Pharma Bloodbath Part VII: The machine that applies for you, and why it is quietly applying against you.

[Bookmark this one. I'm building a free library of career advice on this newsletter, one Bloodbath at a time, and this is the instalment to save before you hand €65 a month to a robot that may quietly torpedo you. Everything here is free. For the version with your name on it, morethancareer.de has the actual goods: Career Coaching (job-search strategy, interview preparation and salary-negotiation training with tailored scripts), Reverse Recruitment, and CV and LinkedIn rewrite and rebranding. But read this first, ideally with the bot switched off.]

Dear #MoreThanCareer community,

Welcome back to Part VII, the newsletter your prospective employer's HR team reads in incognito mode, with the brightness turned down, like it's something to be ashamed of. House style, for the newcomers: data, dark humour, and the load-bearing assumption that nobody in talent acquisition is going to tell you the unflattering parts out loud. I will. It is cheaper than therapy and lands in roughly the same place, except I don't validate your feelings, I validate your sources.

Today's patient on the slab: AI-based automated mass-applying. The tools — simpleapply.ai, LazyApply, Sonara, AIApply, LoopCV, Jobright and a clone army of near-identical others with names that sound like they were also generated by AI — promise the one thing every job seeker has prayed for since the invention of the online form: to never see a Workday login screen again as long as they live. It is a beautiful dream. It is also, on the evidence, the career equivalent of paying a stranger to gamble with your name while you sleep, then tipping him. My working thesis, defended where the data earns it and walked back where it does not: as your primary strategy, mass-applying hurts you more than it helps. Viel hilft viel (more helps more) is a touching German folk belief. It is wrong about painkillers, schnapps, sunbathing, and job applications, in ascending order of how badly it ends.

Let us look closely at the mechanism, because a remarkable number of people are paying for this without knowing what they bought, which is also, incidentally, how gym memberships work.

First: what these tools actually do (or, how to outsource your dignity for €65)

The model is the same across the category. You upload your CV. You set preferences. The tool scans the boards, scores listings against your profile with what it generously calls intelligence, fills in the forms, and submits. There are tiers: a manual mode where you approve each match, a semi-auto mode where the AI fills the form and you press send, and a full-auto mode where the software applies to everything it deems a match while you, in the marketing's actual words, sleep.¹

I want to dwell on that selling point, because it is magnificent. The headline benefit is that the product works while you are unconscious. So does a dripping tap. So does a slow gas leak. So does compound interest on a payday loan. The category of things that operate diligently while you sleep is not, historically, a category you want to join lightly. Pricing runs from about €20 to €250 a month, and LazyApply cheerfully advertises somewhere between fifty and three hundred applications a day.² The numbers are real. Whether they are working for you is the entire question, and it is the one question the pricing page treats like a relative it owes money to.

The volume problem, in numbers that should ruin everyone's afternoon equally

Here is the context that turns this whole category radioactive. Applications through LinkedIn rose roughly 45% in a single year, with the platform now processing in the order of eleven thousand applications per minute.³ Read that again, slowly, because the human mind refuses it on first contact. Eleven thousand. Per minute. One US consultant got over 1,200 responses to a single posting within days and took the listing down, which is the professional equivalent of unplugging the smoke alarm because it keeps reacting to the fire.³ A recruiter told CNBC that popular roles now pull three to five hundred applications inside three days, some sailing past a thousand over a weekend.⁴

Sit with the irony, because it is exquisite, the kind you'd frame. The tools that promise to get your application seen are, in aggregate, the precise reason your application will never again be seen by anything possessing a heartbeat. Every auto-applier is feeding the flood that every other auto-applier is drowning in. It is a circular firing squad, everyone has paid a monthly subscription to attend, and the bullets are CVs. Operation gelungen, Patient tot (operation successful, patient dead): the software did exactly what you asked, with admirable precision, and the outcome is measurably worse than if you had spent the afternoon lying on the floor contemplating the ceiling. This is not a bug. This is the feature, running at industrial scale, against everyone who bought it, simultaneously, forever.

