You do not need to call yourself an AI expert to show that you can use AI thoughtfully. A small portfolio can demonstrate a useful task, the decisions you made, and how you checked the result.
For an experienced worker learning AI in their 40s, 50s, or beyond, this can make an abstract claim such as “comfortable with AI” more concrete. It does not guarantee an interview, overcome discrimination, or replace qualifications that a job requires.
Choose one task worth showing
Start with a task relevant to the work you want to do. You might prepare a clear customer information sheet, organize a set of fictional requests, or improve instructions for a process you understand.
A portfolio is a collection of work samples. CareerOneStop’s portfolio guidance recommends explaining each project, your contribution, and its results. You can apply that idea to an ordinary workplace task without building a machine-learning model.
Choose a sample you can explain and check yourself. If the exercise depends on knowledge you do not yet have, narrow it or get appropriate help before presenting it as evidence of competence.
Use material you can safely share
For a first practice project, invented information is often enough. Label it “independent practice using fictional data.” Do not describe it as work commissioned by a client or deployed by an employer.
Before reusing actual workplace material, check ownership, confidentiality, and permission to share it. Removing names or a logo does not settle those questions. Creating a new fictional example may be simpler than trying to sanitize an old employer document.
If you later want to sell a template based on your experience, treat permission and ownership as a separate check. A useful portfolio sample is not automatically yours to commercialize.
Build a case study on one page
Use these headings to make your contribution visible:
- Task: what needed to be produced, and for whom?
- Starting material: what information did you use, and was it fictional, public, or shared with permission?
- AI’s role: what did the tool draft, suggest, sort, or explain?
- Your work: what did you decide, verify, correct, or create yourself?
- Evidence: what can a reader inspect in the finished sample?
- Limits: what was not tested, and what would you check next?
Include the date and the tool or model name when available. Keep a copy of the instructions and the relevant output so that you can explain the process later.
A beginner project you can try
Create six fictional requests for a small equipment-lending service. Include questions about collection times, changes to a reservation, and a report of damaged equipment. Do not use real customer details.
Ask an AI assistant to group them:
Group these fictional requests into Collection questions, Reservation changes, Reported damage, or Needs clarification. Preserve each request’s ID. Explain briefly why you chose the category. Do not answer the customer, invent service policies, or take action. Flag a request if more than one category could apply. Requests: [insert your six examples].
Compare every classification with the original message. A request to cancel a reservation belongs with reservation changes; it is not automatically a request for a refund. A damaged item needs the appropriate person to assess it, not an AI-generated repair instruction.
Keep the original requests, the proposed categories, and your corrected version. Your case study can explain the categories you chose and the errors you actually found. This is a practice exercise, not evidence that a real organization saved money or improved service.
Describe results without manufacturing success
If you measured time, explain what you timed and include checking and correction. Do not claim a percentage improvement when you have no comparable baseline. “Created a checked category list for six fictional requests” is a clear, limited outcome.
Avoid letting AI promote the exercise into a larger achievement. “Tested a draft classification prompt” does not mean “built an automated customer-service platform.” Name any help you received and distinguish your work from a team’s contribution.
You can ask AI to shorten the case study, with this condition: “Use only my notes. Preserve limitations. List missing evidence instead of inventing results.”
Share a version that is easy to inspect
A short PDF or simple page is sufficient. Use readable text, descriptive links, and captions that explain the sample. Check that the file opens, the links work, and the intended viewer can access it. Review comments, file properties, hidden spreadsheet sheets, and screenshots for private details before sharing.
A portfolio supports your application; follow the employer’s requested format. Use our AI resume-checking guide to keep the accompanying claims accurate.
Age discrimination is not something an applicant can solve with a better document. US readers can find information about protections and complaint routes through the Equal Employment Opportunity Commission. Your portfolio’s job is to show relevant work honestly.
Featured photograph by Ashley West Edwards on Unsplash. Photograph used for illustration.
AI disclosure: This article was drafted and revised using AI. Read our full disclaimer.