To sort applications with AI, start where applications actually land: in the shared bewerbung@ mailbox or in your ATS. Monday, 07:40. Carla runs operations at a 180-seat restaurant in Hamburg and opens Outlook: 83 new line cook applications since Friday, another 14 waiting in Personio, two of them already asking when they will hear back. The job ad has been live for nine days, she hired the last cook five months ago, and the first service of the week starts in three hours. This guide shows exactly how to hand the reading, parsing and shortlisting to an AI employee, step by step, without losing one good candidate and without a single automated rejection. Below you will see how Superkind’s ready-made AI team for applications reads, scores and schedules on its own.
The pattern is the same for HR teams in a mid-sized manufacturing plant hiring shift workers, a retail chain filling 40 open shelf-stocker roles or an agency with six developer vacancies. Different roles, same bottleneck: a human has to read every CV once before anything else happens.
Key takeaways
AI reads, a person decides - Superkind reads every incoming application in the bewerbung@ mailbox and in Personio, extracts the key facts, scores against the must-haves you wrote on the vacancy and prepares the shortlist. Interview invites and every rejection wait for a human click.
GDPR Article 22 and the EU AI Act stay respected - no decision is based solely on automated processing: Superkind drafts the shortlist with the cited CV line behind every score, and your hiring manager confirms it before any candidate hears back.
Explain it once, it keeps running - a rule like “Line cook: German B2, weekend availability, two years of line experience - everything else goes on the maybe list” turns one round into a daily routine.
Get a ready-made AI team for applications - Superkind comes with a hiring team (Application Scout, Shortlist Reviewer, Interview Scheduler) that reads, scores and schedules automatically. Start for free with unlimited users, or build your own.
Keep candidate data in Europe - CVs carry personal data under the strictest category. Superkind is hosted in Europe, GDPR compliant and never trains on your data.
Your AI team for applications
What does it mean to sort applications with AI?
Sorting applications with AI means an AI reads every incoming CV and cover letter, pulls out the facts the hiring manager actually needs (name, location, earliest start date, must-have skills), scores each application against the criteria written on the vacancy, groups them into clear yes, maybe and clear no, and hands a human the shortlist to confirm before any candidate is invited or rejected. The AI does the reading, the parsing and the comparing. The invitation, the rejection and the hire stay with a person. One example is Superkind: a ready-made AI team that reads every application in your bewerbung@ mailbox and in Personio, scores it against your criteria and prepares the shortlist, hosted in Europe.
The difference to a generic prompt in ChatGPT: an AI employee does not stop at summarising a CV. It knows which role the application belongs to, which must-haves this vacancy carries, which applications from last round are still open and when a candidate is waiting too long for a reply.
How it works with Superkind: your AI team for applications
Superkind is the AI employee for your team: you write to Superkind by email (Outlook or Gmail), in Microsoft Teams, in Slack or in the Superkind web app, like you would write to a colleague, and Superkind does the task in your programs. Hosted in Europe, GDPR compliant, with 1,500+ integrations into the systems where hiring work already lives: Personio, SAP SuccessFactors, Workday, Outlook, Gmail, Teams, Slack, SharePoint, Google Drive and Excel. No new ATS to buy, no copy-paste between tools.
You do not start from scratch: Superkind comes with ready-made AI employees for hiring teams. They run on their own in the background whenever a new application arrives, and only the moments that matter (rejections, invitations) wait for your OK. Use them right away, adapt them, or build your own.
Application Scout
AI employee for intake and parsing
Monday morning the bewerbung@ mailbox holds 83 new applications and the Personio portal 14 more. The Application Scout reads every message as it comes in, pulls name, location, earliest start date, desired salary and the three most relevant skills from the CV, tags it with the right job opening, and drops the structured summary into the Teams channel for that role. No one types a line.