Does it get you more interviews? The honest, annoyingly thin answer

Now I have to be straight with you, because the rest of your feed will not be — the rest of your feed is on commission. The hard data on whether mass-applying produces more interviews (as opposed to more applications, which is true the way "I really increased the temperature in here" is true of an arsonist) is weak, and most of what exists is published by companies selling rival products. Treat all of it, mine included, with the suspicion you would bring to a horoscope written by someone who has seen your bank balance.

Caveat duly nailed to the wall, hammer still warm: one documented LazyApply user reported five thousand applications and twenty interviews, about 0.5%.⁵ Five thousand. That is not a job search, that is a hostage situation, and the hostage is the recruiter's inbox. One early provider, Wonsulting, reportedly killed its bulk auto-apply feature in August 2025 after clients averaged roughly one interview per fifty applications, about 2%, a number so dispiriting the vendor looked at its own product, sighed, and pulled the plug.⁶ When the dealer won't touch his own supply, you have learned something about the supply. These are anecdotes and vendor claims, not a controlled trial, and I will not dress them in a lab coat and pretend otherwise. But they all point the same direction with unsettling consistency, and I have yet to find a single credible source pointing the other way. If you have one, the comments are open, and I will eat my words on camera with a side of Currywurst and good grace.

What we can say with confidence is structural. Recruiters keep reporting the identical thing: AI applications have started to look suspiciously similar, because they are all built by asking the same handful of models to cram the same job advert's keywords into a CV.³ When every application is the same beige paste extruded from the same machine, the recruiter cannot tell the qualified from the keyword-optimised, so they retreat to harder filters: referrals, internal candidates, and screening tools with the warmth and curiosity of a parking meter. The crowning irony, the one that belongs in a museum, arrived when the AI companies themselves got the fear. Anthropic, maker of the very model that may be ghost-writing your cover letter at this exact moment, reportedly asked its own applicants not to use AI assistants on their applications.⁷ When the arsonist starts handing out smoke detectors at the door, you do not need a second opinion.

How well does the AI match you to roles? Less well than the confident little slider implies

The matching is mostly keyword overlap wearing a borrowed lab coat and a name badge that says "Intelligence." The better-engineered tools do score fit before applying, in fairness. Independent testing by the job-search engine Adzuna found one of the stricter filters (ApplyIQ) rejected roughly one in five conventional applications and only fired above an 80% match threshold.⁸ That is the premium end of the market, and it quietly indicts the entire rest of it: a tool is only worth its subscription when it spends most of its time heroically refusing to do the precise thing you bought it to do. Picture a taxi that earns your loyalty by declining four fares in five and charging you monthly for the privilege of being turned down.

At the budget end, the reviews read like a blooper reel with the laugh track surgically removed. Users report tools applying to roles outside their stated location and language, and submitting them for senior posts when they are entry-level, all in cheerful defiance of the settings they had carefully entered by hand.⁹ Mit Kanonen auf Spatzen schießen (shooting sparrows with cannons) at least implies the sparrows are in danger. Full-auto mode misses the sparrow, the field, the farm, and on a bad day the hemisphere, then bills you and asks how it did.

And the quality of the generated CV?

Polished, grammatical, beautifully formatted, and increasingly indistinguishable from the four hundred other polished, grammatical, beautifully formatted CVs landing in the identical inbox before the recruiter has finished their first coffee. The German careers press is refreshingly blunt about it: AI is useful for structure and for the keywords an applicant tracking system craves, but fully generated documents lose the personal voice recruiters are screening for, and carry a real risk of inventing detail when the model confidently fills a gap it does not understand and cannot be embarrassed by.¹⁰ A CV that delights the machine and bores the human has aced the mock exam and failed the only paper that was ever marked. Außer Spesen nichts gewesen (nothing to show for it but the expenses), and the expenses, unlike the interviews, arrive monthly and on time.

