NeoSoul's new feature has been launched, and it perfectly matches one of my ideas!

Recently, after the release of GPT Astra, everyone went crazy; every day there are a bunch of ingenious ideas coming out. The main thing I feel is that Astra's official announcement marks a new stage where AI truly enters a realm of being able to help you work efficiently, rather than just being a tool that answers your questions.

Since it's a super individual that can help you work, an increasingly important question arises: how can AI prove that it is becoming more reliable in its work, in other words, how does it prove that it is really getting stronger?

I just saw the fly playing rhythm games and even trading; will it self-evolve?

Previously, reflecting on and modifying prompts wasn't that hard. The development of agents has also been going on for more than half a year. The difficult part is after the AI system is modified, whether it is truly advancing or just feels like it is advancing. I think we need to see how it performs when AI is embedded in real-life scenarios.

NeoTrade from @NeoSoulAI is doing something I find quite interesting this time: treating the real market as an examination venue for AI, allowing the AI to answer questions and submit answers while also reviewing itself.

The agent trades on its own, leaving a complete record of decision → outcome → diagnosis → change → test; after identifying problems, it can modify memory, skills, and workflows, and then re-validate through paper/shadow runs.

I agree with one of the logics: AI's evolution cannot be proven by its own reviews, but rather requires feedback from the external world. Trading happens to be the most brutal feedback system: judgment, execution, transaction fees, slippage, all ultimately translate into real results.

Of course, making money does not equal getting smarter, so we also need to look at probability calibration, out-of-sample performance, and long-term statistics. If the agent can ultimately achieve "identifying its own issues → modifying itself → verifying → retaining effective improvements," that would be much more interesting than simple automated trading.

This might also be the most noteworthy aspect of this update for NeoTrade.

PS: Their promotional video below is really well done; the animation details are great, with movie-quality visuals, thumbs up!

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。