Shortlist Reviewer
AI employee for pre-selection
Carla, the hiring manager, does not want 83 CVs; she wants the ten she should actually read. Every morning at 9:00 the Shortlist Reviewer scores the new applications for each open role against the must-haves and nice-to-haves written on the vacancy, cites the line in the CV behind every score, and posts the top ten in Teams. Carla approves the shortlist. Rejections are never sent automatically.
Interview Scheduler
AI employee for candidate communication
Candidates on the shortlist want a reply within 48 hours, and silence loses good people. The Interview Scheduler sends the confirmation of receipt to every new applicant on its own, and once Carla has approved the shortlist it offers three slots from her Outlook calendar for a 20-minute first call. Rejections wait for Carla to click send so every no stays personal.
Once a shortlist is approved: send a confirmation to every new applicant, offer three slots from Carla's calendar to the top ten, hold rejections for Carla.
Reasoning ⌄11 interview invites sent with three Monday to Friday slots each. 97 confirmations of receipt sent this morning. 62 rejection drafts prepared and waiting for Carla.
Every employee is ready to use right away. Or build your own on the platform.
From the first application to your own app
The same path is told once on this topic. Each stage builds on the last:
- 1
Start
Ready in minutes. Sign up for free at app.superkind.ai, connect the bewerbung@ mailbox and Personio, forward one open vacancy and watch Superkind parse the next application into a structured card in Teams.
- 2
Relief
Tasks become routines. Every new application goes through the same scoring path, the confirmation of receipt goes out on its own, and the shortlist waits for Carla’s OK, because that is how you set it up. Rejection emails are never sent automatically.
- 3
Knowledge
Learns from day one. Your Company Brain grows with every hire: which phrasings in a cover letter signal a reliable line cook, which schools train the strong apprentices, which must-haves were worth it. When the recruiter changes roles, the knowledge stays.
- 4
Growth
When it gets more complex, you do not start over. Build a custom hiring app with its own interface on the Superkind platform, on top of what Superkind already knows about your win rate per source and per role. Built yourself or with the Superkind team, and it all belongs to you.
Send Superkind your first stack of applications
Start for free in Outlook, Teams or Slack.

What you need
Check these points before you start. Most of them already sit in the hiring workflow you run today.
Checklist before the first round
- Access to the inbox where applications land: Outlook or Gmail for the bewerbung@ mailbox, and the career portal if you post there
- Access to your ATS or HR system: Personio, SAP SuccessFactors, Workday, Rippling, Greenhouse, Lever, Ashby, Recruitee, Workable, SmartRecruiters or Breezy HR
- A written vacancy with three to five must-have criteria per role (not vague wishes) and two or three nice-to-haves
- An agreed approver per role: the team lead or hiring manager who confirms every shortlist and sends every invitation and rejection
- Teams or Slack where the hiring team already talks, so the shortlist lands in the channel instead of another inbox
- One paragraph in your privacy notice that tells candidates AI helps with intake and sorting and that a person at your company makes every decision
- A free Superkind account at app.superkind.ai, and if you have a works council, a short info note so they can see the audit log
- Thanks to the 1,500+ integrations in Superkind, these systems connect without new tools or custom code
Step by step with Superkind
Five steps from the first forwarded application to a confirmed shortlist. Each step is something you can do today.
Step 1: Connect the mailbox and the ATS
Sign up at app.superkind.ai and connect the systems applications flow through: Outlook or Gmail for the bewerbung@ mailbox, Personio or your other ATS for the career portal, Teams or Slack for the hiring channel. An admin confirms which mailboxes and portal areas Superkind may read. No API keys to copy, no custom code.
- Check - Superkind only sees the mailboxes and roles you allowed, and the audit log records every application it opened.
- Tip - start with one role (the one with the most applications per month) and the three best past hires for that role.
Step 2: Write the must-haves in one short message
Tell Superkind, once, how to sort for this role. Keep it to criteria that describe the job, not the person. Three to five must-haves is enough.
