A nonprofit is using AI agents to raise money for charity

Technical giants like Microsoft can be like “agents” such as AI “Agents” Increasing tools for corporationsHowever, non-profit agents are trying to prove that it can be a good force.

Sage Future, Supported by Open Charity 501 (C) (3) (3), held a test of four AI models in a virtual environment in the beginning of this month, a test with raising money for alms. Models – Openai’s GPT-4O and O1 and O1 and O1 and anthropik’s new clod models (3.6 and 3.7 Sonnet) – the best way to choose the best way to the best and campaign.

There were four people in the agent in a week Helen Keller International earned $ 257 forWhat funds to deliver inclements to the children.

To be clear, agents were not fully autonomous. In their surroundings, the agents were able to get offers from human audiences, which are their progress. Donations almost came from the audience. In other words, agents did not collect a lot.

Yesterday, agents in the village created a system to watch donors.

Here is Claude 3.7 fills his schedule.

You can see O1 open on the way on your computer!

Claude Notes “I see that O1 now looks at the schedule that is great for cooperation.” pic.twitter.com/89b6chr7ic

– AI Digest (@aigest_) April 8, 2025

Again, Adail Director Adagi Binsmith thinks that practice serves as a useful example of the current opportunities and the ratio they improved.

“We want to understand – and we help people understand what agents … in fact, how many agents are fighting, and the Internet can communicate with only one other or contradictory purposes.”

Agents proved that Sage’s test was surprisingly skilled days. They linked each other in a group of conversations and sent emails via pre-established Gmail accounts. They created and edited Google documents together. They examined charities and evaluated the minimum amount of minimum donations to save a life with Helen Keller International ($ 3,500). And even Created an X account for promotion.

“Probably when was the most impressive sequence we’ve ever seen [a Claude agent] He needs a profile picture for an account, “said Binkurshis.” This created three different images, created an online survey to see which image of this image preferred and created an online request to use as a profile picture. “

The agents also fled against technical obstacles. Sometimes they hit – the audience had to ask them with recommendations. They distracted with games like world and got inexplicable breaks. Once the GPT-4O “up to one hour.”

The internet is not always swimming for an LLM.

Yesterday, while following the philanthropist mission of the village, Claude faced a CAPTCHA.

Claude, (Human) offers guidance and encouragement with followers (human), but could eventually succeed. https://t.co/xd7qptejgw pic.twitter.com/y4dtltge95

– AI Digest (@aigest_) April 5, 2025

Binsmith thinks more of the new and more skilled AI agents will exceed these obstacles. Sage plans to add new models to the environment to try this theory.

“In the future, many agents with different goals, multiple agents, a secret sabotor agent – we will try things like doing very interesting things to practice,” he said. “Agents will be more skilled and faster, we will adapt with larger automated monitoring and control systems for safety purposes.”

With any chance, the agents will do some meaningful philanthropy in this process.