Same uncomfortable truth as résumés: humans don't read cover letters first anymore — AI does. The application portal screens them, the recruiter copilot tags fit-signals against the job description, and the hiring manager (if they read it at all) sees a one-paragraph AI summary before they ever open your actual letter. So if AI is the first reader, the smart move is to refine your cover letter using AI. WaxFrame runs a hive of them — ChatGPT, Claude, Gemini, and more — on your draft at once, catching what one model gives you a pass on.
Free 3-round trial — no signup, no credit card. Unlock unlimited rounds with a $19 lifetime license. Open WaxFrame ↗ · Cover Letter Template ↗
True story: WaxFrame's creator landed his current job by doing this process by hand, end to end — and the cover letter went through the same gauntlet as the résumé. He took a draft, opened ChatGPT, Claude, and Gemini in separate tabs, fed each one the same draft and the same job description, collected the notes, reconciled them himself, then ran the rewrite back through. Four rounds in, the letter that came out the other side was the one that opened the door.
That manual back-and-forth was the prototype for WaxFrame. The app exists because the result was that much better than any single AI on its own — but the manual version was tedious enough that nobody would do it twice. WaxFrame is that workflow with the tedium removed: paste the letter once, paste the job description once, the hive reviews in parallel, a Builder synthesizes the consensus, you decide what to keep.
You are not the first person to use a hive of AIs to refine a cover letter. You're just the first person to do it without 47 browser tabs and a notepad.
Modern cover-letter screening is layered. The application portal often runs the letter through an ATS that keyword-matches against the job description. A recruiter copilot then tags fit-signals — does the candidate explain the role connection, do they mention the company, does the close ask for a next step. Only then does a human spend 5-10 seconds reading either the actual letter or the AI's summary of it. Every layer has different failure modes: ATS misses you for missing the right phrase, the copilot tags "generic" if you don't reference the role specifically, the human bounces if the hook doesn't land.
A single AI catches some of these. ChatGPT tends to over-formalize. Claude polishes voice but can soften specifics. Gemini is fast but sometimes glosses over the role-fit connection. Grok pushes harder on tone; DeepSeek is methodical but quiet on flair. WaxFrame runs all of them on your draft at once. Each returns numbered, specific suggestions — "tighten paragraph 2," "the hook is generic," "the close is weak," "you didn't mention the company by name," "the job description's 'cross-functional' keyword isn't in your letter." A Builder AI synthesizes the strongest into a refined version. You review every change. Three or four rounds is typical for cover letters.
The result reads like one author wrote it — but it's been pressure-tested by half a dozen, including against the bots that screen it first.
Will: tighten the hook so it lands in three seconds, sharpen the role-fit connection between your background and the specific job, calibrate tone for the company culture, kill generic openers ("I am writing to express my interest"), kill the filler phrases hiring managers skim past, surface where you're underselling a real accomplishment, flag where you're overselling without evidence.
Won't: fabricate experience you don't have, invent numbers or accomplishments, lie about your skills, or make claims the hiring manager could disprove in 30 seconds on LinkedIn. WaxFrame's Builder is explicitly instructed not to add facts that aren't supported by what you provide.
Open WaxFrame, pick the Cover Letter template, paste your draft into the Starting Document screen. Paste the job description into the Reference Material screen — the hive will check your letter against what the employer actually asked for, every round. Pick a Builder AI (Claude or ChatGPT work well here). Run 3-6 rounds, reviewing what changed after each one. Export when you're satisfied.
Same template, but click Start from Scratch on the Starting Document screen and leave the text box empty. Before Round 1, open the Notes drawer and paste three things: (1) why you're a fit for this specific role, (2) one concrete accomplishment with numbers if possible, (3) what specifically excites you about the company or position. The hive will build a first draft from those talking points. Run rounds the same way.
The single biggest mistake people make with AI cover letter tools is not feeding them the job description. WaxFrame's reviewers cite against your reference material every round — if you paste the JD, every reviewer will flag where your letter does and doesn't match the stated requirements. If you don't, the hive will produce a competent but generic letter.
Paste the full posting. The first 500 chars of "About Us" boilerplate isn't worth as much as the actual requirements list, the team description, and the application instructions. The hive can read 10,000+ characters of reference material without slowing down.
Round 1. Six reviewers flag: weak hook, three filler phrases, role-fit connection is implied but not stated, missing the company name, one accomplishment is buried. Builder rewrites with all five fixes. You review and accept four of five changes.
Round 2. Reviewers flag: the new opener is stronger but a touch formal for this company's voice, the second paragraph is tighter but loses the metric you wanted to keep, close is still generic. Builder rewrites. You accept.
Round 3. Two reviewers return NO CHANGES; the other four flag minor polish (a comma, a stronger verb). Builder applies. You export to .docx.
Total time: ~15 minutes. Total cost: usually under 50 cents in API calls on a paid Builder, or free if Gemini Flash is your Builder on AI Studio's free tier.
Document type set to "Cover letter." Target audience pre-scripted ("Hiring manager at [company] reviewing applications for a [job title] role"). Desired outcome pre-scripted ("Reader wants to interview me. Three paragraphs: hook, role connection, confident close with a clear call to action"). Scope & constraints already populated ("Three paragraphs max. No 'I am writing to express my interest' openers. No filler phrases. No 'I look forward to hearing from you' sign-offs"). Reference Material field labeled for the job description. Best AIs For: Claude for voice and authenticity, ChatGPT for punchy rewrites, Grok for tone calibration.
You can override any of these once the template loads. They're starting points, not constraints.
1. Open WaxFrame. waxframe.com ↗ — runs in your browser, no install, no signup. Works on desktop and laptop (Chrome, Firefox, Edge, Safari at 1366×768 or wider).
2. Apply the Cover Letter template from the Template Catalog ↗ — auto-populates the Goal fields.
3. Want the deep version of this workflow? See the Cover Letter playbook ↗ — every field, every option, with reasoning.
4. Cost question? See AI API Pricing ↗ — live per-token rates for every supported provider.
WaxFrame is fully open source under the AGPL-3.0 license. Every line of code is public and auditable.
💡 Your API keys go directly from your browser to each AI provider. WaxFrame never sees them. Read the source on GitHub →
About
Built with ❤️ by WeirDave and Claude.
Your WaxFrame Pro license is active.