Demo · illustrative

What is safe to hand to AI—and what stays human?

The Daily Work Audit turns one real week into a task-by-task answer: automate now, revisit later, or keep human on purpose.

The Human Agency Scale (H1–H5) was introduced by Stanford's SALT Lab ("Future of Work with AI Agents," 2025). The HAS Audit — task inventory, scoring, the three scores, the action list — is the applied framework taught at AI Start Academy. The composite founder below is illustrative.

The framework

How much human involvement do you want?

H1 · Full automation

AI runs it end to end.

H2 · Human reviews

AI produces; a person approves.

H3 · Equal partner

AI and human iterate.

H4 · Human leads

AI assists the person.

H5 · Human only

The human is the point.

Desired involvement × current AI capability

Green light

You want it automated and AI is capable now.

R&D opportunity

You want leverage, but capability needs work.

Keep human

AI could help, but the human contribution is deliberate.

Low priority

AI is not capable and automation is not valuable.

Fictional composite

Daniel Reyes runs two businesses on a team of six.

CredentialCo is a certification product that other trainers license. Ascend Academy sells live cohort courses. Daniel first reports eleven recurring tasks totalling 51 hours a week.

The calibration catch: “Running the cohort” occupies 34 of 51 hours—two-thirds of the week in one vague entry. It is too coarse to score. Splitting it reveals the real opportunity.

The 34-hour line, split

Cohort taskHoursDesired HASAI capabilityZone
Live teaching and delivery14H52Keep human
Student Q&A between sessions7H24Green light
Grading and feedback7H33R&D
Onboarding, reminders and scheduling4H14Green light
Cohort announcements2H24Green light

The three scores

A practical view of leverage and judgment.

ARS · 33%

A third of the week is ready to hand off today—review, Q&A, logistics and communications, not teaching or selling.

FLS · 30 hours

Green-light and R&D work combined: the total that may be automatable now or soon.

HMS · 41%

The share that remains human-led on purpose, just above the illustrative 20–40% reference range.

This week’s action list

Three bounded actions—not “automate everything.”

TaskApproachDeployIllustrative time reclaimed
Student Q&AAI uses the course FAQ and escalates anything outside it.Within 7 daysAbout 5 hours
Onboarding and schedulingAn approved welcome, calendar and reminder sequence.Within 7 daysAbout 3.5 hours
Cohort announcementsAI drafts from cohort information; Daniel reviews before sending.Within 7 daysAbout 1.5 hours

The figures are part of the fictional composite, not measured client results.

The takeaway

The opportunity was hiding inside one vague calendar line.

The answer is not “automate teaching.” It is to preserve the work where the founder is the product, while removing the logistics, review and communications buried inside it.

This is what a Business Opportunity Audit produces for a founder’s own week: a task-by-task answer to what is safe to hand off and what is not.

View the Business Opportunity Audit on the website →