Run fast and reliable fact checks on any draft before it goes live. Editors and marketing teams can compare each sentence against a trusted source of facts and get a clear report of issues to fix.

The flow starts by receiving two inputs, the facts and the text to review, either by a manual test or by another workflow. A Code node safely splits the text into sentences and throws an error if the text is empty. Each sentence is paired with the facts and sent to an Ollama model that is built for claim checking. The results are merged by position, filtered to keep only problems, and aggregated into one list. A second Ollama model then writes a short summary that groups issues by type so the team can act fast.

Setup is simple if you already run Ollama. Pull the models and enter your Ollama credential in n8n. Pass in a clean facts document and the draft text. Expect faster reviews, fewer errors, and a repeatable check you can call from any content workflow. Use it for blog posts, product pages, sales decks, and internal briefs.

• Dual entry points with a manual test trigger and an execute by another workflow trigger
• Code node validates input and splits the draft into sentences safely
• Split Out node processes each sentence as a separate claim
• Merge nodes align each claim with the shared facts for accurate checks
• Ollama bespoke minicheck model evaluates each claim against the facts
• Filter keeps only sentences marked as incorrect for a focused review
• Aggregate collects all flagged items into a single payload
• Second LLM step drafts a short summary of issues using an Ollama model
• Designed to return either a summary or raw results for agent workflows

• Reduce manual review from 60 minutes to 10 minutes per article
• Flag incorrect claims at the sentence level to cut rework by 50 percent
• Summarize issues into clear categories so editors can fix problems faster
• Reuse the same check from other n8n flows to standardize content QA
• Scale to long drafts by splitting text into sentences and processing in parallel

1. Import the template into n8n: Create a new workflow in n8n > Click the three dots menu > Select 'Import from File' > Choose the downloaded JSON file.
2. You'll need accounts with Ollama. See the Tools Required section above for links to create accounts with these services.
3. Install and run Ollama on your machine or server. Make sure the API is reachable at a URL like http://localhost:11434.
4. Open a terminal and pull the required models: run ollama pull bespoke-minicheck and ollama pull qwen2.5:1.5b.
5. In n8n, double click the Ollama Chat Model node. In the Credential to connect with dropdown, click Create new credential and follow the on screen steps to connect your Ollama instance.
6. Repeat the credential step for the Ollama Model node so both LLM nodes use a valid Ollama credential.
7. Open the When Executed by Another Workflow node and confirm the workflow inputs include facts and text. These two fields will be passed in when called by other flows.
8. For testing, open the Set node and place sample facts text in the facts field. Use the Manual Trigger to run a test.
9. If the run fails at the Code node, check that the text field is present. The Code node throws an error when text is empty.
10. Review the results at the Filter and Aggregate nodes. If nothing is flagged, confirm the model outputs the expected yes or no signal and adjust the Filter condition if needed.
11. Check the final summary at the Basic LLM Chain node. If it is too long or not needed, remove the summary step and return the aggregated raw results as noted in the sticky note.