Recruitment screening and interview preparation automation

Recruitment teams waste enormous amounts of time on screening. A hiring manager posts a role on Monday morning, and by Tuesday they're drowning in 200 applications. Most get a glance before rejection. The ones that make it through to interview prep consume hours of preparation work: reviewing CVs, researching candidate backgrounds, prepping interview questions, and compiling feedback forms.

What if you could automate 80% of this friction? This workflow combines three focused AI tools to screen applications at scale, prepare interview briefs automatically, and generate structured feedback templates, all without anyone manually moving information between systems.

The goal is simple: a candidate submits an application, and within minutes you have a screening decision, an interviewer brief document, and pre-built interview notes waiting in your system. No copy-pasting. No manual handoffs. Just data flowing through your recruitment pipeline at the speed of API calls.

The Automated Workflow

This workflow uses four key stages: application intake, CV analysis and screening, interview brief generation, and feedback template creation. The orchestration happens in either Zapier, n8n, or Make (Integromat), depending on your preference and technical comfort....

Why these tools?

Cogram records meetings and generates summaries with structured output, useful for capturing interview decisions. Hyperwrite provides fast text generation and content creation for briefs and frameworks. Micro1 AI is a lightweight tool for document classification and structured extraction. Together, they handle the entire recruitment pipeline without expensive dedicated HR software. For more on this, see Turning Meeting Notes Into Action: Tools for Automating P....

Stage 1:

Application Intake and Trigger

Your workflow starts with an application submission. This could come from a form, email, or your existing ATS webhook. For this example, we'll use a standard form submission via a webhook.

Set up the trigger in your orchestration tool:

Copy
POST /webhook/application-received
Content-Type: application/json

{
  "candidate_name": "Alice Johnson",
  "email": "alice@example.com",
  "role": "Senior Product Manager",
  "cv_url": "https://storage.example.com/alice-cv.pdf",
  "cover_letter": "I'm excited about this opportunity...",
  "submission_date": "2024-01-15T10:30:00Z"
}

In Zapier, this becomes your first step: a Webhook trigger that catches incoming applications. In n8n, you'd set up a Webhook node in "Listen" mode. In Make, use the HTTP module to receive the POST request.

The webhook becomes your single entry point. Everything downstream waits for this signal.

Stage 2:

CV Extraction and Screening with Micro1 AI

Once an application arrives, you need to extract structured data from the CV and assess fit against your job requirements. Micro1 AI is designed for exactly this task: document classification and field extraction.

API call to Micro1 AI:

Copy
POST https://api.micro1.ai/v1/extract
Authorization: Bearer YOUR_MICRO1_API_KEY
Content-Type: application/json

{
  "document_url": "https://storage.example.com/alice-cv.pdf",
  "extraction_template": {
    "fields": [
      "full_name",
      "email",
      "phone",
      "years_experience",
      "current_role",
      "key_skills",
      "education",
      "previous_companies"
    ]
  },
  "classification": {
    "check_against": "Senior Product Manager role requirements",
    "output_scores": true
  }
}

Micro1 returns structured JSON with extracted fields and a relevance score. If the score is below your threshold (say 60%), the workflow can automatically reject the candidate and send a rejection email. If they pass, move to stage three.

Example Micro1 response:

Copy{
  "extracted_data": {
    "full_name": "Alice Johnson",
    "years_experience": 7,
    "current_role": "Product Manager at TechCorp",
    "key_skills": ["product strategy", "user research", "cross-functional leadership", "data analysis"],
    "education": "MBA from LSE"
  },
  "classification": {
    "relevance_score": 0.82,
    "decision": "pass",
    "reasoning": "Strong relevant experience and skills match for senior PM role"
  }
}

Store this output in a database or spreadsheet for later reference. In n8n, use a PostgreSQL node. In Make, use a Google Sheets module.

Stage 3:

Interview Brief Generation with Hyperwrite

The candidates who pass screening now need an interviewer brief. This is where Hyperwrite comes in. You'll send the extracted CV data and role details to Hyperwrite, and it generates a structured brief in seconds.

