Home › Blog › How to Use AI to Prepare a Portfolio Walkthrough for an Interview (Prompts and Example, 2026)

How to Use AI to Prepare a Portfolio Walkthrough for an Interview (Prompts and Example, 2026)

To prepare a portfolio walkthrough with AI, give a chatbot your raw notes on each project and ask it to shape them into a spoken story (the problem, the constraints, your decisions, the outcome and what you would change), cut that story to the time you have, and then act as a critical reviewer who asks why you made each choice. AI is fast at structure, timing and predicting questions.

To prepare a portfolio walkthrough with AI, give a chatbot your raw notes on each project and ask it to shape them into a spoken story (the problem, the constraints, your decisions, the outcome and what you would change), cut that story to the time you have, and then act as a critical reviewer who asks why you made each choice. AI is fast at structure, timing and predicting questions. It cannot know which work is yours, what really happened or how you talk, so the choice of projects, the facts and the final wording stay with you.

The method suits designers, writers, marketers and engineers asked to "walk us through something you built". For what the round itself looks like, see our guide to portfolio review interviews.

What AI can and can't do for a portfolio walkthrough

Part of the walkthroughAI helps byYou still have to
Choosing projectsScoring each project against the job descriptionPick work you made and can discuss in depth
Story structureTurning messy notes into problem, constraints, decisions, outcome, reflectionSupply every fact, including what went wrong
TimingCutting a rehearsal transcript to a word budgetRehearse with a timer until it fits
Critique questionsPredicting 'why this and not that' questions for each projectAnswer from memory of the real decisions
Confidential workMaking an already-anonymized version read naturallyCheck your NDA and what you are allowed to show
DeliveryFlagging jargon and sentences too long to saySound like yourself and talk to the panel, not to a script

A seven-step workflow from project notes to rehearsal

  1. Ask for the format. Ask the recruiter how long you have, how many projects they want, whether it is a presentation or a conversation, who will be in the room and whether you present your own slides. Formats vary widely between companies and roles, and the answer sets your time budget.
  2. Write raw notes for four to six candidate projects. For each: who had the problem and how you knew, the constraints, your exact role, two or three decisions and the options you rejected, the outcome and how you know, and one thing you would change. Half-sentences are fine.
  3. Strip confidential details before pasting anything. Replace client names, product codenames, unreleased features, internal figures and colleagues' names with placeholders such as [CLIENT].
  4. Choose the projects with Prompt 1. Two or three projects that show range usually beat five shallow ones, unless the recruiter asked for a specific number.
  5. Build a speaking outline per project with Prompt 2, then match your slides to it with Prompt 3. Keep the outline as short bullets.
  6. Record a full run-through on your phone, get a transcript and cut it to time with Prompt 4. A calm speaking pace for most people is somewhere around 120 to 150 words a minute, which gives you a word budget per project.
  7. Run the critique round with Prompt 5, answering out loud. Fix the slide or the story wherever an answer felt weak, then repeat with a colleague or friend if you can.

Six copy-paste prompts for a portfolio walkthrough

1. Pick the projects that fit this job

Here is a job description for a [role] role and short notes on [number] projects from my portfolio. Do not rewrite anything yet. Score each project from 1 to 5 on: relevance to the main problems in this job description, how clearly my personal role shows, evidence of outcome, and whether it contains a real trade-off I made. Then recommend the [two or three] projects that together show the most range, and say what each one proves that the others don't.

Why it works: it makes the chatbot judge against the job instead of praising everything, and the last line stops you from picking three versions of the same project.

2. Turn raw notes into a speaking outline

Using only the facts in my notes below, write a speaking outline for one portfolio case study in five parts: the problem (who had it and how we knew), the constraints, my decisions (two or three, each with the option I rejected and why), the outcome, and what I would change. Use bullets of no more than 12 words, in first person singular for my own actions. Where a fact is missing, write [ASK ME] instead of guessing. Notes: [paste notes].

Why it works: the [ASK ME] rule keeps invented metrics out, and short bullets give you cues to talk from rather than a script to recite.

3. Match the slides to the story

Here is my speaking outline and a list of my slides, one line describing each. Tell me which slides don't support any point in the outline, which points have no visual, and where I show process artifacts (sketches, journey maps, wireframes) without saying what decision they led to. Suggest a slide order for a [number]-minute slot.

Why it works: most walkthroughs show too much process; this finds the slides that display effort without explaining a decision.

4. Cut a rehearsal to time

This is a transcript of me presenting one case study. It ran [X] minutes and I have [Y] minutes. Cut it to about [target] words without removing any decision I made, the outcome or what I would change. Shorten the background first. List every cut so I can put back anything important.

Why it works: background is usually what runs long, and the list of cuts lets you restore anything that mattered.

