Last spring, a wine estate manager near Libourne reached out to me with a problem that made me laugh—not unkindly. She was hiring seasonal workers for harvest and had received 187 CV applications for 12 positions. Her process: print them all, highlighter in hand, manually filter by experience, then call candidates one by one to confirm availability. She spent 14 hours screening CVs that could have been handled in 30 minutes.
She didn't have an HR department. She didn't have a hiring budget. What she had was frustration and spreadsheets that contradicted each other.
This is where most SMBs get it wrong about AI and recruitment. They think they need expensive ATS platforms (Applicant Tracking Systems) costing €150–300/month, or they assume they can't automate hiring without a dedicated person managing it. Both beliefs are wrong.
Here's what I've seen across 40+ SMBs, restaurants, hotels, and craft businesses: recruitment doesn't fail because you get too few applications. It fails because you can't process the ones you do get.
A hotel manager in Bordeaux told me she was screening 60 applications for three front-desk positions. Her criteria were straightforward: 2+ years experience, bilingual French-English, available to work weekends. But instead of filtering, she was reading full CVs, making judgement calls, losing context between candidates, then realizing three days later she'd already rejected someone perfect because she'd forgotten their details.
She was spending 6 hours per hiring cycle. She hired four times a year. Do the maths: 24 hours annually wasted on a task that should take 90 minutes.
The problem isn't technology. It's workflow. You need to separate signal from noise before a human ever touches the CV.
For the wine estate, I built a simple automation using Make (the no-code platform I prefer over Zapier for this specific use case—better document parsing, more reliable PDF handling). Here's what actually happened:
Step 1: Centralized intake. Applications came to a single email address. Make automatically captured attachments, extracted text from PDFs, and stored candidate data in Airtable.
Step 2: Automated scoring. I connected Claude AI (via Make's built-in integration) to read each CV and score candidates against pre-defined criteria: years of vineyard experience, certification status, language skills, previous harvest work. Claude generated a structured score: 0–100 points. No more subjective filtering.
Step 3: Threshold routing. Candidates scoring 70+ were automatically moved to a "Call" column in Airtable. Candidates scoring 40–69 went to "Maybe—human review." Below 40: rejected, but with an automated email thanking them (because you're not a monster).
Step 4: Prep for humans. For candidates who passed the threshold, Make generated a one-page summary for the hiring manager: key qualifications highlighted, gaps noted, interview questions auto-populated based on CV content.
Total setup time: 3 hours. Total monthly cost: €0 (Airtable free tier, Make's basic plan €10, Claude API costs less than €2 per hiring cycle).
Result: She went from 14 hours screening to 1.5 hours conducting actual interviews. The 12 positions filled in 8 days instead of 21.
You don't need fancy software. You need filtering that's consistent and fast. The wine estate didn't care about a beautiful ATS dashboard—she cared about not losing a good candidate because she was tired of reading.
A pastry shop in Saint-Émilion implemented the same system for apprentice hiring. A hotel chain across Gironde adopted it for seasonal staff. A restaurant in Libourne uses it for kitchen positions.
They all say the same thing: "We finally know who we're actually interviewing."
Don't ask Claude (or ChatGPT) to decide if someone is a good fit. Ask it to score candidates against your criteria. You decide what matters. The AI just applies it consistently. That's where the safety is—and where the legal clarity is.
If you're hiring for a kitchen role and you weight "demonstrated leadership" at 30%, that's your call. The system just ensures every CV is measured the same way.
Once screening is automated, consider adding interview scheduling (Make + Calendly integration), reference check automation (email templates with scoring rubrics), or offer letter generation in Airtable. But start with screening. That's where the time leak actually is.
You don't need an HR department. You need a process that works without one.
If you're hiring regularly and drowning in applications, book a free 1-hour audit. I'll show you exactly where you're losing time and what a no-code system would cost to build for your specific hiring cycle.
```Prêt à automatiser votre business ?
Audit gratuit 1h — on identifie ensemble les 3 automatisations les plus rentables pour vous.
Réserver l'audit gratuit