AI Workflows  |  Career Operations  |  Multi-Model Tooling  |  Process Documentation

---

Most people use AI to polish a resume they’ve already written. That’s not what this is.

This project started with a harder problem: I didn’t have a clear answer to the question “what job title should I be targeting?” I had experience across operations, digital marketing, home health agency management, and content work — but no obvious single label that captured all of it. So I used AI not just to format a document, but to help me think through career positioning from scratch, identify the right target roles, and then build role-specific resume variants optimized for each one.

Over several months, that process evolved into a repeatable multi-model workflow involving Claude, ChatGPT, and Gemini — each used for what it does best. This documents how it works.

---

Phase 1: Figure Out What to Apply For

Before building a resume, I had to answer a more fundamental question: what roles actually match my background, and what are they called?

I gave Claude and ChatGPT a detailed picture of my experience — what I had done at each job, the skills involved, the scope of responsibility — and asked them to identify job titles that would be a realistic and strong match. Not aspirational titles. Actual titles that hiring managers use when they’re looking for someone with my specific profile.

That process surfaced three to five target titles, including Content Operations Manager, AI Operations Specialist, and Marketing Operations roles. Each title represents a different emphasis on the same underlying skill set. Knowing that upfront meant I wasn’t building one generic resume — I was building a targeting system.

---

Phase 2: Build the Base Resume

With target titles identified, I fed AI the raw material: my old resume, my LinkedIn profile, and detailed notes on what I actually did in each role — the decisions I made, the systems I built, the results I produced. Things that weren’t on any existing document but were real and relevant.

From that, I had AI construct a properly formatted, achievement-oriented resume built around the language hiring managers and ATS systems actually respond to. Not a summary of job duties. A document that shows scope, impact, and relevance to the target role.

The output came into Google Docs, where I made manual edits — fixing small things, adjusting language, confirming that everything was accurate. That human review pass is part of the workflow, not an afterthought.

---

Phase 3: Build Role-Specific Variants

A resume optimized for a Content Operations role is not the same document as one optimized for an AI Operations role, even if the underlying experience is identical. The emphasis shifts. The language shifts. The skills you lead with shift.

For each target title, I built a dedicated variant. Same core experience, repositioned for a different reader. The process for each one followed the same pattern:

• Provide AI the target title and a sample job description for that role category
• Ask it to identify the skills and language that matter most to that audience
• Reframe the existing resume content to lead with those elements
• Review, edit, and lock in the variant

Over time, each variant became more refined. The Content Ops resume got tighter with each iteration. By the time I had applied to a dozen Content Ops roles, the base document needed minimal editing for each new application.

---

Phase 4: Match Each Application to a Specific Job Description

For each role I considered applying to, I ran a two-step process before submitting anything.

First, I asked AI to evaluate the job description against my profile and give me a direct answer: is this a match, and should I apply? That pre-screening step saved significant time. Not every posting that looks relevant actually is — and having a structured analysis of fit before investing time in an application is a better use of effort than applying broadly and hoping.

Second, if the answer was yes, I gave AI both the job description and my base resume variant and asked it to identify any gaps, missing keywords, or language mismatches. ATS systems filter on specific terms. If a job description uses “workflow automation” and your resume says “process improvement,” that’s a gap worth closing before you submit. AI catches those mismatches faster and more reliably than manual review.

The result was a resume that was meaningfully tailored to each application without starting from scratch every time.

---

Phase 5 (Optional): Company Vetting Layer

At a certain point I added another step to the workflow: before committing time to tailoring a resume for a specific application, I had Claude research and analyze the company itself.

I gave Claude the company name or the job description and let it use web search to find the information on its own — Glassdoor ratings, Indeed reviews, news, anything publicly available that would give a real picture of the company behind the posting. Claude would then synthesize everything into a direct assessment of whether the company was worth applying to.

What it surfaced was useful. Patterns like consistently low management ratings, high turnover signals in employee reviews, or a wide gap between what the job description promised and what current employees described. Things that are easy to miss when you’re moving fast through a job search and just trying to get applications out.