The DACH twist your feed is quietly not telling you about

Now the part that matters most for this audience, and that the American coverage skips with the breezy confidence of a tourist ordering a Maß and expecting a small one. These tools are built for the Anglo-American model: LinkedIn Easy Apply, Indeed, ZipRecruiter, Workday. The German Bewerbung is a different beast entirely, with different teeth and strong opinions about formatting. It expects an Anschreiben (cover letter) tuned to the specific employer, a structured Lebenslauf, and frequently Arbeitszeugnisse (formal employer references), often submitted through a company's own portal or by email rather than via a one-click board. A tool engineered to machine-gun LinkedIn Easy Apply at the US market is, in Frankfurt, a jet ski at a hiking club. Impressive machine. Wrong terrain. Everybody has stopped to look, and not in the way you wanted.

The behaviour data agrees. A Stepstone survey reportedly found around six in ten applicants in the German market already using AI somewhere in their search.¹¹ But German employers and candidates remain markedly more sceptical of it than their American cousins, who would put AI in the Brötchen if it polled well. Survey data points to the top German worry being precisely that applications will be chewed up by a machine and never reach a human at all.¹² In a culture that still respects a considered, personal application, a visibly mass-produced one is not neutral. It is a small, legible note pinned to your CV that reads: I could not be bothered to write this, so I outsourced caring about you to a €65 subscription, and I would do it again. German recruiters, may they live a thousand years, can read that note from across the room, upside down, in the dark, while doing something else.

The regulation nobody mentions: the EU AI Act polices their robot, not yours

Here is an asymmetry worth your full attention, because it is about to bend this whole arena like a paperclip. Under the EU AI Act (Regulation 2024/1689), AI used in recruitment and hiring decisions is classified as high-risk, which triggers the strictest tier of obligations: documented bias testing, human oversight, transparency to candidates, logging, registration — the entire bureaucratic Buffet, all courses, no skipping the salad.¹³ The AI-literacy obligations took effect on 2 February 2026, and the core high-risk requirements are scheduled for 2 August 2026, though a proposed "Digital AI Omnibus" may yet shift parts of that timeline, so check the current state before you bet the Schrebergarten on the date.¹⁴ Article 86 even hands candidates a right to an explanation of significant automated decisions, meaning you may soon be legally entitled to know precisely how a robot rejected you. Closure, of a sort. Cold comfort, but itemised.

Now read the asymmetry, because it is the whole game. The law regulates the employer's robot, the one filtering you. It says essentially nothing about your robot, the one mass-applying on your behalf at three in the morning. So the recruiter is being legally frogmarched towards transparency, human oversight and trusted sources at the exact moment candidates are sprinting in the opposite direction, arms windmilling, straight into full automation. This is not a fair fight, and the candidate's side has elected, with tremendous enthusiasm and a paid subscription, to run flat out at a door that is being bricked up in real time, while cheering. Volume-spraying cannot reach human judgement, structured criteria, or a referral. Those are precisely the channels the law is herding everyone back towards. Pech gehabt (tough luck).

A brief international comparison, since not all of you are in Leverkusen

United States: ground zero, patient one, and the gift shop. The LinkedIn flood is largely an American event, the tools are mostly American products, and the arms race is most advanced there, the way an infection is most advanced at the site of the wound. It is also where "apply to everything that moves, and several things that don't" is gospel and, by the available evidence, least rewarded. Faith and results have rarely shared a postcode less.

United Kingdom and the rest of Europe: structurally closer to DACH than to the US, with portal-based, reference-heavy hiring, though the UK's fondness for Easy Apply parks it somewhere in the awkward middle, like a country that hasn't decided which queue it's in. And the EU AI Act reaches anyone whose AI output touches candidates in the EU, regardless of where the head office keeps its espresso machine.

India and China: colossal application volumes and, in tech especially, serious automation on both sides of the table. For a DACH reader these are less a destination than a postcard from the future, and the postcard reads "wish you weren't here": the bot-versus-bot end state, where the human applicant's signal rounds cleanly down to zero and everyone involved is tired. Frame it on the wall as a warning, not a holiday brochure.

The throughline is identical everywhere, dispiriting in any language. Where applying is frictionless, volume explodes, signal collapses, and employers flee to the channels automation cannot follow. The accents change. The physics does not. The physics never does.