5. Play a tough reviewer

Act as a senior [design lead / creative director / engineering manager] reviewing my portfolio for a [role] role at a [type of company]. I will present one case study. Then ask me six questions, one at a time: what exactly I did versus the team, an alternative I rejected, how I knew it worked, a disagreement with a stakeholder, what I would change, and one direct critique of the work itself. Wait for each answer. At the end, name my weakest answer and the slide that would have helped it.

Why it works: the questions afterward test whether you understand your own decisions. Our pages on UX designer interview questions and product designer interview questions list more role-specific ones.

6. Make anonymized work read naturally

My case-study notes below already use placeholders such as [CLIENT] and [PRODUCT]. Rewrite them so they read naturally with neutral descriptions (for example 'a regional bank'), turn any remaining absolute figures into relative ones (for example 'about a third fewer support tickets'), and flag anything that could still identify the client. List every change.

Why it works: you remove the names yourself first, and the AI then makes the anonymized version readable and points out details you missed. Only you can check the result against your agreement.

Before and after: the first 30 seconds of a case study

Openings are where walkthroughs lose the room. Here is a transcript of a first attempt (an example we wrote for a product designer), the edit applied, and the result.

First attempt (spoken): "So this project was for a fintech client, and basically we did a ton of research, like interviews and surveys and a competitive audit, and then we did a lot of wireframes, and there were a few rounds of testing, and the stakeholders had a lot of opinions, and eventually we launched the new onboarding flow, and it did really well, conversion went up a lot, and everyone was really happy with it."

Edit applied: Prompt 2 on the designer's notes, plus one line: "Make the first 30 seconds state the problem, my role and the result before any process." The designer then filled in the [ASK ME] gaps from memory.

Improved opening: "This is the onboarding flow for a consumer budgeting app. The problem: more than half of new sign-ups never connected a bank account, so they never saw the product working. I was the only designer, with a product manager and two engineers, over eight weeks. I'll show you two decisions: moving the bank connection after a first quick win, and cutting the sign-up form from nine fields to four. After launch, the share of new users connecting an account went from about 40 to about 55 percent. At the end I'll tell you what I'd change."

What changed: the problem, the role and the result come first; a list of activities became two decisions the panel can ask about; vague praise became a before-and-after figure from the designer's own notes. No number? Describe what you observed instead, such as fewer support tickets on the topic.

Check the AI's draft before you rehearse it

Rehearse it out loud: a three-run drill

Plan three timed runs over two or three days, each one recorded.

RunWhat you doWhat you look for
1. ColdPresent every project start to finish with a timer, no stoppingWhere you ran long, looked at notes or lost the thread
2. CutPresent again after the cuts from Prompt 4Finishing each project with a little time to spare
3. QuestionedPresent one project to the AI reviewer (Prompt 5), answering out loud before typingWhich question made you hesitate; fix that slide or story

Portfolio rounds rarely stand alone. Expect "tell me about yourself", "why this company" and behavioral questions around the walkthrough, and rehearse those as spoken answers in an AI mock interview built from your résumé and the job description, or start with typed answers in the free practice demo. For slide design and timing advice, see our guide on preparing a presentation for a job interview, and for the wider picture, our guide on how to use AI for interview prep.

Where CoPilot Interview fits

CoPilot Interview is an AI interview assistant for rehearsing interviews out loud. Its AI mock interview builds questions from your résumé and the job description, reads them aloud and scores each spoken answer out of 10, which suits the behavioral and motivation questions that usually surround a portfolio review. The walkthrough itself should come from your own slides and your own preparation. The free plan gives 15 minutes a day of live help in sessions of up to 15 minutes, with AI mock interviews included. Paid plans are Standard at $24.99 a month and Pro at $39.99 a month. It runs as a desktop app for Windows 10 and 11 and macOS 12 or later, a free iPhone and iPad companion on the App Store (iOS 16 or later) and an Android companion in beta. In any live interview, follow your employer's rules on AI tools.

Disclosure: CoPilot Interview is our product, so weigh what we say about it accordingly.

Frequently asked questions

How long should a portfolio walkthrough be?

As long as the slot allows, minus real time for questions. Ask the recruiter for the length and the number of projects, then plan to finish each project slightly early, because the discussion afterward is where much of the evaluation happens.

How many projects should I show in a portfolio interview?

The number they ask for. If they don't specify, two or three projects that show different skills usually work better than five shallow ones.

Can I use ChatGPT to write my portfolio case studies?

Yes, for structure, editing and timing, as long as every fact is yours and you can talk about each decision without the text in front of you.

What if my best work is under NDA?

Show an anonymized or redacted version only if your agreement allows it. Otherwise choose another project, or describe the process in general terms and say why you can't show the screens.

Should I read from speaker notes during the walkthrough?

Short cue bullets are fine where the format allows, but don't read a script; reviewers want to hear you reason.

Rehearse it out loud before the real one

The AI mock interview builds questions from your resume and the job description, reads them aloud and scores each spoken answer out of 10. Included on the free plan.

Start a mock interview