Hire on the signal, not the noise.
qsift reads every interview, screen, and note for a role - then surfaces the candidates worth your time and flags the conflicts worth a second look. No ATS to learn.
Built by a hiring manager who's screened hundreds of applicants with it - not a demo.
Free to start. No credit card. For teams hiring 5-20 roles a year, not running a recruiting department.
- Read 48 applications & 19 interview notes
- Calibrated to Senior Backend · Berlin
- Ranked on signal, not keywords
- Flagging cross-stage conflicts…
Conflict flagged. Phone screen rated comms “excellent”; onsite panel said “needs work.” Worth a closer look.
Every ATS is a filing cabinet. You go to it to do the work.
Greenhouse, Ashby, Lever, Comeet - all competent record-keepers. None of them read your interview notes and tell you what to ask next. None flag when a CV claim contradicts what the phone screener wrote. None are built for a hiring manager with twenty minutes between meetings to make a call.
Decide on the candidate, not the platform.
qsift captures interview feedback in your own words, calibrates it against the role, and surfaces conflicts across every stage - so the signal comes to you, instead of you digging for it.
Less plumbing. More signal.
Cross-stage flag engine
Claude reads every interviewer's note and surfaces the inconsistencies a rules layer misses - claim-versus-observation conflicts, late-stage drops after a strong start, scoring jumps between consecutive stages. Acknowledge or dismiss inline.
CV states 6 yrs React; phone screen noted "mostly Vue, ~2 yrs hands-on React."
NL-first feedback
Type a paragraph after the call. Claude extracts strengths, concerns, a score calibrated to the role, and a recommendation - all editable before you save.
⌘K Ask qsift
The palette is context-aware: at a position it knows the pipeline; at a candidate it knows their applications; at an application it sees every note. Real RAG over your data.
Pipeline with owners
Drag-and-drop stage builder. Each stage carries a default owner, so new applications auto-assign. Exponential weighting means F2F outweighs the CV screen in the overall score.
Interview scheduling
One-click schedule for any interview stage. Add to Google Calendar via deep link or download an .ics, with owner and candidate pre-filled as attendees.
Email templates
Acknowledgment and rejection templates with mail-merge - name, position, sender, org. Auto-prompted right after a reject, sent through your own mail client, and logged.
CV + JD extraction
Upload a CV - Claude pulls name, email, role, and location. Paste a job description - Claude fills the whole position: responsibilities, requirements, salary, employment type.
We're not trying to be Greenhouse.
They have multi-channel posting, billing, and a marketplace of integrations. We have the AI copilot they don't - plus the basics any user notices in the first five minutes.
| Capability | qsift | Greenhouse / Ashby / Lever |
|---|---|---|
| Cross-stage conflict detection (LLM) | ||
| NL-first feedback extraction | ||
| Context-aware ⌘K palette | ||
| Pipeline with stage owners | ||
| Reject / advance / source / JD / scheduling / email | ||
| Multi-poster (LinkedIn / Indeed / job boards) | source field | |
| Marketplace of integrations |
The things people ask first.
Where does my candidate data live?
How does the AI actually work?
Do I need to import my whole ATS?
Who is it for?
How do I sign in?
On the roadmap
- WhatsApp / Telegram ingestDictate notes from your phone - qsift parses them and attaches them to the right candidate.
- Real-time email + calendar syncManaged email delivery with tracking, plus two-way Google Calendar sync.
- Public careers page + job feedsAuto-syndicate openings with structured Schema.org markup for Google for Jobs.
- MCP serverTalk to your hiring pipeline from Claude Desktop - ask, advance, and flag from anywhere.
Make the next hire the easy part.
Free to start. One role. No ATS to learn - your data stays yours, export or delete it anytime.
Start free