AI-Assisted Role Design and Hiring Audit
Design a role, test it against AI, then hire only for the remaining work.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 94%
This audit begins by defining the outcomes and responsibilities of a proposed role, then drafting a job description that a human reviews and personalizes. The finished description becomes the input to a second AI analysis. Give the model a specific role, a restricted list of approved tools, and instructions to break down every responsibility. For each item, ask how AI could perform the work, which prompts would be required, and which knowledge files or business context must be supplied. Test those recommendations with real tasks rather than accepting them theoretically. The result reveals which responsibilities can be automated, which can be AI-assisted, and which still need human ownership. Redesign the position around the remaining work, hire at the appropriate capacity, and use the final job description to generate and iteratively refine a 90-day onboarding plan.
Origin
Rick Mulready developed this process while hiring a new team member, using multiple AI models to draft the role, audit how much AI could perform, and create the onboarding plan.
Core principles
- 01Define the work before deciding whether a person or AI should perform it.
- 02Use explicit tool constraints to make an AI feasibility audit realistic.
- 03Break responsibilities down individually rather than evaluating the role as one block.
- 04Retain people for judgment, coordination, and quality assurance where they add the most value.
- 05Turn the final division of work into a concrete onboarding plan.
How to run it
- 1
Define the Required Outcomes
Write down what the business needs the role to accomplish and the responsibilities attached to those outcomes. Do not begin with a fashionable title.
Pro tip Separate recurring production work from judgment, communication, and ownership responsibilities.
Watch out An unclear role produces an equally unclear automation audit.
- 2
Draft and Edit the Job Description
Use AI to create a first draft from your notes, then correct, personalize, and complete it. The edited version becomes the authoritative specification.
Pro tip Treat the AI draft as a time-saving starting point rather than a finished hiring document.
Watch out Unreviewed job descriptions may omit critical expectations or misrepresent the role.
- 3
Set the Audit Parameters
Tell AI which tools are permitted and ask it to analyze every responsibility within those constraints. Request explicit prompts and required knowledge files.
Pro tip Compare several models when the role or proposed workflows are consequential.
Watch out An unconstrained audit may recommend tools, access, or capabilities the business cannot actually use.
- 4
Test Each Proposed Workflow
Run representative tasks through the suggested prompts and tools. Evaluate accuracy, consistency, review time, and operational risk.
Pro tip Test difficult edge cases, not only ideal examples.
Watch out A plausible written plan is not proof that the workflow works.
- 5
Redesign the Human Scope
Remove responsibilities AI can handle reliably and define where a person will direct, review, coordinate, or perform the remaining work. Adjust the role’s capacity and seniority accordingly.
Pro tip Preserve human quality assurance for customer-facing and strategic work.
Watch out Reducing headcount without assigning clear review ownership can create hidden quality failures.
- 6
Build the Onboarding Plan
Generate a 90-day plan from the revised job description, then adjust its timing to match real expectations. Specify what the hire should learn and deliver at each stage.
Pro tip Tell AI the milestone you expect by a specific day and let it re-sequence the plan around that constraint.
Watch out A generic 90-day schedule may move too quickly or too slowly for the actual role.
In the wild
Rick used AI to draft a job description, edited it, and then asked several models to explain how each responsibility could be completed using five specified tools. Gemini produced the strongest step-by-step breakdown. Rick still hired someone, but in a smaller capacity and with some of the AI-generated workflows incorporated into training.
→ The business retained human capability while reducing the amount of work the new hire needed to perform manually.
After revising the role, Rick asked AI to create a 90-day onboarding plan. He then iterated on the schedule, telling it when milestones were too slow or too fast and specifying where the hire needed to be by day seven.
→ The new team member received a more structured onboarding path aligned with Rick’s actual expectations.
Common mistakes
Automating Before Defining the Role
Without clear desired outcomes and responsibilities, the audit cannot distinguish useful automation from work the business does not need.
Accepting the Audit Without Testing
AI can produce a convincing workflow that fails on real inputs, edge cases, or required integrations. Every proposed responsibility needs practical validation.
Eliminating Human QA
Automating production does not eliminate the need to assign accountability for accuracy, judgment, and final approval.
Is it for you?
Best for
Small businesses designing a new role or reconsidering the scope of an upcoming hire.
Not ideal for
Regulated, safety-critical, or high-trust roles where automated processing cannot be validated or appropriately supervised.
From the transcript
“I want you to break down this entire job description and tell me how you would do it with within the structure.”
“And if there are prompts involved, I want you to write the prompts and also tell me in order to do this or whatever it…”
“I hired somebody in a smaller capacity, but we used some of the things that it told me.”
From the episode
What 7-Figure CEOs Know About AI That You Don’t with Rick Mulready
Rick Mulready