I don’t run this step on every application anymore. The tradeoff is real — adding a vetting filter reduces the number of roles that make it through to the application stage, and when you’re trying to maintain application volume, that friction adds up. But I use it selectively, particularly for roles where I’m about to invest significant time in tailoring or where something about the posting feels off.

The lesson from running this regularly: not every open role is worth your time, and the job description alone doesn’t tell you that. The company behind it matters. This step makes that evaluation systematic rather than gut-feel — and lets Claude do the research so you don’t have to.

---

Why I Used Three Different Models — And What Each One Does Best

Claude, ChatGPT, and Gemini are not interchangeable. Running this workflow across all three taught me where each one has a real edge.

• Claude produces a finished, formatted Word document directly — ready to download, open, and use. No manual reformatting required. For final resume output, this is the most efficient path.
• ChatGPT is strong for structured analysis — evaluating a job description, identifying keyword gaps, and giving direct apply/skip recommendations. Good for the pre-screening and review layer.
• Gemini inside Google Docs is useful for in-document editing — adjusting language, tightening copy, and fitting everything onto one page without leaving the document. When you’re already working in Google Docs, this is the fastest iteration loop.

The workflow uses each tool at the stage where it performs best rather than defaulting to one model for everything.

---

The Core Prompts That Drive the Workflow

These are the prompt patterns I used at each stage. Adapt the bracketed sections to your own background and target roles.

Phase 1 — Role Identification:

Here is a summary of my professional background: [PASTE EXPERIENCE SUMMARY]

Based on this, what job titles should I be targeting? I want realistic matches — roles where my background is a strong fit, not a stretch. List the top 3 to 5 titles and explain why each one fits.

Phase 2 — Base Resume Build:

Here is my existing resume: [PASTE RESUME]
Here is my LinkedIn profile: [PASTE LINKEDIN TEXT]
Here are additional notes on what I did in each role: [PASTE NOTES]
Target title: [YOUR TARGET TITLE]

Build me a clean, one-page resume optimized for this target title. Use achievement-oriented language. Prioritize skills and experience that are most relevant to this role. Format it professionally.

Phase 4 — JD Pre-Screening:

Here is a job description: [PASTE JD]
Here is my resume: [PASTE RESUME]

Is this a strong match for my background? What are the alignments and the gaps? Should I apply? Give me a direct answer.

Phase 4 — ATS Optimization:

Compare this job description to my resume and identify any keyword gaps or language mismatches that could cause ATS filtering. Suggest specific edits to close those gaps without misrepresenting my experience.

JD: [PASTE JD]
Resume: [PASTE RESUME]

Phase 5 — Company Vetting (Optional):

I am considering applying to [COMPANY NAME]. Here is the job description: [PASTE JD]

Using web search, find publicly available information on this company — Glassdoor and Indeed ratings, employee review themes, recent news, and anything else relevant to evaluating them as an employer.

Based on what you find, is this a company worth applying to? What are the green flags and red flags? Give me a direct assessment.

---

What This Actually Saved

The time savings are real but they’re not the most important part. The bigger value is in the decisions that got better.

Starting from a clear targeting framework meant I wasn’t applying randomly and hoping something stuck. Having role-specific variants meant each application led with the most relevant version of my experience. Running each JD through a pre-screening step meant I stopped spending time on applications that weren’t going to go anywhere.

The resume itself became almost a secondary output. The primary output was a clearer, more deliberate approach to the entire application process.

---

Why This Works

Job searching without a targeting framework is expensive. Every application to the wrong role costs time you could have spent on a better match. Every resume that isn’t optimized for ATS costs you a screening you never get to have.

AI doesn’t solve the job search problem. But it compresses the iteration cycle on every decision inside it — what to apply for, how to position yourself, whether a specific role is worth your time, and how to close the gap between your resume and what a specific employer is looking for.

That compression, applied consistently over months, adds up to a fundamentally more efficient process.