.

Nine questions to ask before anything gets built: an A.I. system, a product, a business. Open source under the MIT License.

Why E.I.

A.I. makes almost anything cheaper to build. It does not say whether the thing should exist, or who pays when it goes wrong.

E.I. asks both halves together. Empathy asks whose day it changes. Innovation asks whether it can be built. The other seven questions sit around those two.

From A.I. to E.I.

The nine questions

Ethical, Empathy and Ecosystem run down the side. Impact, Innovation and Investment run across the top. Each question has a paper with measures and three public cases.

E × I
ImpactDoes it work, and for whom?
InnovationCan it be built?
InvestmentOn whose clock, and who owns it at the end?
EthicalWhat will we not do to get there, and can an outsider check what we claim?
Can you prove it worked?

Measured, not asserted

A claim arrives with a measure, a baseline, and a date it was taken.

01 · Purpose · Read the paper →
Is there a line you will not cross for money?

Hard stops

A written list of harms disqualifies a venture, and nothing offsets them.

02 · Purpose · Read the paper →
At the end, who owns the asset?

Ownership at the end

The operator ends up holding the asset, not the payment schedule.

03 · Patient Capital · Read the paper →
EmpathyWhose day does this change, and were they in the room when it was designed?
Can you name one real person this helps?

The first beneficiary has a name

One person's day, written down, before anything gets built.

04 · Purpose · Read the paper →
Were the people it affects in the room?

Built with, in the room

Design with them, not for them.

05 · Playbook · Read the paper →
How long will this take, and does the money agree?

The problem sets the clock

The horizon follows the problem, which makes patient money a term structure.

06 · Patient Capital · Read the paper →
EcosystemDoes it still work after we leave, and who holds the asset at the end?
Who pays the costs that never reach your P&L?

Measured at the wide boundary

Count where the labor, land, capital and care sit.

07 · Purpose · Read the paper →
Does it still work when you walk away?

The Minimum Viable Ecosystem

The smallest set of parts that keeps working after we leave.

08 · Playbook · Read the paper →
Do you own what the business runs on?

Own the ground

We buy the building, because long-horizon money has to rest on an asset.

09 · Patient Capital · Read the paper →

Marking

Green means we have the evidence and it carries a date. Yellow means someone owns the gap and there is a date to close it. Red stops the work until it clears. Early work sits at yellow for a long while, and we mark it that way.

Green: evidenced and dated
Yellow: a work order with an owner and a date
Red: absent or contradicted

Guardrails for A.I., people and business

The aim is one framework that holds in all three places. Version one was written for ventures. The next versions get written in the open.

A.I.

Systems

The nine questions, asked of a model or an agent before it ships and each time it changes.

People

Teams

A review a team can run in an hour, with a named owner and a date against every open answer.

Business

Companies

The answers written into the business model, next to revenue and cost, where decisions get made.

Use it

E.I. is released under the MIT License. Use it inside your own product, change the questions, publish your version. Keep the copyright notice.

Data
ei-framework.json

All nine questions, papers, measures and cases, machine-readable.

To propose a change, write to contact@cei.inc with the question, what you would change, and why.

Changelog

October 2026

Released under the MIT License, as its own site.

September 2026

Empathy moved to the middle row and Innovation to the middle column. The question where they meet now sits at the center: were the people it affects in the room?

September 2026

Two scores, one for ethics and one for viability, replaced by nine questions answered one line each.