# Check a claim against its source before it becomes a post > Source: https://tareno.co/agents/use-cases/check-facts-before-posting > Language: en > Use an AI agent to compare a press summary, a budget or an event listing with its primary source, then draft a measured post with the source named. Approval stays in Tareno. ## Short answer **Can an AI agent check a claim against its source before I post it?** Yes. Give the agent the claim and the primary source, and it reports where they differ before any text is drafted. It does not decide which source is right and it cannot judge whether the underlying science or policy is sound, so a person reads the post and approves it in Tareno. Speed is the enemy of accuracy on social media: the post that appears first gets the replies, and the one that corrects itself later gets the screenshot. The cure is a short, boring step before drafting, in which an agent compares what you want to say with what the primary source says. The step looks the same for a local newsroom, a tourist office and a university department. This page lists the guides that build it into a post, and the habits that keep it honest. ## The same job in four industries - **Local news:** An editor checks a vote count against the council's minutes and finds the article and the minutes disagree before the post goes out. - **Tourism:** A regional office verifies every event date and opening hour against the organizer's official page and pins only what holds up. - **Academia:** A department compares a press summary with the abstract and limitations of the paper and posts the measured version. - **Civic data:** A volunteer compares a city's budget summary with the full document and explains the difference in a thread with page numbers. ## How it works 1. Name the primary source: the document, the official page or the abstract. The summary is the claim to be checked. 2. Ask for a table: each claim, the supporting passage, and whether the summary goes further than the source. 3. Resolve differences yourself. The agent reports; the editor decides which source the account follows. 4. Draft the post at the strength of the source and name the source in the text, then have the person closest to the material read it. ## Guides that put it to work - [Can Grok Bot write Facebook Page posts that link to your local news stories?](https://tareno.co/agents/grok-bot/facebook): Checks a local story against the council's minutes before a Page post. - [Can Grok Bot check a match result before your supporters' club posts it on X?](https://tareno.co/agents/grok-bot/x): Verifies a match result against the league's page before the report. - [Can Grok Bot check a city budget against its document and write a Bluesky thread?](https://tareno.co/agents/grok-bot/bluesky): Compares a budget summary with the full document and writes a thread with page references. - [Can Grok Bot check a university press summary against the paper before a Mastodon post?](https://tareno.co/agents/grok-bot/mastodon): Compares a university press summary with the paper's abstract. - [Can Grok Bot turn a regional event calendar into verified seasonal Pinterest pins?](https://tareno.co/agents/grok-bot/pinterest): Verifies event dates for seasonal pins. - [Can Grok Bot check a claim that keeps appearing in your Threads replies?](https://tareno.co/agents/grok-bot/threads): Checks a claim that keeps appearing in replies against named sources. ## What to watch out for - Without a named source, an agent checks against general knowledge, which is what you are trying to avoid. - A source that cannot be reached is a gap, not a pass. Ask the agent to say which sources it could not read. - Do not let a post describe motives. State what the documents show. ## Questions ### Which agents suit fact-checking work? Any agent that can read documents and compare texts. Grok Bot is the focus of several guides here because of its source-checking, but the workflow is the same with others: name the source, ask for a comparison table, decide, then draft. ### Can the agent say whether a result is true? No. It can say whether a summary matches a paper or a document. Whether the paper is right is a question for the authors and the field, which is why an author or an expert reads the final post. ### Why name the source inside the post? Because readers can then verify it, and because it keeps the post at the strength of the source. A post that names its source is easier to correct honestly if the source is later updated. ### Does this slow posting down? By a few minutes per post, and it removes the correction posts that cost more. The table also becomes a record of what was checked, which matters if someone asks later. ## Related - [Grok Bot × Tareno](https://tareno.co/agents/grok-bot) - [Approval workflows](https://tareno.co/features/approval-workflows) - [Content boards](https://tareno.co/features/content-boards) - [Read the replies to a post and turn them into a decision](https://tareno.co/agents/use-cases/read-replies-and-triage) - [Publish photos and slides with image descriptions that are checked](https://tareno.co/agents/use-cases/photos-with-image-descriptions) - [Generate posts from a code repository without publishing a secret](https://tareno.co/agents/use-cases/posts-from-your-code-repository) Approval model: Tareno saves drafts directly. Scheduling, publishing, edits to scheduled posts and deletions become pending actions that a person approves in the Tareno dashboard; no agent can approve them.