• AI Academy
  • Posts
  • 📚 Analyze Docs for Insights with AI

📚 Analyze Docs for Insights with AI

How we use Claude's projects feature to improve our Newsletter

Reading Time: 5 minutes

Hello AI Enthusiast,

In this edition, we're exploring how to use Claude's projects feature. We'll show you how we used it to analyze our newsletters and keep consistency across multiple issues over time.

What are Claude's Projects?

Claude's Projects are like digital file cabinets for long-term tasks. They allow you to organize, analyze, and collaborate with your team on large amounts of information within a dedicated space. You can upload documents, create custom instructions, and engage in ongoing conversations with Claude about your project's content. The feature is only available to paying accounts.

The Problem

We found it challenging to consistently analyze our newsletters. Spotting subtle shifts in tone, audience focus, or topic trends across multiple issues was time-consuming and prone to oversight. We needed a more efficient and objective way to track these changes over time.

How We Did It: A Step-by-Step Guide

Here's how we used Claude's projects feature to create our Newsletter Analyzer:

Step 1: We Set Up the Project

We created a new project in Claude by navigating to the Projects section and clicking "Create Project".

Claude's Projects section

Claude's Project creation button

Step 2: We Wrote a Project Description

We provided a description focused on how our team could use this project:

Purpose: To help our content team track and improve our newsletter strategy over time.

Use this project to:
1. Analyze tone shifts to ensure consistency with our brand voice
2. Track audience engagement by identifying changes in content focus
3. Monitor topic trends to align with reader interests and industry developments
4. Generate insights for editorial planning and content optimization

How to use: Upload newsletter issues here for automated analysis. Review reports for actionable insights to guide our content strategy.

Project description section

Step 3: We Created Custom Instructions

We crafted detailed instructions for our analyzer to specify what to pay more attention to:

Tone Analysis Instructions:
- Focus on detecting shifts between formal, informal, and technical tones
- Highlight significant tone changes across newsletters

Audience Analysis Instructions:
- Identify shifts in target audience
- Use context clues to infer audience changes

Topic Analysis Instructions:
- Track frequency of key topics
- Break down topic prominence over time
- Flag emerging trends or recurring themes

Output Preferences:
- Create structured report with visual elements
- Use clear, neutral language for objective insights
- Summarize feedback into actionable points

Other Behaviors:
- Maintain formal, professional tone in reports
- Flag data inconsistencies for review

Project’s custom instructions

Step 4: We Uploaded Documents

Next, we upload our newsletter PDFs for analysis and organize documents chronologically for easier tracking:

Document upload interface

Step 5: Test and Refine

We tested our analyzer with a sample analysis and asked for visuals. Here is the prompt:

Group and analyse the topics discussed in the newsletter editions 

Claude's analysis output

#1 Generated visual

#2 Generated visual

Artifacts appear on the right side at this stage, allowing us to follow how Claude is analyzing the file. This helps us monitor the process and adds a level of explainability to the results.

Step 6: The Newsletter Analyzer is ready to go!

With our analyzer set up and refined, we started using it for ongoing newsletter analysis

Newsletter Analyzer chat

When giving instructions to an AI model like Claude’s Projects, it’s important to first determine the output you want to achieve. To do this, crafting the right prompt is key. If you’re interested in learning more about prompting, validation, and prototyping with AI, be sure to join our next Bootcamp cohort starting on November 11th.

Create Your Own Claude Project

Our Newsletter Analyzer is just one example of how Claude's projects feature can be used. Here's how you can create your own project for your specific needs:

  1. Identify Your Goal: Determine what you want to analyze or track over time. This could be customer feedback, industry reports, or competitor communications.

  2. Create a New Project: Give your project a name and briefly explain the project's purpose and how your team should use it

  3. Set Up Custom Instructions: Define what you want Claude to focus on in its analysis and specify the format and style of the output you need

  4. Upload Relevant Documents: Gather the materials you want to analyze and upload them to your project

  5. Use Regularly: Incorporate the project into your workflow and continue to upload new documents as they become available

Start simple and iterate. You can always expand the project's scope as you become more comfortable with the feature.

Want to get even more practical? Explore hands-on AI learning with AI Academy:

We'll be back with more AI tips soon!