For Line Cook Nordhafen: must-haves are German B2, at least two years of hot-line experience and weekend availability. Nice-to-haves are HACCP certificate and Italian cuisine.
Superkind confirms with “Noted.” From now on every new application for that role is scored against those criteria, with the exact line in the CV cited behind every score.
Step 3: Let Superkind read every incoming application
As soon as a new mail arrives at bewerbung@ or a new application lands in Personio, the Application Scout reads it: name, location, earliest start date, desired salary, must-have match. The structured card lands in the Teams channel for that role, the candidate gets a confirmation of receipt within minutes, and the application is filed in Personio under the right opening. No one types a line.
Forward every new application in bewerbung@ to Superkind, send the standard confirmation of receipt and file it in Personio under the matching opening.
Step 4: Approve the shortlist in Teams
Every morning at 9:00 the Shortlist Reviewer posts a card in the hiring channel: the top candidates with the scored criteria, the cited CV lines and a clear “yes / maybe / no” recommendation. Carla reads the shortlist, promotes one maybe to the yes list, demotes another and clicks Approve. Only now does anything move.
- Confirmed yes list - the Interview Scheduler offers three slots from Carla’s Outlook calendar to each candidate on the yes list.
- Maybe list - stays open for a second look, with a reminder in three days.
- Clear no list - Superkind drafts a personal rejection for every candidate; the email is sent only after Carla clicks send.
Step 5: Keep the audit trail
Every step Superkind did sits in the audit log: which CV was read, which criteria were applied, which score was given, which line in the CV the score came from and who approved. For GDPR Article 22 and for the EU AI Act Annex III logging duty, that record is what you need. Export it to SharePoint or share it with the works council whenever they ask.
Export the hiring log for the last line cook round to our SharePoint folder HR > Audit 2026.
Answer: “Done. 142 applications read, 11 invited, 62 rejected, every rejection confirmed by you. Log saved to SharePoint.”
Set it up as a routine
Once one round has gone through cleanly, stop giving the instructions each time. Tell Superkind once what should happen for every open role.
For every open role in Personio: read each new application the moment it arrives, send the confirmation of receipt, file it under the right opening, and post the top ten in the role’s hiring channel every morning at 9:00. Wait for my OK before any invitation or rejection goes out.
- Daily - every new application parsed and filed on arrival, shortlist posted at 9:00 for each role still open.
- Weekly - a status card: applications in, shortlisted, interviewed, rejected, open roles with low volume so you can refresh the job ad.
- Monthly - a source report per role: which channel brings the strongest candidates, so the next ad budget goes to the right place.
When the process grows, for example a hiring team across five sites, a scoring model per role family or a shared candidate pool, you do not start over. On the Superkind platform you build a custom hiring app with its own interface, on top of everything Superkind already knows about your roles. Built yourself or with the Superkind team, and it belongs to you. The 1,500+ integrations keep the app connected to Personio, Outlook and SharePoint without new middleware.
Data protection and EU hosting
Safe by default
CVs are personal data. Superkind keeps them in Europe and never uses them to train models.
- Hosted and processed in Europe
- GDPR compliant
- Aligned with the EU AI Act
- No training on your data
- DPA through the Trust Center
A CV carries name, address, date of birth, photo, salary expectation, references and sometimes health or trade-union information. Few data categories are as sensitive, and few are copied across as many tools in a day. Superkind is built for that.
What this means for an application: the CV Superkind reads, the scores it writes and the shortlist it drafts never leave the EU and are not used to train a model. Confirm the AI note in your privacy notice and the information duty under EU AI Act Article 26 (deployers of high-risk AI inform affected persons), and the legal footing for sorting applications sits solid.
