Alloquy LogoAlloquy
← Back to BlogAI & Productivity
AI & Productivity

AI Prompts for Your Portfolio That Stick to Real Work

A
Alloquy Team
Published on Alloquy
AI Prompts for Your Portfolio That Stick to Real Work

The best portfolio AI prompts do three things: they hand the model your real evidence, they tell it exactly what to produce, and they forbid it from guessing. Everything below is built on that pattern. You will find prompts for headlines, bios, case studies, site design and a recruiter chatbot, plus examples for a student, a career changer, a returning professional and a veteran of many projects.

The most common failure is not a weak prompt. It is a blank page. People open a chat window with nothing but a job title, and the model fills the gaps with plausible but invented detail. Fix the input first, and the prompts become easy.

Alloquy
alloquy.app
Build a Portfolio From Real Work
Alloquy organizes your work artifacts into a shareable portfolio, with an AI assistant that answers recruiter questions using verified evidence.
See the live demo

Key Takeaways

  • Collect your raw evidence (slide decks, capstone papers, spreadsheets, GitHub repos, client reports) before you write a single prompt.
  • Every prompt should say “use only the facts I provide” and tell the AI to flag what is missing instead of filling it in.
  • Ask for one deliverable at a time: a headline set, a bio, a case study, a chatbot instruction. Short, specific prompts beat one giant request.
  • A portfolio chatbot needs a permission to say “I don’t have that information.” That sentence protects your credibility.
  • Review every generated number, employer and award against your original files before anything goes public.

Table of Contents

Gather Your Evidence Before You Prompt

A prompt is only as honest as the material behind it. Guidance from GoDaddy suggests collecting your resume, selected project evidence, your role and contributions, outcomes, your audience and goal, contact details, visual preferences and any confidentiality limits. It also recommends keeping a separate list of unknowns and requiring the AI to flag missing evidence rather than guess.

For each project you plan to show, Atoms.dev recommends gathering the project name and link, approved screenshots, your exact role, collaborators, constraints, process, deliverables, dates and verifiable outcomes.

In practice, that looks like a folder with real files in it:

  • A college student: the capstone paper, a group project slide deck, a club budget spreadsheet, and a supervisor’s email about a summer internship.
  • A career changer: old performance reviews, a client report, a training deck you built for new hires, and a list of tools you used.
  • A returning professional: project docs from before the break, a volunteer event plan, and any course certificates earned during the gap.
  • An experienced worker: years of quarterly business review decks, KPI dashboards exported to spreadsheets, and promotion packets.

Choose about five representative projects rather than everything you have ever touched. Fewer, better-documented projects produce sharper prompts and a more readable portfolio.

Write the “unknowns” list

This step feels odd, but it is the one that prevents invention. Before prompting, write down what you cannot prove: a revenue figure you only remember roughly, a team size you are unsure of, a launch date. Then paste the list into your prompt with the instruction to leave those items as placeholders. Ashley Scruse’s portfolio workshop uses the same idea: use only facts from your context and project files, and leave TODOs where facts are missing.

The Ground-Rules Block to Paste Into Every Prompt

Put these rules at the top of every portfolio conversation. They work for any model.

You are helping me write portfolio content from my real work.
Rules:
1. Use only the facts and documents I provide below.
2. If a number, date, employer, client, or award is not in my material, write [MISSING: describe what is needed] instead of guessing.
3. Do not use filler phrases such as "passionate about" or "results-driven".
4. Describe my role exactly as I state it. Do not upgrade my title or scope.
5. After the draft, list every claim that depends on a number so I can check it against my files.

Rule 5 matters most. It turns the AI’s output into a checklist you can verify line by line.

Prompts for Headlines and Bios

Headline options

Project headlines get skimmed first. A prompt pattern described by Ruul asks for six headline options in outcome-led, problem-solution or approach-led formats. Adapt it:

Using only the project notes below, write six case-study headlines: two outcome-led, two problem-solution, two approach-led. Under 12 words each. Do not add any number that is not in my notes.

Notes: [paste notes]

Pick the one that matches what you can actually prove. If your notes contain no measurable outcome, an approach-led headline is the honest choice.

Bio in three lengths

One bio is never enough. You need a full version for your About page, a short one for a profile card, and a one-liner for the top of the page. A bio template on Promptitin uses three lengths from the same summary: roughly 200 to 250 words, about 75 to 85 words, and under 25 words. Try:

From the professional summary below, write three bios: a full version (200-250 words), a short version (about 80 words), and a one-liner (under 25 words). Write in first person. Avoid "passionate about" and "results-driven". Use only the facts in the summary.

Summary: [paste summary]

If you want a tighter bio, Ruul’s freelance prompts suggest a 50 to 75 word version written strictly from supplied facts.

Aim at one reader

Tell the model who is reading. A prompt that says “a hiring manager skimming for 15 seconds who wants to know whether I have run a project like theirs” produces tighter copy than one that says “make it professional.” The same advice appears in a developer’s guide to AI-built portfolios, which also recommends capping each project description at three sentences.