The verdict, data-based and without a cliffhanger

So: does AI mass-applying hurt candidates? On the evidence, as a primary strategy, broadly yes, and for structural reasons rather than vibes or moralising. It inflates the volume that triggers harder filtering, it produces beige applications recruiters have trained themselves to ignore on reflex, it surrenders all control over where your name ends up, it fits DACH hiring culture like Lederhosen at a board meeting, and where it runs through platforms like LinkedIn it can breach the terms of service outright, since LinkedIn's User Agreement flatly prohibits bots and automated activity and reserves the right to restrict or ban accounts that use them.¹⁵ You can torch the professional network you spent a decade building in order to save yourself forty minutes of form-filling. Der Schuss geht nach hinten los (the shot goes off backwards), and on this particular platform the recoil can take your account with it, neatly, in one go.

The honest caveat, which is the entire reason this is not just more doom dressed up for engagement: the efficacy data is thin, and the narrow use is fine, even smart. Using AI to draft a first version, tune keywords for an ATS, or fill the mechanical fields of a form you then read and personalise is sensible and saves genuine time. The harm lives almost entirely inside one little syllable: auto. The moment a machine is submitting on your behalf to roles you have never seen, using a CV you have never read, you have outsourced not the drudgery but the judgement. And the judgement was the only part of any of this that anyone was ever going to pay you for.

Five things that actually work — calibrated to where you are

1. Apply to fewer roles, properly, and sneak round the front door entirely. The most robust finding in all of recruiting data is that referred candidates are roughly four times more likely to get an offer, and that referrals, while only a thin slice of all applications, account for a fat slice of actual hires. SHRM's reporting describes about one in ten referrals converting to a hire, against job boards routinely needing fifty to sixty applicants to produce a single one.¹⁶ Ten well-aimed applications with one warm introduction will, on this evidence, flatten three hundred bot-fired ones and not even break a sweat. This holds for everyone, but it is scripture for executives and C-level readers, for whom mass-applying is not merely useless but radioactive with a long half-life. At that altitude, roles move through network and reputation, and a visible auto-apply trail tells the market you have profoundly misread your own position in it. Nothing whispers "do not, under any circumstances, promote this person" quite like being caught spray-applying with a bot.

2. Juniors and career-starters: spend the time the bot saved you on actual humans. Chasing an internship, traineeship or entry-level pharma role in DACH? The temptation to let a bot blanket the market is strongest for you and the payoff weakest — a cruel combination, because your CV is the least differentiated in the pile and a single human champion is worth more to you than to anyone. Pick fifteen target employers. Find one real person at each. Engage with substance, not a connection request reading "I'd love to add you to my professional network," which is the digital equivalent of a handshake from someone reading the instructions for handshakes off their phone. If you need visa sponsorship, confirm which employers actually sponsor before you apply, rather than discovering after three hundred auto-submissions that almost none of them do. A bot will not check. A bot does not care. A bot has never lain awake worrying about a residence permit, and it never will, the lucky thing.

3. Use AI as a drafting assistant, never as a submission agent. Semi-auto, never full-auto, no exceptions, no "just this once." Let the model build the scaffolding and suggest the keywords, then write the human half yourself: the specific, slightly embarrassing, entirely true reason you want this role at this company. It is also the only mode that survives the DACH appetite for a considered Anschreiben. The machine supplies the what; you supply the why. Nobody has yet automated a convincing why, and on the day they do, your CV will be the least of our collective problems and we'll have bigger meetings to attend.

4. Guard your platform like it owes you money and is about to skip town. Before you install any browser extension that acts on your behalf inside LinkedIn, read the User Agreement, all of it, because that account is a real asset and the tool you are about to unleash on it is, by LinkedIn's own crisp definition, contraband.¹⁵ Losing a decade of network to save one afternoon of clicking is not a trade. It is a tragedy, and unlike most tragedies it arrives with a recurring subscription fee and an auto-renewal you will forget to cancel.

