Come Creare un Flusso di Screening dei Candidati AI in WordPress

Hiring gets messy when every job application lands in the same inbox. A candidate submits a form. HR opens the entry, checks the role, reads the work experience, reviews the resume, writes notes, updates a spreadsheet, and sends a reply. That works for a few applicants. Once the number grows, good candidates can sit unnoticed while the team repeats the same review steps again and again.

AI can help screen applications faster, but it should not make the final hiring decision by itself. A job application has context. A resume may be incomplete. A candidate may have strong experience but a weak written answer. That is why the better workflow is not “AI rejects candidates automatically.” The better workflow is AI reviews the application, prepares a score and summary, sends it to HR for approval, and only then sends the right email to the candidate.

What This No-Code AI Job Screening Workflow Does

A no-code AI job application screening workflow automates the first-stage review process. Instead of checking every form entry manually, HR gets a clean summary of the candidate, their role, experience, AI score, and CV link. HR can then approve or disapprove the candidate from the review email.

Here is the full workflow:

Passo	Node	Cosa fa
1	Invio modulo riuscito	Starts the workflow when a candidate submits the job application form
2	Agente AI	Reviews the form answers and returns a structured screening result
3	Parser JSON	Converts the AI response into usable fields
4	Gmail: l'intervento umano nel processo	Sends the candidate profile to HR and waits for approval
5	Router	Splits the workflow after the HR review step
6	Condizione	Checks whether HR approved or disapproved the candidate
7	Google Sheets Add Row	Saves the application record
8	Gmail Send Email	Sends the right email to the candidate

This is useful for HR teams, small businesses, agencies, and WordPress site owners who collect job applications through a form.

Cosa ti serve prima di creare il flusso di lavoro

Before building the workflow, make sure these tools are installed, activated, and connected.

• A supported form plugin: Install and activate a form plugin supported by Bit Flows. You can use Bit Form, Contact Form 7, WPForms, or Elementor Form. The form should collect the candidate’s name, email, position, experience, CV, and answer.
• Bit Flows plugin: Install and activate Bit Flows on your WordPress site. 
• AI model: You need an API key from OpenAI or another supported AI provider, such as DeepSeek, Claude, or Gemini.
• Gmail account access: You need a Gmail account connected with Bit Flows so the workflow can send the HR review email and candidate follow-up emails. Follow the Bit Flows Gmail integration guide to create the Gmail connection.
• Google Sheets access: You need a Google Sheet connected with Bit Flows so the workflow can save candidate records. Follow the Bit Flows Integrazione di Google Sheets guide to connect Google Sheets and use the Add Row action.

To install a plugin in WordPress, go to Plugin → Aggiungi nuovo, search for the plugin, click Install Now, and then click Attivare. If you have a ZIP file, use the Upload Plugin to upload the file, then install and activate it.

Step 01: Create the Job Application Form

Create a job application form using your form builder. Here, I have used the Bit Form plugin and built the form in 4 steps.

• Step 1: Personal Information
• Step 2: Work Experience
• Step 3: Resume Upload
• Step 4: Signature

Set the resume upload field to accept PDF, DOC, or DOCX files. Keep the file size limit reasonable, such as 5 MB or 10 MB. Add this confirmation message after submission:

Step 02: Add the Form Submit Success Trigger

Create a new flow and name it: No-Code AI Job Application Screening Workflow. Add the Bit Form node: Trigger – Submit Success.

Cliccare Capture Response, then submit a test application.

Real sample data helps you map the fields correctly in later nodes.

Step 03: Add and Configure the AI Agent

After the form trigger, add the Agente AI node. This step reviews the candidate’s submitted form data and creates a screening result for HR. First, configure the Modello di chat. Add your OpenAI, DeepSeek, Claude, Gemini, or other supported AI model API key in the connection settings. 

Then select the model you want to use for the screening.

Dopo, aggiungi Memoria semplice. Use a unique key such as the candidate email, form entry ID, or submission ID. Keep the Context Window Length around 5 so the AI Agent can keep the current form data, role requirements, and scoring instructions together while processing the application.

In the AI Agent settings, use:

• Source for Prompt: Define Prompt
• Response Format: JSON Object

Nota: Update the AI prompt based on your own job description, required skills, experience level, and hiring criteria before testing the workflow.

Use this short system prompt:

You are an HR screening assistant.
Review only the submitted job application data. Do not invent skills, experience, education, company names, achievements, or certifications.
Do not make the final hiring decision. Do not reject any candidate automatically.
Score the candidate only as an internal HR review signal. Return valid JSON only. Do not include markdown, explanation, or extra text outside the JSON.Copia

Use this in the main Scoring Rules prompt:

Scoring Rules:
- Role skill match: 40 points
- Relevant work experience: 20 points
- Responsibilities quality: 15 points
- Candidate answer quality: 10 points
- Communication clarity: 10 points
- CV or proof availability: 5 points

Recommendation Rules:
- 80 to 100: Shortlist
- 60 to 79: Manual Review
- 0 to 59: Not Matched

Return only valid JSON in this format:

{
  "candidate_summary": "",
  "matched_skills": [],
  "missing_skills": [],
  "experience_note": "",
  "score": 0,
  "recommendation": "",
  "hr_note": "",
  "cv_review_note": ""
}Copia

This setup is important because the next step, JSON Parser, needs a clean JSON response. If the AI adds extra text before or after the JSON, the parser may not read the result correctly.

