The simplest way to use Browse AI data with an AI assistant is to export it and upload it directly. No API keys, no webhooks, no code. Just download your data from Browse AI and bring it into Claude, ChatGPT, Gemini, or any other AI chat interface.
This approach is ideal when you want to ask questions about your scraped data, use it as reference material for a project, or get quick analysis without building an automated pipeline.
📖 When you're ready to automate this process, see our guides on enriching data with Claude, analyzing data with Claude, and connecting to any LLM via API.
Step 1: Export your data from Browse AI
Browse AI lets you export your scraped data in multiple formats. To export:
Go to your robot from the dashboard and click Tables.
Click the Export button in the top-right corner of your table.
Choose your format (see comparison below).
Which format should you use?
Format | Best for | Why |
JSON (recommended) | Most AI assistant use cases | Preserves field names, structure, and relationships. LLMs parse it accurately and can reference specific fields by name. |
CSV | Simple flat tables, spreadsheet workflows | Good for straightforward tabular data. Works well when your data is a single flat table without nested fields. |
✅ Why JSON is recommended: AI assistants understand JSON natively. Field names are preserved (so the AI knows that "price" is a price, not just a column), nested data stays intact, and there's no ambiguity from commas or line breaks in your data. JSON files also tend to be more token-efficient than formatted tables, meaning you can fit more data into a single conversation.
Step 2: Upload to your AI assistant
Each platform handles file uploads slightly differently.
Claude (claude.ai)
Open a new conversation at claude.ai.
Click the paperclip icon or drag and drop your JSON/CSV file into the chat.
Add your prompt. Claude will read the file and respond based on its contents.
Claude supports files up to 30MB and can handle large datasets within its 200K-token context window (roughly 150,000 words of data).
ChatGPT (chatgpt.com)
Open a new chat at chatgpt.com.
Click the paperclip icon and select your exported file.
Add your prompt. ChatGPT will process the file using its Advanced Data Analysis feature for CSV files, or read JSON files directly.
Google Gemini (gemini.google.com)
Open Gemini at gemini.google.com.
Click the + icon to attach a file.
Upload your exported JSON or CSV and add your prompt.
Step 3: Ask the right questions
The quality of your analysis depends on how you prompt the AI. Here are effective patterns for different use cases:
Quick analysis
I've uploaded a JSON export from Browse AI containing scraped [product listings / job postings / competitor pages / etc.]. Please analyze this data and tell me: 1. How many records are included? 2. What are the main patterns you see? 3. Are there any outliers or surprising findings? 4. Summarize the key takeaways in 3-5 bullet points.
Competitive intelligence
This JSON file contains product data scraped from [competitor name]'s website. Compare their offerings and help me understand: - Their pricing strategy (budget vs. premium positioning) - Feature gaps compared to our product: [list your key features] - How they describe their value proposition - Any patterns in their product naming or categorization
Data summarization and reporting
I've uploaded scraped data from [source]. Please create a structured report with: - An executive summary (3-4 sentences) - Key findings organized by theme - A comparison table of the top entries - Recommended next steps based on what you see
Content research
This file contains content scraped from [websites/blogs/forums]. I'm using it as research for [project goal]. Please: 1. Identify the main topics and themes covered 2. Note any common opinions or recurring points 3. Flag anything that contradicts or is inconsistent 4. Suggest angles or gaps I should explore further
Using exports as project context
Beyond one-off conversations, you can use Browse AI exports as persistent reference material that an AI assistant draws from across multiple conversations.
Claude Projects
Claude Projects let you upload reference files that persist across all conversations within that project.
Create a new Project in Claude (available on Pro, Team, and Enterprise plans).
Click Add content in the project knowledge section.
Upload your Browse AI JSON exports as project knowledge files.
Add project instructions like: "This project contains scraped data from our competitor websites. Use this data to answer questions about competitor pricing, features, and positioning."
Every new conversation in this project can now reference your scraped data without re-uploading.
This is particularly useful when you want to:
Build a research hub from scraped industry data that your whole team can query
Keep competitor data as a living reference you update with fresh exports
Create a knowledge base from scraped documentation or help articles
ChatGPT custom GPTs
You can create a custom GPT with your Browse AI data as its knowledge base:
Go to chatgpt.com/gpts/editor.
Create a new GPT and upload your JSON exports under the Knowledge section.
Write instructions telling the GPT how to use the data (e.g., "You have access to competitor product data. Help users compare features, understand pricing, and identify opportunities.").
Share the custom GPT with your team.
Google NotebookLM
Google NotebookLM is designed specifically for analyzing uploaded documents:
Go to notebooklm.google.com.
Create a new notebook and upload your Browse AI exports as sources.
Ask questions about your data. NotebookLM will cite specific entries from your uploads.
✅ Tip: For project context use, re-export your data from Browse AI periodically (weekly or after major scrapes) and replace the old files. This keeps your AI assistant working with fresh data without any code or automation.
Tips for getting the best results
Tip | Why it helps |
Tell the AI what the data is | "This is scraped product data from Amazon" gives the AI context to interpret fields correctly. |
Specify what you want | Instead of "analyze this," say "identify the 3 cheapest options and explain why they're priced lower." |
Request specific output formats | "Present this as a comparison table" or "give me bullet points" helps get actionable output. |
Ask follow-up questions | The AI remembers the uploaded data for the whole conversation. Drill deeper without re-uploading. |
Use JSON for complex data | If your data has nested fields, URLs, or long text content, JSON preserves everything cleanly. |
Split very large exports | If you have thousands of records, upload in batches or filter to the most relevant subset first. |
When to upgrade to automated pipelines
The manual export-and-upload approach is great for ad hoc analysis and research. Consider moving to an automated integration when:
You find yourself exporting and uploading the same data repeatedly
You need results delivered to your team automatically (Slack, email, CRM)
You want real-time processing as soon as data is scraped
You need to enrich individual records at scale (hundreds or thousands of rows)
When you're ready, check out:
How to enrich Browse AI data with Claude for adding AI-generated fields to each record
How to analyze Browse AI data with Claude for automated reports and knowledge bases
How to automate workflows with Browse AI and Claude for trigger-based pipelines
How to connect Browse AI to any LLM for OpenAI, Gemini, Mistral, and other providers