5. For the 50-plus reader, ignore with prejudice anyone telling you to "just apply to everything." It is the single worst possible advice for you specifically, and it is dispensed freely and confidently by people who have never once been on the receiving end of the bias you navigate every week. Volume strategies bury your one decisive asset, depth, under a heap of generic submissions that read like everyone else's and erase the very thing that makes you worth hiring. Your edge is the warm network you have built over decades and the framing of your experience as institutional risk management: the hard-won judgement an organisation cannot afford to get wrong, the kind whose absence shows up in a regulatory filing eighteen months later. A bot cannot transmit thirty years of pattern recognition down a fibre-optic cable. A fifteen-minute coffee with a former colleague can do it before the foam settles. Choose the coffee. It is warmer, it is more honest, and it is blessedly, gloriously free.

For the recruiters and hiring managers reading this over someone's shoulder

You are on the receiving end of the deluge, and the EU AI Act is about to convert your defences from a productivity preference into a legal obligation with teeth. Happily, and this is rarer than a clean audit, the measures that actually work are the same ones that march you towards compliance. A free lunch, in regulation, sighted in the wild — photograph it.

Lean hard into structured, skills-based screening with documented, auditable criteria. It cuts through the beige AI paste and satisfies the Act's documentation and oversight demands in one motion.¹³ Feed your referral programme like it's the family business, because it is your highest-converting channel and it is largely immune to the bot flood by design. Source directly rather than waiting for inbound volume you can no longer trust further than you can throw a rack of servers. Keep a human at the decision point, not only because Article 86 will require you to explain significant automated rejections, but because the one thing the candidate's bots cannot counterfeit, no matter how many credits they buy, is a real conversation. And quietly retire "application count" as a health metric and never speak of it again. In 2026, a flood of applications is not proof your role is desirable. It is proof the auto-appliers have found your address and told their friends.

The bottom line

The tools are neither evil nor magic, and anyone selling them as either is selling. They are a lever that faithfully multiplies whatever you feed in, and most people are feeding in a generic CV and precisely zero judgement, multiplying it three hundredfold, and then sitting back, puzzled, as the inbox stays as silent as a held breath. The market is not broken because candidates got lazy. It is strained because the cost of applying collapsed to nearly nothing, and value, reliably, ruthlessly, always slides towards its marginal cost. The way to stay valuable in a market drowning in the free thing is exactly what it has always been: be the scarce thing. Specific, human, introduced, and impossible to mistake for a template that 311 other people also submitted before lunch and forgot about by dinner.

You read all the way to the end of an article arguing against the easy button. That alone vaults you ahead of everyone who let a robot read it for them and filed the summary under "later," somewhere they'll never find it again. Now go and have one real conversation. It will outperform the bot, it will not cost you €65, and it will not, at any point, get your LinkedIn account suspended. Versprochen (promised).

Your Turn

Save this article to your free career-advice library (I'm building one here, instalment by instalment, and you're early) and repost it so it reaches someone whose thumb is hovering, right now, over a €65 "Subscribe" button, blissfully unaware.

Then, because this newsletter runs on the conversations it starts, and a thoughtful comment helps other readers more than a like ever could:

Have you used an AI auto-apply tool? What was your real interview rate? Numbers, not vibes. I am collecting them, lovingly, like a coroner with a clipboard.

Recruiters: can you spot a mass-applied submission, and what gives it away? Asking for the entire profession.

All readers: has a fully AI-generated Bewerbung ever actually helped you, or only the human-edited kind?

50-plus crowd: what has actually got you into the room lately — the application or the relationship? Be honest, we're among friends.

If this was useful, subscribe so the next Bloodbath lands in your feed, and follow the More Than Career page to keep an eye on where this is headed. The real depth, though, lives at morethancareer.de: Career Coaching (job-search strategy, interview preparation and salary-negotiation training with ready-to-use scripts), Reverse Recruitment, and CV and LinkedIn rewrite and rebranding. The human-edited kind. The only kind worth paying for.

Prost. Und bewerbt euch mit Verstand. (Cheers. And apply with your brain switched on.)