Step 04: Parse the AI Result with JSON Parser

Add JSON Parser after the AI agent node: Set the JSON input source to the AI Agent response. Here is the JSON result:

The JSON Parser turns the AI Agent’s JSON response into separate usable fields, such as candidate summary, score, recommendation, and HR note, so they can be mapped in the next workflow steps. 

Step 05: Send the application to HR with Human in the Loop

In Bit Flows, add the next node: Gmail → Send and Wait for Response / HITL.

If Gmail is not connected yet, click Aggiungi connessione, aggiungi il Google Client ID e Segreto del cliente from Console Google Cloud, authorize the account, then select that Gmail connection for this step. For detailed setup, follow the official Bit Flows Gmail integration guide.

Now, map the HR email, subject, and review message fields. This node sends the candidate profile to HR and pauses the workflow until HR clicks a response button. Use this configuration:

Keep the HR email focused. HR does not need every technical field from the flow. They need the candidate, role, score, summary, notes, and CV link.

Add the Router After Human in the Loop

The Router separates the Human in the Loop result from other workflow paths.

Create at least two paths:

• Path 1: HR Decision Path
• Path 2: Google Sheets Record Path

You can also keep Google Sheets after the approval result if you only want to save the final-reviewed applications.

Step 06: Create the Google Sheet

Now, create a Google Sheet named: Job Applications – AI Screening

Before saving the application record, select your Google Sheets connection, choose the Job Applications – AI Screening spreadsheet, select the worksheet, and map each candidate value to the correct column. For more help, follow the official Bit Flows Google Sheets integration guide. 

Use a simple sheet first. You do not need too many columns while testing.

Save the record in Google Sheets

Step 07: Add Conditions for Approved and Disapproved Status

Add a Condition node after the Human in the Loop path.

Condition 1 (approved):

• Field: status from Gmail Human in the Loop
• Operator: Equal to
• Value: approved

Condition 2 (disapproved):

• Field: status from Gmail Human in the Loop
• Operator: Equal to
• Value: disapproved

Check the exact value your Human in the Loop node returns before setting these. Copy the value directly from the node output rather than typing it.

Step 08: Send the Approved Candidate Email

On the approved condition path, add: App: Gmail, Event: Send an Email

Do not include the AI score in this email. The score is an internal review signal.

Step 09: Send the Disapproved Candidate Email

On the disapproved condition path, add:

Send this email only after HR disapproves of the application from the Human in the Loop step. Do not let AI send rejection emails automatically.

Step 10: Test the Full Workflow

After setting up the workflow, test it with different candidate types. This helps you confirm that the AI score, HR approval step, Google Sheets row, and candidate emails are working correctly.

Test the workflow with three types of candidates.

Test Candidate	Expected AI Result	Expected HR Action	Expected Email
Strong Front End Developer	Score 80+ and Shortlist	HR approves	Shortlisted email
Average candidate	Score 60 to 79 and Manual Review	HR decides manually	Depends on HR decision
Weak role match	Score below 60 and Not Matched	HR reviews before rejection	Disapproved email only if HR rejects

First, test with a strong candidate, review the AI screening result, and approve or decline it from the Human in the Loop email.

After HR approves or declines, check that the workflow sends the correct candidate email, adds the record to Google Sheets, and shows a successful run in the logs.

This test proves that AI prepares the screening result, HR makes the final decision, and the workflow sends the right email automatically. 

Can AI Read the Uploaded Resume File?

AI can analyze a resume only if the workflow sends the actual resume content to the AI model or to a document-processing API.

In many form workflows, the uploaded CV field gives you a file URL, file path, or file name. That does not always mean the AI Agent has read the PDF or DOC file. If your AI prompt only includes the file name, the AI should not claim it reviewed the resume.

OpenAI supports file inputs through base64 data, file IDs, or external URLs in the Responses API. Gemini can process PDF documents within its document processing limits. That means a deeper CV analysis workflow is possible, but you need the extra file-processing step.

Errori comuni da evitare

Avoid letting AI make the final hiring decision or send rejection emails automatically. Always use Human in the Loop before the final candidate email. Make sure the AI response is parsed with JSON Parser, the approval status matches the exact value returned by the review step, and the AI score stays internal. Also, do not assume the CV was reviewed if the workflow only receives a file name or URL. Keep the Google Sheet simple at first and check workflow logs after every test run.

Conclusion: A Smarter Way to Review Job Applications

A no-code AI job application screening workflow works best when AI handles the repetitive review work, and HR controls the final decision.

Start with one role, one form, one Google Sheet, and two candidate emails. Test approved and disapproved paths carefully. Once the workflow works, add more roles, better scoring rules, resume text extraction, and interview scheduling.

Bit Flows fits this kind of workflow because it can connect the form submission, AI Agent, JSON Parser, Human in the Loop, Google Sheets, Gmail, Router, Conditions, and logs inside one visual workflow.

Build the first version with human approval first. Then, improve the resume analysis step once the core hiring flow runs correctly.

Domande frequenti