---
title: "Fact-check claims, surface hidden assumptions, and separate evidence from spin."
description: "Turn passive reading into active critical thinking. Highlight bold claims in essays, whitepapers, market reports, or news articles and ask Vezz me what the argument depends on."
canonical_url: "https://vezz.me/use-cases/check-claims-and-assumptions-in-articles/"
type: "pseo-use-cases"
category: "use-cases"
topic: "Critical Reading & Research Workflow"
intent: "Best for analysts, journalists, researchers, and careful readers evaluating persuasive essays, vendor benchmarks, and policy arguments."
---

# Fact-check claims, surface hidden assumptions, and separate evidence from spin.

Turn passive reading into active critical thinking. Highlight bold claims in essays, whitepapers, market reports, or news articles and ask Vezz me what the argument depends on.

> **Best fit:** Best for analysts, journalists, researchers, and careful readers evaluating persuasive essays, vendor benchmarks, and policy arguments.

## Live Webpage Walkthrough

- **Source:** Industry Benchmark Whitepaper — Performance Claims (`vendor-blog.example.com/2026-database-benchmark`)
- **Recommended Context Mode:** Detailed

**Highlighted Passage:**
> In our benchmark of 10 million analytical queries, Engine X delivered 8.4x lower p99 latency at half the compute cost of Legacy Warehouse Y.

**User Question (Shortcut):** "What unstated assumptions or benchmark design choices could make this 8.4x claim misleading?"

**Vezz me On-Screen Answer:**
Check the methodology section for these 4 common benchmark levers:
1. Warm cache vs cold storage: Were Engine X's results served from pre-warmed RAM/SSD caches while Warehouse Y ran cold scans?
2. Indexing & tuning asymmetry: Was Engine X tuned with custom sort keys/projections while Warehouse Y used default unindexed tables?
3. Concurrency level: Was p99 measured at 1 concurrent user or 100 concurrent users?
4. Hardware normalization: How was 'half the compute cost' calculated—list on-demand pricing vs reserved instances?

**Suggested Follow-Ups:**
- "Look at the surrounding headings—did the author disclose concurrency or cache state?"
- "What is the strongest steelman case for when Engine X genuinely wins?"
- "Give me 3 questions to ask their sales engineer about this benchmark."

## Custom Prompt Preset for Vezz me

- **Preset Name:** Assumption & Evidence Auditor
- **Recommended Context Level:** Detailed
- **JSON Preset Download:** https://vezz.me/prompts/presets/assumption-and-bias-auditor.json

```text
You are a skeptical research analyst. For the selected passage: 1) Separate stated empirical facts from author inference or rhetoric, 2) List the top 3 unstated assumptions required for this claim to hold true, and 3) State what concrete evidence or control group you would check next.
```

**Why it works:** Trains you to spot base-rate neglect, survivorship bias, and cherry-picked baselines while reading.

## Step-by-step workflow

1. **Highlight the thesis or headline statistic**: Select the claim, chart caption, or causal statement that the article hinges on.
2. **Ask what the claim depends on**: Press Ctrl+Shift+X and run the Assumption & Evidence Auditor prompt.
3. **Generate the steelman counterargument**: Follow up by asking: 'What is the strongest informed counterargument to this paragraph?'

## Separate correlation, causation, and denominator tricks

Persuasive writing frequently switches between percentages and absolute numbers, or treats observational correlation as causal proof.

- Ask what the denominator or baseline period is when a big percentage change is cited
- Check for survivorship bias and selection effects in case studies
- Keep the article text right beside the critique so you can verify whether the author addressed the caveat later on the page

## Build balanced research briefs faster

When writing a strategy memo or literature review, you need both the author's claim and its boundary conditions.

- Use the Counterargument & Steelman prompt to surface the strongest opposing view
- Copy both the claim summary and the key caveat into your research log
- Always verify factual claims against primary datasets and original citations

## Frequently asked questions

### Does Vezz me automatically tell me if a claim is true or false?

Rather than acting as an infallible oracle, Vezz me is best used to surface hidden assumptions, logical leaps, missing baselines, and what primary evidence you should verify next.

### Can I use custom prompts to switch between summarizing and critiquing?

Yes. You can keep both an Executive Summary prompt and an Assumption Auditor prompt saved in Vezz me and switch between them depending on what you are reading.

## Related guides

- [Assumption & bias auditor prompt](https://vezz.me/prompts/assumption-and-bias-auditor/): Download our JSON prompt preset for critical reading.
- [Counterargument & steelman prompt](https://vezz.me/prompts/counterargument-steelman/): Download our JSON prompt preset for generating balanced counter-perspectives.
- [Vezz me for Substack & Medium](https://vezz.me/for/substack-and-medium/): How to read long-form essays and opinion pieces with an AI companion.