ChatGPT and Claude turned a messy portfolio into something hiring managers actually clicked
ChatGPT and Claude were given a job no designer enjoys: take a heap of screenshots, half-written project notes, and a stale “About” page, and turn them into a portfolio that wins callbacks. The Language Model duo didn’t add fake gloss. They built structure, shaved word count, and surfaced the only thing hiring managers care about—outcomes. With a few disciplined Software prompts, the site stopped feeling like a gallery and started reading like proof.
ChatGPT and Claude as the “editor + strategist” pair
The portfolio’s owner, Maya, was a mid-level product designer with scattered work. ChatGPT handled the architecture: page map, case-study sections, consistent metadata. Claude smoothed the voice so it felt human, not demo-day hype. The handoff between the two tools looked like a small, reliable assembly line.
Prompt — Site architecture from fragments
Context: Case-study notes, 20 screenshots, resume bullets, outdated About.
Task: Propose a portfolio IA: Home, 3 Case Studies, About, Contact, Writing.
Constraints: 1 line per page; ensure each case study has Problem → Role → Process → Outcome (metric).
Output: Markdown site map with page-by-page content checklist.
Prompt — Tone cleanup with Claude
Context: Draft case study copy from ChatGPT.
Task: Rewrite to sound like a person after a real project debrief.
Constraints: Keep metrics; remove buzzwords; ≤ 140 words per section; vary sentence length.
Output: Polished case-study section.
Maya stopped overexplaining deliverables and started front-loading results. Employers clicked deeper instead of bouncing.
A case study that reads like a result, not a diary
Hiring managers love numbers and chronological sanity. ChatGPT forced both.
Prompt — Case Study Skeleton (copy/paste)
Context: Redesign of onboarding for a fintech app; goals: reduce drop-off.
Task: Generate the sections and short copy:
- Summary (≤ 60 words),
- Problem (≤ 80),
- Role (≤ 40),
- Process (3 bullets, ≤ 12 words each),
- Outcome (3 bullets with metrics),
- Artifacts (list of 5 with short labels).
Constraints: Plain English; outcomes must include % or Δ time or revenue; avoid “delight,” “game-changer,” “revolutionize.”
Output: Markdown block ready for Webflow/Notion.
Her best project opened with: “Cut onboarding time by 37%, increased first-week activation from 48% to 66%.” No adjectives needed.
Turning screenshots into a narrative instead of a collage
Most portfolios dump images with no context. ChatGPT created micro-captions that sell the thinking behind each artifact.
Prompt — Micro-captions that carry your thinking
Context: 6 screenshots: flow, early wireframes, final UI, experiment results, A/B matrix, emails.
Task: For each screenshot, write: “What we tested,” “What changed,” “Why it worked” (≤ 25 words per line).
Constraints: No design jargon; one metric per caption if available.
Output: Table: Image | Tested | Changed | Why it worked.
Each image became a line in a story arc; recruiters could skim and still understand.
About pages that stop sounding like cover letters
Claude fixed the most underrated page on the site.
Prompt — Human About page
Context: Designer with 6 years in SaaS, prefers small product teams.
Task: Write an About (≤ 120 words) that reads like an intro at a coffee chat, not a resume.
Constraints: Mention one project lesson and one working preference; no empty adjectives.
Output: Final paragraph + 3 “How I work” bullets.
Response rate on inbound messages climbed after the About rewrite; hiring managers referenced her “how I work” bullets in calls.
Before vs after: portfolio performance that matters
| Aspect | Before (gallery feel) | After (results-first) |
| Home page bounce | 63% | 39% |
| Avg. time on case study | 38s | 2:04 |
| Click to “Contact” | 1.9% | 6.7% |
| Recruiter replies (2 weeks) | 1 | 6 |
| Employer feedback | “Nice visuals” | “Clear outcomes—let’s talk” |
ChatGPT turned LinkedIn bullets into portfolio intros
Consistency across channels sells. ChatGPT aligned portfolio blurbs with LinkedIn highlights so both reinforced the same metrics.
Prompt — Align LinkedIn ↔ Portfolio
Context: LinkedIn bullet: “Improved onboarding conversion by 18% Q2.”
Task: Generate a 30–40 word portfolio “Intro stripe” that sets the case study up.
Constraints: Start with the number; mention role and constraint; no fluff verbs.
Output: One-liner for hero stripe.
Every case study now opens with a number, then context. Skimmers get the win first.
A writing page that hiring managers actually read
Short posts prove thinking without adding delay. ChatGPT produced outlines for 3 bite-size write-ups (200–400 words) tied to her work.
Prompt — Writing mini-outlines
Context: Topics: choosing KPIs for onboarding, when to stop iterating on copy, running design QA without hating life.
Task: For each topic, generate a 5-point outline (≤ 8 words per point) + 1 example.
Constraints: No “best practices” clichés; aim for tactical.
Output: Markdown outlines with a suggested title.
Those posts lifted search traffic and gave interviewers conversation hooks.
Chatronix: The Multi-Model Shortcut
Maya bounced between tabs—ChatGPT for structure, Claude for tone, Perplexity AI to verify a benchmark. It cost time. Moving to Chatronix cut the friction:
- 6 best models in one chat: ChatGPT, Claude, Gemini, Grok, Perplexity AI, DeepSeek.
- 10 free prompts to iterate sections before publishing.
- Turbo Mode with One Perfect Answer to fuse structure + tone into a single draft.
- Prompt Library with tagging & favorites so “Case Study Skeleton,” “Micro-captions,” and “About v3” live one click away.
Prompt-engineer script to rebuild a portfolio in 48 hours
Context: Mid-level product designer; assets = screenshots, notes, resume bullets; goal = results-first portfolio that converts to interviews.
Inputs/Artifacts: 3 projects with rough notes; 20 images; 5 metrics you trust.
Role: You are portfolio editor + hiring manager.
Task: Produce, for each project:
- Summary (≤ 60 words, lead with metric)
- Problem (≤ 80), Role (≤ 40)
- Process (3 bullets, ≤ 12 words)
- Outcome (3 bullets with %, Δ time, or revenue)
- Micro-captions table for 6 images (Tested/Changed/Why)
Constraints: No clichés (revolutionize, game-changer, seamless, unlock); plain English; vary sentence length; avoid design-speak; link each claim to evidence (doc/screenshot).
Style/Voice: Calm, confident, specific; reads like a real debrief.
Output schema: Markdown per case study; global “About” (≤ 120 words) + “How I work” (3 bullets); “Writing” outlines (3 topics).
Acceptance criteria: Each case study leads with a number; every image caption explains thinking; About sounds human; total rewrite fits into a weekend.
Post-process: Suggest a homepage hero layout (headline with metric, 3 links to case studies) and a contact panel CTA (email + Calendly).
Steal this chatgpt cheatsheet for free😍
— Mohini Goyal (@Mohiniuni) August 27, 2025
It’s time to grow with FREE stuff! pic.twitter.com/GfcRNryF7u
The portfolio that recruiters actually bookmark
The work didn’t change. The story did. ChatGPT gave the shape; Claude gave the polish. With a results-first template, employers finally saw what mattered in the first scroll. The clicks weren’t vanity—they turned into replies, calendar invites, and offers. This really works when you write for the reader who decides in 30 seconds.