API call to Hyperwrite:

Copy
POST https://api.hyperwrite.com/v1/generate
Authorization: Bearer YOUR_HYPERWRITE_API_KEY
Content-Type: application/json

{
  "prompt": "Generate an interview brief for the following candidate applying for Senior Product Manager:\n\nCandidate Name: Alice Johnson\nYears of Experience: 7\nCurrent Role: Product Manager at TechCorp\nKey Skills: product strategy, user research, cross-functional leadership, data analysis\nEducation: MBA from LSE\n\nInclude:\n1. A brief profile summary (2-3 sentences)\n2. Key areas to explore in the interview\n3. Potential strengths to validate\n4. Red flags or gaps to address\n5. Suggested interview questions (5 questions)\n\nFormat as markdown.",
  "max_tokens": 1200,
  "temperature": 0.7
}

Hyperwrite generates a formatted brief. You now have a document that an interviewer can read in two minutes and be prepared.

Typical Hyperwrite output:

Copy## Profile Summary
Alice is a mid-to-senior PM with 7 years of experience and relevant product strategy expertise. Her MBA and track record at TechCorp suggest strong analytical and leadership capabilities for this senior role.

## Key Areas to Explore
- Experience scaling products from 0-1 and 1-N
- Approach to cross-functional stakeholder management
- Data-driven decision making in ambiguous situations
- Vision for product roadmapping in competitive markets

## Potential Strengths to Validate
- User research methodologies and incorporation into strategy
- Leadership of technical teams without engineering background
- Demonstrated product impact through metrics

## Red Flags or Gaps
- Need to understand reasons for staying 4 years at one company
- Limited public evidence of B2B experience (confirm if needed for role)

## Suggested Interview Questions
1. Walk me through a product launch where you had to balance stakeholder demands.
2. Describe your user research process and how it shaped product decisions.
3. Tell me about a time you disagreed with engineering on priorities.
4. How do you measure product success?
5. What attracted you to this role from your current position?

Save this brief to a shared document storage, a Slack channel, or your ATS system. In n8n, use a Google Docs node. In Zapier, use a Zapier Document step or send to Slack.

Stage 4:

Interview Feedback Templates with Cogram

Before the interview happens, generate a structured feedback template that interviewers will complete after meeting the candidate. Cogram is useful here because it can create templates that include prompts for the kinds of information you want to capture.

API call to Cogram to generate a feedback template:

Copy
POST https://api.cogram.com/v1/generate_template
Authorization: Bearer YOUR_COGRAM_API_KEY
Content-Type: application/json

{
  "template_type": "interview_feedback",
  "candidate_name": "Alice Johnson",
  "role": "Senior Product Manager",
  "interview_type": "first_round",
  "fields_to_include": [
    "overall_impression",
    "technical_competency_score",
    "communication_score",
    "cultural_fit_score",
    "key_strengths",
    "areas_of_concern",
    "would_recommend_next_round",
    "notes"
  ],
  "scoring_scale": "1-5"
}

Cogram returns a structured feedback form. Deploy this as a Google Form, Typeform, or a custom form in your system.

Example template structure:

Copy{
  "template_id": "feedback_alice_johnson_2024",
  "candidate": "Alice Johnson",
  "role": "Senior Product Manager",
  "interview_date": "2024-01-16",
  "interviewer_name": "required field",
  "sections": {
    "overall_impression": {
      "type": "text",
      "prompt": "What was your initial impression of the candidate?"
    },
    "technical_competency": {
      "type": "rating",
      "scale": "1-5",
      "prompt": "How well did the candidate demonstrate product management knowledge?"
    },
    "communication": {
      "type": "rating",
      "scale": "1-5",
      "prompt": "How clearly did they articulate their thinking?"
    },
    "cultural_fit": {
      "type": "rating",
      "scale": "1-5",
      "prompt": "Do they align with our company values?"
    },
    "recommendation": {
      "type": "multiple_choice",
      "options": ["Strong yes", "Yes", "Maybe", "No", "Strong no"]
    }
  }
}

This template is now ready before the interview even starts.