Prompts for Case Studies From Real Projects

A case study is a story with proof attached. Sense Central recommends preparing each project around the challenge, approach, solution, outcome and lessons learned. Ruul’s pattern is a 400 to 500 word first-person case study with a brief situation, the core challenge, your approach, the outcome with specific numbers when available (directional language when not), and one thing you would do differently.

Five components of a grounded project case study

Draft a 400-500 word first-person case study from the material below. Structure: (1) situation, (2) core challenge, (3) my approach and why, (4) outcome, using specific numbers only if they appear in my material and directional language otherwise, (5) one thing I would do differently. Mark anything unsupported as [MISSING].

Material: [paste slide deck text, report excerpts, notes]

For design-specific structure, our six-block case study template for designers shows how to lead with the outcome, and the portfolio case study guide covers the full format.

Show how AI was used

If AI helped with your project or your writing, say so plainly. An engineering education paper from ETIS argues for “proportional provenance”: explain where AI helped, where it created risk, and how your judgment governed the result. A case-study method from A Square Solution adds two sections worth borrowing: Verification (what you checked and how) and Failures and iteration (what went wrong and what improved). Those sections show judgment, which a polished final deliverable alone does not.

Add a short section titled "How I verified this" to the case study. Use only the checks I list below. Do not invent verification steps.

Checks I performed: [list]

Prompts by Career Stage

The same structure works for everyone. What changes is the raw material.

College student or recent graduate

You may feel you have no experience. You have a capstone, lab write-ups, a club role and a part-time job. Paste in the capstone abstract and your own notes:

I am a recent graduate. Below is my capstone paper abstract, my role in the team, and my notes on what I personally built. Write a three-sentence project description for a hiring manager. Describe my contribution only. Do not claim team results as mine.

For how to turn an internship deck into something recruiters can explore, see our walkthrough on turning one bullet into a queryable case study slide deck.

Career changer

The job is translation: old duties described in the language of the new field. A teacher’s lesson-planning spreadsheet becomes evidence of curriculum design and data tracking, provided the spreadsheet really shows it.

Here are three documents from my previous career: [paste excerpts]. List the transferable skills (skills that carry over between jobs) that these documents actually demonstrate for a [target role]. For each skill, quote the line in my document that supports it. Do not list a skill without a supporting line.

The “quote the supporting line” rule is the guardrail. It keeps the result grounded in what you did.

Returning professional

A career break should be stated plainly, not hidden. Use your old project docs as the anchor and add any recent learning or volunteer work as separate, labeled items.

Using the project documents below from before my career break and the list of recent courses and volunteer work, write a short portfolio introduction that states the break in one neutral sentence and then moves to my strongest evidence. Do not invent activities during the break.

Experienced professional

Your problem is volume. Ten years of decks and dashboards need ranking.

From the project summaries below, pick the five that best show [leadership / cross-functional impact / technical depth] for a [target role]. Explain each choice in one sentence using only the facts I gave. Flag any project where the outcome is not documented.

For developers, these developer portfolio examples show what hiring teams tend to look at, and the ten AI portfolio examples cover other fields.

Prompts for Building the Portfolio Site

Design prompts are separate from career-evidence prompts. Keep them apart. GoDaddy advises specifying who the site is for, the action you want visitors to take, the pages and content to include, the projects to feature, the visual mood, references and brand elements.

Build a portfolio site for [role]. Audience: [hiring managers at mid-size SaaS companies]. Main action: [book an intro call / download my resume]. Pages: Home, Projects (five), About, Contact. Mood: [calm, minimal, high contrast]. Use only the copy I supply. Where copy is missing, insert a visible TODO.

One practical point from a personal AI portfolio build: keep your static project pages separate from the AI chat, so the site still works and can be indexed if the chat backend is down.

Prompts for a Portfolio Chatbot

A portfolio chatbot answers recruiter questions about your work. Its prompt is less about style and more about boundaries. Developers who have built these describe a consistent pattern.

  • Build from one data source. One builder’s write-up keeps the profile and projects in a single source, builds the system prompt (the standing instructions the model reads before every conversation) from that data, supplies real project numbers, and has the bot politely decline unrelated questions.
  • Ground answers in retrieved passages. This technique is called RAG, short for retrieval-augmented generation. In plain words: before answering, the system looks up the most relevant passages from your approved documents and writes its answer from those, not from memory.
  • Allow “I don’t know.” A portfolio assistant case study treats “I don’t have that” as a first-class answer, and says to tell visitors when something, such as salary, is not in the portfolio.
  • Mark gaps in the data. Another honest-chatbot build encodes strengths, moderate skills and gaps in the data itself, because prompts alone can drift.

A starter chatbot instruction:

You answer recruiter questions about [Name]'s work. Use only the portfolio material provided. Every factual claim must come from that material. If the answer is not there, say "That isn't in [Name]'s portfolio" and suggest contacting [Name] at [email]. Never guess salary, dates, or responsibilities. Do not speak as [Name]; speak about [Name]'s documented work.

