AI projects7 min read

Turn a finished project into a clear portfolio example

A portfolio entry should help someone understand what you did and how you think. Build that explanation from the work you actually completed.

Career Auntie illustrated character

The Career Auntie team

Updated

Start free: five credits every month. No card needed.

Browse courses
Jump to the worked example
On this page

Choose the question your project will answer

A useful project begins with a question that a reader can understand. Can you turn sample sales data into a reliable weekly report? Can you organise unclear handover notes into actions and unresolved questions? Can you design a small campaign proposal and explain why you chose its messages? Each question points towards an output and a way to review it.

Start with a role you want, but narrow the exercise to one part of that role. You are creating an example of work, not demonstrating an entire occupation in a weekend. Read the responsibilities in an advert and choose a task you could practise with suitable information. If you already have a strong example from paid or voluntary work, you may need a clearer explanation of that experience rather than another project.

Define what a finished first version will contain before asking AI for help. This prevents the assistant from expanding a short report into an elaborate product you cannot complete. Write down your audience, intended output, available time and main constraints. A clear brief makes later choices easier to explain because the reader can see what you were trying to achieve.

Collect evidence during the work

Keep a small record as you go rather than trying to reconstruct the story at the end. Save the source material you are allowed to use, an early draft, a meaningful change and the final output. Record the reason for the change while it is still fresh. You do not need to preserve every AI response or upload a long conversation that distracts from the work.

For a spreadsheet project, evidence might include a sample input table, a formula check and a corrected chart. For a writing project, it could be the brief, the draft that missed the audience and the revision that addressed the problem. For a process guide, it could be a checklist and notes from trying to follow it. Choose evidence because it explains a decision, not because it looks impressive in a screenshot.

Keep private material out of anything you share. Crop screenshots carefully, review visible tabs and comments, and avoid including names or identifiers that are not yours to publish. Invented examples should remain clearly labelled. A reader should be able to tell whether they are looking at a practice exercise, a proposed solution or work that was actually used.

Make your contribution visible alongside AI

Describe AI as part of the process. Name the type of help you used: suggesting an outline, producing alternative wording, explaining a formula, organising notes or challenging an assumption. Then describe the decisions you made. Which sources did you choose? What did you reject? What did you change after checking the result? What remained your responsibility?

“The AI produced a report” says little about your contribution. A more useful explanation is: “I defined the reporting question, checked the totals against the input table, removed an unsupported explanation and rewrote the conclusion around what the figures could actually show.” Use that pattern only when it describes work you really did. It is a model for clarity, not a sentence to copy as an achievement.

Do not present a prompt as proof that you completed the task. The prompt explains what you asked for; the output and your review show what happened next. Equally, you do not need to hide ordinary AI assistance. An honest account of tool use and judgement gives a reader more to discuss than a vague claim that everything was done unaided.

Show a check that could have found an error

A useful check has a clear purpose. Recalculating a total independently can reveal an arithmetic error. Comparing a summary against its sources can reveal a missing condition. Testing a checklist with a new example can show whether its instructions are usable. Asking the same assistant whether its answer is correct may offer suggestions, but it is not the same as independently verifying the result.

Choose checks that fit the output and record what they found. Include an awkward example: a missing value, a contradictory note, a request outside the policy or a message that could be misunderstood. If a check changes your output, keep the before and after. If it finds no problem, describe what was checked without implying that every possible error has been ruled out.

Some things cannot be established in a practice exercise. You may not know whether customers would respond to the campaign, whether staff would use the process or whether a model works on real data. State those limits. A proposed measurement plan can be useful, but label it as a next step. Do not turn expected benefits into invented results or business savings.

Write a summary an employer can scan

Once the work is complete, organise the explanation around six questions. What was the problem? How did you approach it and use AI? What did you personally contribute? How did you check it? What did you produce or observe? What remains uncertain, and what would you do next? Keep each answer focused on details that help the reader understand the work.

Use a descriptive title such as “Checked weekly sales report using invented shop data”. Explain the context in a sentence, then move quickly to your actions. A reader should not have to decode a long tool list before understanding what you made. Include only links or screenshots that support the explanation, and give enough context that the entry still makes sense without opening every attachment.

