Build an AI helper that can read web pages and call a simple API to answer user questions. It is great for quick research, content prep, and idea suggestions without manual browsing or copying text.

A manual test start loads two sample prompts. Each prompt feeds an AI agent that uses the OpenAI chat model. One agent can call a web scraper tool that sends a request to Firecrawl and returns clean page content as markdown with only the main text. The second agent can call a public API to suggest activities using query parameters like type and participants. Both tools plug into the agents as actions, so the model decides when to fetch data and how to use it. This pattern removes the need for subworkflows and can cut node count by up to ten in similar builds.

You will need an OpenAI API key and a Firecrawl API key. The activity API does not require a key. Expect faster research, cleaner outputs, and less switching between tabs. Common uses include pulling key facts from a page, drafting quick summaries, and suggesting ideas based on simple filters. Once tested, adapt the prompts to match your team’s needs.

• Manual test start triggers two parallel AI requests for quick validation
• Prebuilt prompts with Set nodes to control the agent questions
• OpenAI chat model powers both agents for natural language reasoning
• Web scraper tool posts to Firecrawl and returns clean markdown with only main content
• HTTP tool hits a public API with query parameters for type and participants
• Agents decide when to call tools and merge results into a final answer
• Optimized responses reduce extra formatting work

• Reduce manual research from 60 minutes to 10 minutes
• Cut node count by up to 10 using the HTTP tool approach
• Improve answer relevance by letting the AI fetch fresh data
• Handle more research requests in less time without extra staff
• Connect AI with web pages and an API in one place

1. Import the template into n8n: Create a new workflow in n8n > Click the three dots menu > Select 'Import from File' > Choose the downloaded JSON file.
2. You'll need accounts with OpenAI, Firecrawl and Bored API. See the Tools Required section above for links to create accounts with these services.
3. In the n8n credentials manager, create an OpenAI credential. Get your API key from your OpenAI account API page. Name the credential clearly so you can find it later.
4. Create a Firecrawl HTTP Header credential. Get your API key from your Firecrawl dashboard API page. Add an Authorization header with your key and save.
5. Open the OpenAI Chat Model nodes and select your OpenAI credential in the Credential to connect with dropdown.
6. Open the Webscraper Tool node and select the Firecrawl HTTP header credential. Confirm the method is POST and that fields are set to markdown.
7. No credential is required for the activity API. Open the Activity Tool node and review the query parameters for type and participants.
8. Review the Set nodes to adjust the sample prompts. Keep them short and clear so the agent knows when to call the tools.
9. Click Test workflow to run. Check the execution data and review the agent outputs. You should see scraped markdown from the web tool and a suggestion from the activity API.
10. If you see an authentication error, recheck the OpenAI and Firecrawl keys. Make sure the header name and value are correct in the HTTP header credential.
11. If scraping returns empty content, try another URL or remove strict page options in the Webscraper Tool body settings.
12. When results look good, replace the sample prompts and URLs with your real use case and save the workflow.