Turn AI Output Into A Shareable Artifact
From the course: Agentic AI For Founders & Senior Operators
Ask Claude to turn the result into a claude.ai artifact. Artifacts are HTML files, so they're more interactive than a standard document, and many companies now use them as their standard way to share work. You can ask for a Google Doc instead if you prefer.
Transcript
All right, so first thing in the CLI, what I've done here is very simple. I took the example that we discussed in web scraping and I said, "Okay, get me the latest Ramp Economics Lab report on AI and jobs. Give me the headline findings." So it did the usual Firecrawl scraping and so on and gave us back this whole list of things in chat. Now let's say I want to share this with my team. So I say, "Okay, turn this into a claude.ai artifact that I can share with my team." So it uses the inbuilt artifact tool and it creates this nice-looking artifact for us that I can see here. And because this is the CLI, it also has context of, you know, my company and my brand and so on. So it's used some of the brand guidelines. It's not great. I would probably never share this, but this is natively what it came up with. And the same thing on the desktop, I ran the exact same prompt to get the values back in chat. So it used the exact same Firecrawl web search and scrape, gave me all of this stuff. I want to share it with somebody. So I say, "Okay, can you turn this to a claude.ai artifact?" Now, to be clear, artifacts aren't compulsory or necessary. You could also say, "Put this into a Google Doc," and it'll do that for you. But artifacts are nice because they're basically HTML files, and they're way more interactive than any standard document would be. Many companies now use these as the standard way to share things. So this is what the desktop app came up with. This is what the CLI came up with. Not very different, actually, if you look at the specific contents. This one just didn't have the inbuilt brand kit. And so you can see here it's also added this nice little view chart data as a table. So it's made this bit interactive. So there will always, again, as with agents, be a little bit of randomness.