The last line keeps the assistant honest about what it is: a guide to your evidence, not a stand-in for you.

Test it like a recruiter

Write ten questions a recruiter would really ask: “What was your role on the migration project?” “What happened when the launch slipped?” “Do you have management experience?” Run them, and compare each answer to your source files. The same portfolio build recommends an evaluation set of representative recruiter questions to test factual consistency. Our guides on adding a grounded chatbot recruiters trust and launching an AI portfolio assistant without infrastructure go deeper on setup.

Check Everything Before You Publish

AI drafts are starting points. Atoms.dev warns against leaving generated employers, clients, awards, testimonials or metrics unverified, and recommends comparing claims with source material, confirming role attribution and permissions, testing links and images, and checking mobile readability and alt text.

A four-pass review works well:

  1. Source pass. Match every number and date to a line in your original document.
  2. Role pass. Confirm the draft says what you did, not what the team did.
  3. Exaggeration pass. Look for superlatives and rounded-up figures you cannot defend in an interview.
  4. Privacy pass. See the next section.

A staged review like this mirrors an AI output review checklist that keeps the original task brief, checks accuracy against source documents, runs a privacy and bias review, then records an approval decision. Keeping a short log of what the AI got wrong is also useful: it shows you where to be most careful next time.

Protect confidential work

A prompt can’t tell what you are allowed to share, so you must decide. Rukh Labs advises reviewing agreements and policies before publishing, not assuming a finished project is safe to show, and requesting written permission when appropriate, such as under an NDA (non-disclosure agreement) or employer rules. It also recommends removing identifiers like names, emails, account IDs and invoice references, and checking surrounding context, since a dashboard can reveal an organization even after names are removed.

Add this to your prompt:

Confidential items that must not appear: [client names, internal tool names, revenue figures]. Use generic descriptions such as "a regional retailer" only where I have written permission to describe the work at that level.

Protect confidential work

If you are unsure whether you may share something, ask your former employer or client, or a lawyer, before it goes live.

Perspective

We built Alloquy around a simple observation: the problem with most AI portfolio prompts is not wording, it is evidence. A clever prompt cannot make up for an empty folder, and it cannot be trusted if it is allowed to improvise.

That is why we start with your real files. You connect or upload slide decks, capstone papers, spreadsheets, performance reviews or client reports, and Alloquy builds a master resume from them. Google Drive access is strictly read-only, and your original files are never stored, edited or altered by us. Resume bullets and cover letters are generated from that linked evidence, not from generic AI claims.

We also think a portfolio should answer questions, not only display work. Each portfolio can include a 24/7 assistant that responds to recruiters using only your verified deliverables. It does not stand in for you. It points to your proof. We see it as a companion to your LinkedIn profile and your PDF resume, not a replacement for either.

The prompts in this guide work in any chat tool. Our aim is to make the discipline behind them, evidence first and no guessing, the default.

Alloquy Team

Build the Master Profile Once

If you would rather not paste documents into a chat window for every application, Alloquy keeps your achievements in one master profile built from your linked files. From there you can generate tailored resumes and cover letters for specific job descriptions in seconds, and share an interactive portfolio with a recruiter-facing assistant. You can start with a free profile and compare plan options on the pricing page. No outcome is guaranteed, but you will never face the blank page again.

FAQ

What is the best AI prompt for a portfolio?

There is no single best prompt, but the strongest ones share a shape: your real material, a specific deliverable, a named reader, and a rule to flag missing facts instead of guessing. Start with the ground-rules block above, then add one request, such as a bio or case study.

Can I use ChatGPT or Claude to write my portfolio?

Yes, any general chat assistant can draft headlines, bios and case studies if you supply real documents and check the output. Treat the result as a first draft. You remain responsible for every claim, so verify numbers, dates and role descriptions against your files.

What should a portfolio chatbot prompt include?

Include your verified project facts, an instruction to answer only from them, and permission to say the information is not available. Add a line telling it to point visitors to your email for anything missing, such as salary or availability. Then test it with realistic recruiter questions.

How do I stop AI from inventing details in my portfolio?

Give it source documents, forbid anything outside them, and ask it to mark gaps as [MISSING]. Then ask for a list of every numeric claim and check each one yourself. Never publish a figure you could not explain in an interview.

Is it safe to put client work into an AI tool?

Only if your agreements allow it. Read your NDA or employer policy first, strip names and identifiers, and get written permission where appropriate. If you are unsure, check with the client, your employer or a lawyer.

Alloquy
Discuss Your Portfolio Evidence
Email Alloquy with questions about turning your work artifacts into a resume and interactive portfolio grounded in your real deliverables.

Sources

#portfolio ai prompts
A

Alloquy Team

Insights and technical analysis from the Alloquy team on AI career intelligence, executive positioning, and verified talent networks.

Build Your Verified Portfolio →