Other ways to sort applications
Superkind is not the only way. Here is an honest comparison of the common options.
| Way | Effort | Error risk | Best for |
|---|---|---|---|
| Superkind | Low: connect mailbox and ATS, write must-haves in plain language | Low: cited CV line per score, approval before every invitation and rejection, EU hosting | Hiring teams across Personio, Outlook, Teams and SharePoint, with 1,500+ integrations and an AI team ready on day one |
| Manual in Excel | Very high: copy data by hand, sort columns, chase missing files | Medium: inconsistent scoring, missed candidates, no audit trail | One role at a time, few applications a month |
| Built-in ATS scoring (Personio, softgarden) | Medium: set up matching rules in the vendor UI | Low in the tool, higher outside it (no email intake, no team channel, no cross-system work) | Teams that run everything inside one ATS |
| Generic ChatGPT or Gemini prompt | Medium: paste CV text, copy result back | Medium to high: no link to ATS, no audit log, data leaves the EU on consumer plans | One-off review of a single CV, no hiring record |
| RPA (UiPath, Power Automate) | High: build and maintain rigid flows | Low in the flow, high when the ATS UI changes | Repetitive copy steps with no judgement needed |
The common gap of the alternatives is cross-system work and audit traceability. For wider context on the German HR stack, see our AI tools for Personio guide and the AI in recruiting piece.
Common mistakes
These five mistakes cost the most time and the most good candidates when teams start to sort applications with AI.
- Letting the AI reject on its own - a rejection has legal effect, which GDPR Article 22 blocks when a machine takes it alone. Always let a person click send on every no.
- Vague must-haves - “cultural fit” or “hands-on mentality” cannot be scored and often hides bias. Write criteria that describe the job: German B2, two years of hot-line experience, weekend availability.
- Skipping the candidate information - under GDPR and the EU AI Act, candidates must be told AI is used in intake and sorting. One paragraph in the job ad or the application confirmation is enough.
- Hiding the audit log from the works council - in Germany the Betriebsrat has co-determination rights on AI use in HR. Share the log and the must-have criteria from day one.
- Never looking at the clear-no pile - a 2026 Personalwirtschaft survey showed about half of HR professionals never re-check what got auto-sorted out. Pull a random five each week and read them.
6 tips for sorting applications with AI
1. Write the must-haves on the vacancy, not after the fact
If you cannot name three concrete must-haves for a role, the ad is not ready. Write them in the brief, send them to Superkind and the same criteria run through intake, scoring and interview notes.
2. Keep nice-to-haves separate from must-haves
A nice-to-have never knocks a candidate out; it only lifts them if everything else is equal. Superkind scores both lists separately so a strong candidate who speaks no Italian is not accidentally rejected for a role where Italian is nice-to-have.
3. Put the shortlist where the team already talks
If the hiring manager lives in Teams, the shortlist goes to Teams. If recruiters run in Outlook, keep it in Outlook. Superkind uses the same rules and approvals across the 1,500+ integrations, so the channel is a choice, not a lock-in.
4. Read the clear-no pile every week
The Personalwirtschaft survey found about half of recruiters never re-check sorted-out CVs. Pull a random five every Friday, read them, and tell Superkind what it missed. The next round sorts better.
5. Make data protection a hiring argument
Candidates increasingly ask where their CV is processed. “Hosted and processed in Europe, DPA under GDPR, no model training on your CV” is a short, honest line in the privacy notice. It also covers the EU AI Act Article 26 information duty for deployers of high-risk AI.
6. Involve the works council early
In Germany, the Betriebsrat has co-determination on the use of AI in HR. Show them the must-haves, the shortlist screenshot, the audit log and the human approval step. The conversation is much shorter than the one you have after launching without them.
Conclusion
Sorting applications with AI does not mean a machine decides who gets a job. It means the reading, parsing and scoring is done for you, the rules you wrote on the vacancy are applied to every CV, and a person approves every invitation and every rejection before a candidate hears back. The fastest way to see whether it fits: take the next open role, point Superkind at the bewerbung@ mailbox and Personio, write the three must-haves and compare its shortlist with the one you would have built by hand on Monday morning. If you want a wider map first, read our guide to AI recruiting tools.
