Just yesterday, I was using AI to pick my NCAA tourney bracket.
As much as I wanted to use Claude to execute this task, there was a problem: I needed realtime data, and Claude couldn’t access the internet.
If only…
I knew it would come someday, it’s only a matter of when.
(For the record, after the first day the AI correctly picked 13 of 16 games correctly, and the bracket sits at the 92% out of all entries.)
I finished my bracket and then wrote about the experience. It was interesting to compare the output of the two models and see how they arrived at each pick.
At some point in the evening, I opened my email.
Skimming through my unread messages, a particular one caught my eye:
Wait, whaaaaaaat?
“Claude can now search the web.”
No way! This is a huge deal.
Here is the full write up:
I knew the day was coming. It was a gap I saw with Claude’s LLMs and Anthropic’s API model.
In the past, each time I have needed to prompt something that required realtime data, I would leave Claude and leverage another source of data.
This is what developers have had to do as they have required real time data while using Claude, they have had to look elsewhere.
Yesterday, Anthropic filled that gap.
Yes!
Okay, I had a few minutes before shutting down for the night, and I wanted to give Claude an opportunity to show me it leveraging its new ability!
I was curious: what would Claude offer me?
Claude came back to me with a list of 3 “interesting” current events and a NYT word puzzle from that day.
YES!
This is a prompt I use a lot. It is one I have built with in the past using other servicees. However, it never quite did what I wanted it to do.
This opens up new potential for me!
Here is the latest AI news from Claude, on March 20th, 2025.
Some of these things I already knew, but the output looked promising.
I had to shut it down at that point, but my mind was buzzing with chatter.
This is big deal for Claude and Anthropic!
I am excited to leverage it!
Conversating Claude 
Leave a comment