The recruiter who reads everything you write.
Donna is an AI recruiter that talks to other recruiters on your behalf. She reads your resume, your GitHub, the things you tell her about what you actually want. Then she has a real conversation with each company's recruiter agent. Both sides ask questions, both sides answer. At the end, you get a match score and a transcript.
The problem we got tired of
Job sites match on keywords. You uploaded a resume that says "Python" so every Python job emails you, regardless of whether the role is senior or entry-level, in your city or three time zones away, building something you'd love or something you'd hate. Recruiters get spammed with five hundred applications per role and pick from the top of the pile.
Nobody is reading anything. The conversation that actually matters, "tell me about your experience scaling a Postgres cluster" / "I'd rather not work weekends" / "what's the team like?", happens late, if at all. Most of it is wasted on people who were never going to be a fit.
What Donna does instead
When you sign up, Donna onboards you in a real conversation. She parses your resume, asks about your priorities, your dealbreakers, the kind of place you'd thrive. She builds a profile that reads like a referral, not a search query.
When a hiring manager posts a job, Donna does the same on the other side. The job's requirements, the team's culture, the actual ideal candidate, written down in plain language.
Then her two agents talk. A candidate agent representing you, a job agent representing the role. Both ask probing questions: technical depth, fit with the team's pace, geographic constraints, salary bands, dealbreakers. Both answer honestly because they're built from your data, not your performance.
Three channels, one ranking
Behind the agent conversations, Donna runs a hybrid retrieval pipeline. Three signals fight for influence on which candidates get surfaced for which jobs:
- Dense semantic match: your profile embedding versus the job's profile embedding. Catches the "this person feels right" signal.
- Ideal candidate match: the hiring manager's description of the perfect hire, embedded and compared to you. Catches aspirational fit, not just the literal JD.
- Sparse keyword match: BM25 full-text search on the structured profile text. Catches specific technologies and credentials that embeddings sometimes blur away.
The three rankings get fused with Reciprocal Rank Fusion. Top candidates get re-scored by a cross-encoder reranker. Then, and only then, the agents start their conversations.
Verification, not vibes
If Donna spots a claim in your profile that looks suspect — three years at a company that didn't exist three years ago, a project described in language that doesn't match anything else in your history — she pauses and asks for evidence. A link, a correction, a "yeah I overstated that". Verified profiles match faster and the agents trust them more.
When she gets stuck, she asks you
Sometimes the recruiter agent asks something Donna can't answer from your profile alone. "What language did you build that distributed cache in?" The conversation pauses. Donna pings you on Telegram, WhatsApp, or the dashboard. You answer once. She continues. Your answer gets stored so the question never has to be asked again.
Donna Pro
For candidates, Donna Pro unlocks two things: detailed feedback on rejections (the actual reasons, with fixability scores) and an anonymized comparison with the candidates who did get shortlisted for the same role. It's $9.99 a month, or ₹400 in India.
For recruiters
Recruiters get pre-screened candidates with full conversation transcripts attached. The "would I want to interview this person" question is already answered before they read the first sentence. We're working on a success-fee model so recruiters only pay when a match leads to a hire. For now, posting jobs and seeing shortlisted candidates is free.
Built by a small team in India and the US. Reach us at our contact page, or read the privacy policy and terms of service.