Sources & References

[1] SimpleApply.ai — feature description (manual, semi-auto, full-auto modes; AI job scoring): https://simpleapply.ai/ and https://markets.financialcontent.com/kelownadailycourier/article/marketersmedia-2025-7-31-simpleapplyai-announces-revolution-in-job-applications-with-ai-powered-automation

[2] Index.dev — LazyApply volume and pricing (50–300+/day): https://www.index.dev/blog/ai-tools-for-job-seekers

[3] Semafor — Recruiters swamped with AI-generated applications (LinkedIn +45%, ~11,000/min, 1,200 applications to one posting): https://www.semafor.com/article/06/24/2025/recruiters-swamped-with-ai-generated-job-applications

[4] CNBC — Recruiters "drinking through a fire hose" (300–500 applications in three days): https://www.cnbc.com/2025/10/29/recruiters-are-drinking-through-a-fire-hose-of-job-applications-experts-say.html

[5] Resumefast — AI auto-apply tools, documented LazyApply user (5,000 applications, 20 interviews) [vendor source, single anecdote]: https://www.resumefast.io/blog/ai-auto-apply-tools

[6] FastApply blog — Wonsulting bulk auto-apply shutdown, ~2% interview rate [competitor source]: https://blog.fastapply.co/auto-apply-jobs-tools-compared-2026

[7] CO/AI — AI job applications flood LinkedIn; Anthropic reportedly advised applicants not to use LLMs: https://getcoai.com/news/ai-job-applications-flood-linkedin-with-11000-per-minute/

[8] GrackerAI — Adzuna testing of ApplyIQ filter (rejects ~1 in 5, 80% match threshold): https://gracker.ai/blog/ai-job-apply-bots-2025

[9] JobCopilot — AIApply review (applied outside set location/language); ResumeJudge — SimpleApply review (applies to wrong seniority, no control): https://jobcopilot.com/aiapply-review/ and https://resumejudge.com/blog/simpleapply-review/

[10] Amadeus Fire — Bewerbung schreiben mit KI (risk of generic/inaccurate fully-generated applications): https://www.amadeus-fire.de/blog/bewerbung-schreiben-mit-ki-chancen-grenzen-und-der-richtige-umgang

[11] Hays — Vom Prompt zum Traumjob (Stepstone: ~6 in 10 DACH applicants use AI): https://www.hays.de/magazin/think-ahead/zukunft-und-innovation/vom-prompt-zum-traumjob

[12] OnlineMarketing.de — German applicant sentiment on employer AI use (top concern: applications processed only by machine): https://onlinemarketing.de/karriere/human-resources/58-prozent-nutzen-ki-im-bewerbungsprozess

[13] artificialintelligenceact.eu — recruitment AI classified high-risk under Regulation 2024/1689; obligations from 2 August 2026: https://artificialintelligenceact.eu/what-the-act-means-for-staffing-businesses/

[14] DLA Piper GENIE — AI literacy obligations (2 Feb 2026) and proposed Digital AI Omnibus deferral; HeyMilo — EU AI Act recruitment timeline and Article 86: https://knowledge.dlapiper.com/dlapiperknowledge/globalemploymentlatestdevelopments/2026/The-Digital-AI-Omnibus-Proposed-deferral-of-high-risk-AI-obligations-under-the-AI-Act and https://www.heymilo.ai/blog/how-the-eu-ai-act-changes-recruitment-and-what-employers-need-to-know

[15] LinkedIn Help — Prohibited Software and Extensions (User Agreement §8.2 bans bots/automation/scraping; account restriction): https://www.linkedin.com/help/linkedin/answer/a1341387/prohibited-software-and-extensions

[16] SHRM — Majority of employee referrals made during work hours (ERIN study: ~1 in 10 referrals converts to a hire vs 50–60+ applicants/hire on job boards): https://www.shrm.org/topics-tools/news/talent-acquisition/majority-of-employee-referrals-made-during-work-hours

© 2 June 2026 Andreas Schulz. All rights reserved.