Automated Fact-Checking Workflow with n8n and Ollama AI

1. What This Workflow Does

This workflow checks if article sentences match given facts.

It helps editors find mistakes fast and save time.

The system splits text, tests each claim with AI, then reports wrong info clearly.

2. Tools and Services Used

• n8n: To run and automate the workflow.
• Ollama AI’s bespoke-minicheck: Special model for fact-checking claims.
• Ollama AI’s qwen2.5:1.5b: Model for summarizing found mistakes.

3. Inputs, Processing, and Output

Inputs

• facts: Background facts for the article.
• text: Full article content with claims.

Processing Steps

• Splitting article text into sentences keeping dates and lists intact.
• Separating each sentence as a claim item.
• Pairing each sentence with background facts for AI review.
• Using AI to mark sentences as correct or wrong.
• Filtering out false claims for focus.
• Collecting wrong claims for summary.
• Using another AI step to summarize errors and give overall accuracy.

Output

A report showing each wrong statement with a final accuracy grade.

4. Beginner Step-by-Step: How to Use This Workflow in n8n

Download and Import

1. Click the “Download” button on this page to get the workflow file.
2. Inside the n8n editor, choose “Import from File” to load the workflow.

Configure Credentials and Settings

1. Add your Ollama API Key in the Ollama Chat Model nodes.
2. Check and update any email, channel, folder, or table IDs if used for output.
3. Verify the Edit Fields (Set) node has correct facts and text fields or replace with own data.
4. Copy the exact JavaScript code for splitting sentences in the Code node if changes are needed.

Test and Activate

1. Manually trigger the workflow using the Manual Trigger node “When clicking ‘Test workflow’”.
2. Check the output to see if claims are correctly marked and summarized.
3. Turn on the workflow toggle to activate it in production.

Integrate for Reuse

1. Use the When Executed by Another Workflow node as an entry for embedding this fact check inside bigger processes.

Consider using self-host n8n to run this workflow securely if needed.

5. Customizations and Edge Cases

• Change background facts in Edit Fields (Set) node to match other knowledge.
• Switch the AI model by updating the Ollama Chat Model node if a different specialty is needed.
• Modify the JavaScript sentence splitting in the Code node for other languages or special formats.
• Add output options to send results by email, Slack, or save to sheets after the summary step.
• Note: Very large texts may need splitting before processing to avoid limits.

6. Troubleshooting Common Problems

• Problem: “Input text is empty” error.
  Solution: Make sure the input text field is exactly named “text” and contains data in the Edit Fields (Set) or the incoming trigger.
• Problem: AI responses are wrong or garbled.
  Solution: Check prompt format and Ollama API Key in the Ollama Chat Model nodes. Confirm the model “bespoke-minicheck” is installed.
• Problem: Sentences do not split well near dates or lists.
  Solution: Adjust the regular expressions in the Code node JavaScript to cover needed cases.

7. Summary

✓ Save many hours by automating sentence-level fact checks.
✓ Find errors you might miss manually.
✓ Get clear reports listing each incorrect claim.
✓ Easily add this workflow to bigger editorial processes.

→ This workflow reads article text plus background facts.
→ It splits text and checks claims one by one using AI.
→ It filters errors and creates a detailed sum-up.
→ You get a fast, usable fact-check result with less effort.