The graph is the product
Job requirements connect to specific pieces of candidate evidence: work experience, projects, skills, education, and certifications. Each connection has an explanation. Gaps stay visible, and every generated application section links back to its sources.
What the score means
Role fit describes coverage in the information you provided. It uses weighted requirements, up to three evidence matches per requirement, and a penalty for missing required evidence. It is not a prediction of hiring or interview outcomes. The workbench shows the calculation so you can assess the result.
The parser uses deterministic text rules and a limited skill vocabulary. It can miss nuance, equivalent qualifications, and career context. Review extracted requirements and use clear, specific evidence.
Your data stays under your control
Anonymous packets are retained only in your browser’s local storage, one packet at a time. Text is processed by the application server during generation and is not saved to the database or logged. Anyone using the same browser profile may see the local packet. Use Delete demo to remove it.
Signed-in packets and candidate profiles are stored in Supabase and protected by owner-based row-level security. Delete packets at any time. Clear your reusable candidate profile in Settings. Saving a demo into an account always requires an explicit action.
AI assistance is optional and off by default. If you explicitly enable it while signed in, relevant requirement text and selected evidence are sent to the configured Gemini or OpenRouter provider. The baseline workflow needs no AI service. Application logs contain operational counts and timing, not resumes or job descriptions.
Generated application materials are drafts. Review before submitting.