Frequently Asked Questions
Yes, as long as a person makes the final decision. GDPR Article 22 forbids decisions based solely on automated processing that produce legal or similarly significant effects on a person, and a rejection falls into that category. AI may read, score and rank; a human must confirm every invitation and every rejection and must be able to explain the reasons. The EU AI Act (Regulation (EU) 2024/1689) lists CV filtering and candidate ranking in Annex III as a high-risk use, which means logging, bias monitoring and information duties for the deploying company.
Yes. GDPR requires you to inform candidates in the privacy notice of the job ad which categories of data are processed and whether automated tools are involved. Under the EU AI Act, deployers of high-risk AI in employment must inform candidates before use. The simplest way: one paragraph in the job ad or the application confirmation that AI helps with intake, sorting and the shortlist, and that a person at your company makes every invitation and rejection decision.
Superkind is hosted and processed in Europe, is GDPR compliant, is aligned with the EU AI Act and never trains any model on your data. The CV, cover letter and the scoring stay in your Personio, SharePoint or Outlook; Superkind reads them to prepare the shortlist and writes the result back. Access rules let you pick who sees which role, and a data processing agreement (DPA/AVV) is available through the Trust Center.
Via the 1,500+ integrations Superkind works directly with Personio, SAP SuccessFactors, Workday, Rippling and the ATS products Greenhouse, Lever, Ashby, Recruitee, Workable, SmartRecruiters and Breezy HR, plus Outlook, Gmail, Teams, Slack, SharePoint, Google Drive and Excel. For a system without a ready API, Superkind works in the program natively via computer use, in the browser or in the Windows client, with your approval wherever you want it.
Any sorting tool can carry bias, so three things matter. First, write must-have criteria that describe the role, not the person (German B2, weekend availability, two years of line experience, not "cultural fit"). Second, let Superkind cite the exact line in the CV for every score, so a reviewer can see why. Third, let a person confirm every invitation and rejection. Superkind supports all three, logs every step in the audit trail and never auto-rejects.
You start for free at app.superkind.ai with up to EUR 150 in credits, no credit card and no sales call. For regular hiring work the Business plan starts at EUR 50 per month for the whole company with unlimited users, which means the whole hiring team, team leads and the works council sit on the same plan. A single application reading plus scoring typically costs 50 to 150 credits, and EUR 100 covers up to 1,000 tasks. As of October 2026.
No. Confirmations of receipt go out automatically, because candidates expect a signal within a day. Shortlist decisions, interview invites and rejections wait for a human click. The Interview Scheduler drafts every rejection in your voice, you read it, you send it. This matches GDPR Article 22 and keeps every no personal.
Related Articles
- AI in Recruiting: Tools, Examples and Tips
- The Best AI Tools for Recruiting
- AI Tools for Personio
- Managing Absences in Factorial with AI
Sources
- EUR-Lex - Regulation (EU) 2024/1689 (EU AI Act), Annex III high-risk list
- EUR-Lex - GDPR Article 22 (automated individual decision-making)
- Bundesministerium der Justiz - Allgemeines Gleichbehandlungsgesetz (AGG)
- LTO - Darf KI ueber Bewerbungen entscheiden? (automatisierte Vorauswahl)
- Human Resources Manager - KI in der Bewerbervorauswahl
- Personalwirtschaft - Umfrage: halbe der Personaler prueft automatisch aussortierte Bewerbungen nicht nach
- softgarden - Datenschutz-Hinweis zum KI-Matching (Art. 6 Abs. 1 lit. f DSGVO)
- Personio - Recruiting features (official product page)
- SRD Rechtsanwaelte - KI im Personalwesen und EU AI Act
- Superkind - Trust Center (hosting, DPA, no model training)
Send Superkind your first stack of applications.
Start for free in Outlook, Teams or Slack. Or let us show you in 30 minutes how it runs in your Personio and bewerbung@ mailbox.

