Training Agents On Reading NEAR Transactions

Captions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.

1. 0:04

   So, let me show you a cool thing that you can start to do with your agent. And you could obviously do this with any chatbot, but what's nice about having your own agent is obviously you start to have your own history and you start to have these different things that go on.

2. 0:17

   And on top of all of this is that I have, you know, primed it and trained it on Near. So, it it knows Near has near skills. Um, it understands kind of what to do. And so one thing I can easily do is I can actually just give it a transaction and then ask it to explain it to me. So maybe I don't understand what happened. Maybe it was even if it didn't fail, maybe I just still wonder like what does this transaction mean? So let's try that.

3. 0:43

   So here I am. I'm in my agent and I'm going to just paste in a transaction ID and say, can you tell me what happened on this near transaction? It's going to go it's going to go think probably I'll look it up. Um, and again, when you have something like Open Claw or any other chatbot, you usually have web tools, so it can go out, it can search. Um, I'm going to assume it's going to use something like near blocks and go pull that information and bring

4. 1:21

   it in. Um, and let's see if it just does it on its own. So, there it goes. So, here's what happened. Uh, basically what it was saying that this was an NFT that was unstaked using your intents. Here are the contracts that are involved. And then it gives me a nice summary at the end just saying this NFT was removed from mistaking. It was an intentbased transaction. Was returned to the user's

5. 1:46

   wallet. Cool. So now instead of having something that was actually kind of maybe confusing and just seeing like the output from the transaction, nice natural language and I can see it all right there.