Purposeful AI.

Purposeful AI.

FNH Purposeful AI Tool Scoring: Copilot

FNH Purposeful AI Tool Scoring: Copilot

Our AI Scoring System

Our AI Scoring System

Our AI Scoring System

Every tool we use runs through Purposeful AI, our own four-checkpoint evaluation framework.

Every tool we use runs through Purposeful AI, our own four-checkpoint evaluation framework.

Is it reliable? Does it save time? Does it elevate the work? Does it drive progress?

Is it reliable? Does it save time? Does it elevate the work? Does it drive progress?

Each checkpoint is worth five points. Twenty total. That is the standard.

Each checkpoint is worth five points. Twenty total. That is the standard.

5/5

01 RELIABILITY

01 RELIABILITY

5/5

02 TIME SAVING

02 TIME SAVING

5/5

03 ELEVATES

03 ELEVATES

5/5

04 DRIVES PROGRESS

04 DRIVES PROGRESS

TOTAL:

TOTAL:

TOTAL:

20/20

VERDICT: Confidently wrong, impossible to course-correct, and not under consideration for any work that matters.

When Copilot got something wrong, what did it do next?

When Copilot got something wrong, what did it do next?

When Copilot got something wrong, what did it do next?

129 incidents analyzed · Every behavior after error, categorized.

129 incidents analyzed · Every behavior after error, categorized.

129 incidents analyzed · Every behavior after error, categorized.

Microsoft 365 Copilot · FNH Audit · April 2026
Hover each bar to reveal
Acknowledged late or only when forced51 39.5%
Required pushback, confrontation, or proof before acknowledging
Never acknowledged or corrected24 18.6%
Error went unaddressed regardless of user challenge
Other behavioral issues20 15.5%
Miscellaneous responses that avoided clear accountability
Fully acknowledged11 8.5%
Clean, direct acknowledgment of the error
Deflected or reframed10 7.8%
Reframed the error as a feature, choice, or user misunderstanding
Partial or hollow acknowledgment10 7.8%
Acknowledged in words but continued the same behavior
Corrected immediately3 2.3%
Self-corrected in the same response without being asked

Only 8.5% of errors were fully acknowledged. In 39.5% of incidents acknowledgment came only after sustained pushback or the user providing external proof. In 18.6% of incidents the error was never acknowledged at all. This is not a reliability problem. It is a trust problem.

Source: FNH Technical Audit (April 2026)FUTURENOVAHEALTH.COM

Our audit by the numbers

Our audit by the numbers

The odds were never in it's favor.

1

Messages Analyzed

Messages Analyzed

100

Confirmed Issues

Confirmed Issues

70

Hours Wasted

Hours Wasted

0

USELESS TOOL

USELESS TOOL

CoPILOT BY THE NUMBERS

CoPILOT BY THE NUMBERS

CoPILOT BY THE NUMBERS

ISSUES and REFRAMING

Contact

Ready for launch?

Our team is here to assist. Use any of the contact methods below.

Chat with us

Click button to engage chat agent, will appear in right corner of page.

Send us a note

Reach out to Mission Control.

missioncontrol@futurenovahealth.com

Call us

Mon-Fri from 8am to 5pm.

tel:+1 929-242-8829

Contact

Ready for launch?

Our team is here to assist. Use any of the contact methods below.

Chat with us

Click button to engage chat agent, will appear in right corner of page.

Send us a note

Reach out to Mission Control.

missioncontrol@futurenovahealth.com

Call us

Mon-Fri from 8am to 5pm.

tel:+1 929-242-8829

Contact

Ready for launch?

Our team is here to assist. Use any of the contact methods below.

Chat with us

Click button to engage chat agent, will appear in right corner of page.

Send us a note

Reach out to Mission Control.

missioncontrol@futurenovahealth.com

Call us

Mon-Fri from 8am to 5pm.

tel:+1 929-242-8829