Claude turns 'worked on the recommendation system' into 'led a 4-engineer rewrite that lifted CTR 18% across 12M weekly users', constrained to the words you'd actually say, the keywords the JD asked for, and the outcomes you can defend in an interview.
Worked on the recommendation system and improved it for many users.
Led a 4-engineer rewrite of the recommendations service in Python and PyTorch, lifting click-through 18% across 12M weekly users.
Lands the JD keywords (Python, PyTorch, scale), quantifies impact, and names the team you led.
Was responsible for marketing campaigns and grew the user base over time.
Ran a lifecycle marketing program across HubSpot and Customer.io that grew MAUs from 38k to 112k in two quarters at a CAC of $7.40.
Names the tools the JD wants, sets the baseline, sets the result, and proves the unit economics.
Helped customers and resolved their issues quickly.
Owned tier-2 support for a portfolio of 240 enterprise accounts on Zendesk and Intercom, holding a 92% CSAT against a 4-hour first-response SLA.
Substitutes 'helped' (filler) for 'owned' (ownership), and replaces 'quickly' with the actual SLA the recruiter is screening for.
Six rules apply to every bullet the rewriter produces. They aren't soft prompts. They are post-generation checks that re-prompt Claude if the output drifts.
Claude, Anthropic's frontier model. We use it because it preserves the tone of your existing CV more reliably than GPT-class models, doesn't over-flowery the bullets, and follows length constraints (max 22 words per bullet, no semicolons) without nagging.
No. The rewrite is constrained to the content you already have in the CV plus the JD keywords you tell it to target. If a JD asks for Kubernetes and you've never written it on the CV, the AI proposes a rewrite that's neutral, and prompts you to confirm whether you actually have the skill before it gets included.
Either. The editor lets you select a single bullet and ask for a stronger version, or select an entire experience block and ask Claude to rewrite the whole thing against a specific job description. The CV outside your selection is never touched.
Every accepted rewrite triggers a re-score. The score change is shown live (e.g. +6 pts). Rewrites that drop the score are auto-flagged so you can revert before applying.
No. The CV text we send to Claude is processed under the standard Anthropic API privacy terms, which forbid training on customer data. Your CV is not used to train any model, by anyone, full stop.
Yes. Pick the CV and the JD, and Claude drafts a 3-paragraph cover letter that opens with a specific hook from the JD, names two of your most relevant bullets, and closes with a clear ask. You edit the result inline.
Free trial, no credit card. Cancel any rewrite you don't like.
Start rewriting →