# Analyze Hundreds of FDA Observations Against Your Own in Minutes

Source: https://www.keypedia.com/media/analyze-hundreds-of-fda-observations-against-your-own-in-minutes
Type: tutorial
Published: August 4, 2026
Updated: August 13, 2026
Authors: Jacob Zheng, KeyPedia Agent

> Benchmark hundreds of FDA observations against agency reference data in the time it takes to read one warning letter, with KeyPedia Predictive and Flagship.

---

[Pattern Match Observations 3 Simple Steps.mp4](https://storage.googleapis.com/gks-blog-videos/blog-videos/6cb01a0710f141c08af5a8a5574f324a.mp4)

KeyPedia turns a stack of FDA observations into a clear read on your own risk. Benchmark hundreds of observations against agency reference data in the time it takes to read one warning letter.

This guide covers two ways to do it: the dashboard, and KeyPedia's AI.

> **Before you start:** you need an active KeyPedia seat, your internal observations in text form, and a few minutes.

## Method 1: Dashboard search

Best when you want fast, visual benchmarking across FDA and other agency records.

**1.** Open the KeyPedia dashboard and enter your observation terms. You can stack several at once, and apply a date range to bound the set.

<center>

![keypedia-obs-search.png](https://storage.googleapis.com/gks-blog-images/blog-images/488b91b8bdca42fa932a9804ca9dd0d5.png)


</center>

**2.** Run the search. KeyPedia returns matching records across FDA and other agency reference data, sorted so you can see what is closest to your finding.

<center>

![keypedia-obs-results.png](https://storage.googleapis.com/gks-blog-images/blog-images/686f881b54a849fd94d6972a22fa7fcb.png)



</center>

**3.** Open Statistics for the aggregate view: total records, companies, and how they break down. That is your reference point for whether your own observation rate is normal.

## Method 2: AI distillation

Best when you want the agent to reason across your observations and the reference corpus rather than scan tables yourself.

**1.** Open a new chat and paste your observation text. The chat takes pasted text, not file attachments, so work in batches you can paste.

![screenshot_chat_paste.png](https://storage.googleapis.com/gks-blog-images/blog-images/de4620a7a0e84e8382a3a72ede1958c4.png)

**2.** Ask it to compare your observations against FDA and other agency reference data.

**3.** Get a side-by-side read on where your observations align with what the agency has cited elsewhere, where they diverge, and where they signal risk.

<center>

![screenshot_chat_output.png](https://storage.googleapis.com/gks-blog-images/blog-images/6e1f9c0fa24942b9965c7681eb18af3d.png)

</center>

> See the video at the top of this guide for the full walkthrough.

> **Pro tip:** benchmarking a supplier list instead of observation text? Use **Bulk add companies** to paste a list or drop a CSV. It takes company names, FEI numbers, Brazil CNPJ, or Canada Business Numbers, and detects the format for you.

<center>

![screenshot_bulk_add.png](https://storage.googleapis.com/gks-blog-images/blog-images/c1d20c94a5694e68aa6b4efb379f7641.png)

</center>

## What to do next

Once you've obtained the observations, you can download them, or save the search for future datasets.

If you want to keep the results, [download the records directly from KeyPedia Library](https://www.keypedia.com/media/how-to-download-gks-data). If you want the search to run again as new filings come in, [set it up as a saved search and alert](https://www.keypedia.com/media/save-a-keypedia-alert-for-artificial-intelligence).

---

![banner-analyze-observations-v2.png](https://storage.googleapis.com/gks-blog-images/blog-images/aaee8bc8369745ca97189ee417a8ba8d.png)




---

Ready to run your own observations? [Start in KeyPedia](https://www.keypedia.com) or [book a demo](https://www.keypedia.com/request-demo).
