Tailoring isn't lying: where the line is

Every résumé tailoring tool can produce a lie if you ask it the wrong question. We have explicit guardrails in our prompt; most tools don't. Here's the line, and what happens when you cross it.

The line

Tailoring is rephrasing what's true to emphasise the parts relevant to a given role. Lying is making the résumé say things that aren't true. The middle ground — which is where 80% of LLM-generated tailoring lives — is “technically defensible but probably misleading.”

Five rules

1. Never invent employers, titles, or dates.A title is a fact. “Senior Engineer” is not interchangeable with “Engineer.” The dates are a fact. The employer name is a fact. None of these get tailored, ever.
2. Years of experience are additive, not multiplicative.If you have 3 years in React and the JD wants 5, you have 3. Tailor what you did with those 3 years; do not write “5+ years of front-end experience” if you don't have it.
3. Skills must be real, not aspirational.The tailored bullet can surface a skill you de-emphasised on the base — but only if you've actually applied it on a shipped project. “Used Kubernetes to deploy services” is fine if you used Kubernetes. It's a lie if you read about it on Sunday.
4. Metrics must be traceable.“Reduced p99 latency by 40%” is a real claim only if you can point to the commit, the dashboard, the teammate who measured it. If the LLM writes a metric that you can't source, delete it.
5. Cover letters never name a role you didn't hold.“In my role as engineering lead” — were you actually the engineering lead? Tail-end LLM hallucination. Catch it.

The signals that catch over-tailoring

Recruiters have gotten good at spotting tailored-too-far résumés. The pattern matches:

• Skills sections that read like a JD.Verbatim phrasing from the JD, in the same order. That's not tailoring; that's reflection. Recruiters notice.
• Bullets that don't match the title or the date.“Led migration to Kubernetes 1.28” under a 2019 role — Kubernetes 1.28 was released in 2023.
• Metrics that scale impossibly.“Reduced costs by 90%” reads as fictional because it almost always is. Costs that drop more than 50% in a single intervention are rare; LLMs over- inflate.
• Inconsistent tone across bullets. One bullet sounds like the candidate; the next sounds like a marketing site. The handoff is the LLM.

The interview is the test

The cleanest test for whether you've crossed the line: can you talk about every claim on the résumé in depth, without flinching, in an interview? If the answer is no for any line, that line is over-tailored.

Recruiters and hiring managers will probe the most relevant bullet. “Walk me through how you reduced latency by 40%.” If you can't walk through it, the interview ends and the position is filled by someone who could.

What tools should and shouldn't do

The tailoring tool's job is rephrasing, reordering, and emphasising. Not authoring. Specifically, the tool should:

• Surface skills you've listed in your master résumé that match the JD.
• Reorder bullets to put the JD-relevant ones first.
• Tweak phrasing to match the JD's vocabulary.

The tool should not:

• Add skills that aren't in your master résumé.
• Invent metrics.
• Combine bullets in ways that imply scope you didn't have.

We enforce these as explicit rules in our system prompt and reject outputs that violate them. The tool ships with these on by default.

The simple test for the user

Read the tailored résumé to a friend who knows your work history. If they squint at any bullet, that's the one to soften.