How our AI works

• Turn your CV or imported profile into a draft Career Memory, which you validate.
• Ask you questions to bring out your achievements and their evidence (AI coach).
• Explain in a few sentences why a job posting matches your conditions.
• Write drafts of outreach messages and negotiation counter-proposals, within the mandate you have set.

• It never sends anything without your explicit approval, message by message.
• It never reveals your identity: only your decision, field by field, allows it.
• It does not assess or rank candidates for companies: no company sees a list or score of candidates.
• It makes no decision that produces legal effects concerning you: no automated decision within the meaning of article 22 GDPR.
• It is not trained with your data: we do not train any model.

Each draft is shown to you before anything is sent: you can edit it, reject it or send it. Your guard rails are your own rules, applied as they are; an opportunity’s score is only used to order the postings offered to you, and you remain free to view all of them.

Every message sent to a company states that it was prepared by an AI agent on behalf of a candidate, who reviewed and approved it.

We use models from Mistral AI, a French company, hosted in the European Union. Only the data needed for the requested task is passed to them; your identity, encrypted in your vault, never is.

A language model can make mistakes or word information poorly. Always review a draft before approving it. To report a problem: contact@nextseed-ai.org.