Complete Workflow in n8n (Example)

Here's how you'd structure the complete workflow in n8n. The visual builder makes this straightforward:

Copy
1. Webhook (incoming application)
   ↓
2. HTTP Request to Micro1 AI (CV extraction)
   ↓
3. Conditional: If score > 60%, continue; else send rejection email
   ↓
4. HTTP Request to Hyperwrite (generate interview brief)
   ↓
5. HTTP Request to Cogram (generate feedback template)
   ↓
6. Google Docs (save brief with timestamp)
   ↓
7. Google Sheets (log candidate data and decision)
   ↓
8. Slack notification (alert hiring manager)
   ↓
9. Email to candidate (interview scheduled, brief attached to interviewer)

In n8n's JSON view, the key glue is passing the extracted_data from step 2 as context into steps 4 and 5. Each API call references the previous step's output using {{ $json.extracted_data }} syntax.

Orchestration Tool Recommendation

For this specific workflow:

Zapier: Best if you want a fully visual interface and prefer not to write code. Limited by the 100-task free tier, so this becomes expensive at scale. Good for prototyping.

n8n: Best if you want self-hosted flexibility and powerful conditional logic. Steeper learning curve, but once set up, it's reliable and free to host yourself. Recommended for teams planning to scale.

Make (Integromat): Good middle ground between Zapier and n8n. Visual builder with reasonable free tier. Excellent scenario templates and better pricing than Zapier at scale.

Claude Code: Not an orchestration tool per se, but if you're comfortable with Python, Claude Code can write automation scripts that call all three APIs in sequence and handle error handling. Less reliable for production without additional infrastructure, but powerful for rapid iteration.

The Manual Alternative

If you want to keep things simple and not set up an orchestration workflow, the tools work independently:

1. When an application arrives, open the CV and paste it into Micro1 AI's web interface. Review the extracted data and score.

2. If they pass, copy the extracted information into a Hyperwrite prompt and generate the interview brief manually.

3. Create a feedback form in Google Forms and share it with interviewers.

This takes 10-15 minutes per candidate instead of 2 minutes automated. At 50 applications per month, that's 7-12 hours of manual work per month. The automation saves that entirely.

Pro Tips

1. Handle PDF Extraction Failures Gracefully

Some CVs are images or scanned PDFs. Micro1 AI sometimes struggles with these. Build in a fallback: if extraction confidence is below 40%, route the application to a manual screening queue with an alert to your hiring team. Don't reject automatically; just flag it.

Copy{
  "if": "extraction_confidence < 0.4",
  "then": "send_to_manual_queue",
  "webhook_alert": "admin@company.com"
}

2. Rate Limit Hyperwrite Requests

Hyperwrite has rate limits (typically 60 requests per minute on paid plans). If you get a burst of applications, you'll hit limits. Use a delay node in your orchestration tool between Hyperwrite calls, or batch process briefs in small groups with 1-second delays between requests.

3. Cache Role Descriptions

Don't re-send the full job description to Micro1 AI and Hyperwrite every time. Store it once in a database and reference it by ID. This cuts API payload size and speeds up requests.

4. Monitor Screening Decisions for Bias

Log all screening decisions and the scores that led to them. Review rejection patterns monthly. If you're rejecting 95% of women or candidates from certain schools at disproportionate rates, your thresholds may need adjustment. This is both an ethics and legal issue.

5. Store Interview Briefs in Version Control

Save generated briefs with timestamps. If a candidate reapplies in 6 months, you have historical context. Use Google Docs or a database, not scattered emails.

Cost Breakdown

If you're processing 500+ applications per month, the per-unit cost is roughly £0.20-0.30 per candidate screened and prepped. Compare this to 15 minutes of a £40/hour recruiter's time (£10 in labour per candidate), and the automation breaks even immediately while freeing your team for higher-value work like phone screens and closing offers.