Review the summary for accidental exaggeration. Replace “improved sales” with the actual result if the project only produced a campaign proposal. Replace “automated the team’s workflow” with a description of the prototype if nobody used it at work. You can still explain what the prototype was designed to improve, while keeping its intended benefit separate from an observed outcome.

A worked outline for an invented reporting project

Worked example

Project: a checked weekly report for a fictional neighbourhood shop.

Problem: the sample sales table was difficult to scan, and the brief asked for a short summary of changes between two weeks.

Approach: use a spreadsheet to calculate differences and ask an AI assistant for alternative ways to explain them to a shop manager.

Personal contribution: choose the comparison, build the calculations and decide which observations belong in the final report.

Checks: compare totals with an independent calculation, review missing values and check every sentence against the table.

Outcome to record after doing the work: the report you actually produced, including any correction you made. Do not claim that a real shop used it or that it changed sales.

Limitations: invented data, a short comparison period and no real user testing. A possible next step is to test the report with another sample dataset.

This is an outline for an exercise, not a completed customer case study. Replace it with your own evidence and results; leave any claim out if you did not perform the work that would support it.

Use the completed-project review in Career Auntie

Career Auntie lets you begin from an original idea, a saved application or a career goal. Choose to generate the first AI brief for one credit, including the later plan and stage prompts. Sign in if asked and the same request continues. Review and edit the brief and its end date before generating that included plan. You can also sign in to write your own brief; generating its AI plan costs one credit. Each AI refinement of the brief costs one additional credit; manual edits are free. After you accept the edited plan, copy the prompts into your own AI tool and bring answers or screenshots back for each stage. External AI tools may have their own charges.

When every stage is complete, the project-summary section appears. “Fill empty sections from my work” can reuse submitted words where they support a section. It cannot supply missing facts or verify screenshots, so review the text and complete the remaining sections honestly. Saving the summary stores it with your project. Previewing it lets you check the proposed profile entry before sharing.

“Add to my profile” copies the saved write-up into Project successes on your profile. Your existing audience settings still apply, and you choose whether to include submitted screenshots. This does not automatically update the separate Career Portfolio used for CV and application generation. Keep that distinction in mind when deciding where you want the example to be used.

Use the entry as the start of a conversation

When discussing the project in an application or interview, connect it to a relevant responsibility. Explain the problem, one decision and one check before describing what you learned. Be ready to show the source or calculation behind a claim. A smaller example that you understand thoroughly is easier to discuss than an ambitious output you cannot explain.

If you later revise the work, review the write-up as well. Adding the project to your profile again updates its saved profile entry; simply editing the private project does not silently rewrite what you previously shared. Keep descriptions accurate as the work changes. The value of the entry comes from the evidence and explanation you can stand behind, not from a promise that publishing a portfolio will produce a job offer.

Common questions

What belongs in an AI project portfolio?

Include a clear problem, an output, your contribution, how AI helped, checks you performed and the result or limitations. Use a few relevant screenshots or links rather than uploading every draft.

Can I include practice projects without work experience?

Yes, provided they are clearly labelled as practice and you explain any invented data or hypothetical setting. Do not describe expected benefits as real business results.

Does completing a project publish it automatically?

No. Completion makes the summary and profile-sharing step available. You review the write-up, save it with the project and separately choose whether to add it to your profile.

Does adding a project to my profile update my Career Portfolio?

Not automatically. The profile showcase and the private Career Portfolio used for CV and application generation are currently separate. Saving or publishing the project summary does not update that private portfolio.

Does a complete stage mean the work has been verified?

No. Completion records that the required evidence was submitted. It does not certify skills, establish that an image is accurate or replace your own checks.

Career Auntie, your illustrated career guide

Your next step

Make your next career step something you can show

Start from your idea, application or career goal. Sign in to generate or write an editable brief, review its plan, then build with prompts for your own tools.

First AI brief, plan and stage prompts: 1 credit. Each AI brief refinement: 1 credit. Manual edits are free.

Start free: five credits every month. No card needed.

Applying regularly?

Make more of every application with Unlimited

Tailor CVs and cover letters for each role, unlock creative designs and prepare for interviews.

Loading Unlimited price…

Get Unlimited

Billed upfront. Renews automatically; cancel future renewals at any time.

Fair usage: 45 credits a day, 1350 per 30-day period.

Compare plans Estimate the credits you need