Interesting article. You point to the ability of AI agents to both discover new APIs, and to optimize workflows through feedback loops. I've been building "AI agent" workflows via n8n (who sponsor the article) for a while, and I've never seen a workflow that it able to do either of these things, although I accept that OpenAI Operator may be much smarter.
So if they don't exist today, are they perhaps visions of the future as opposed to something that is possible today?
API discovery can happen if an AI agent can programmatically access an API catalog and obtain the best choice for the task it needs to finish. You could create a workflow where one of the steps would be to discover which API(s) to use.
With n8n, you can already implement API discovery by connecting more than one tool to an AI agent. Behind the scenes, n8n uses LangChain, which follows the ReAct framework to determine at runtime which is the best tool to fulfill the user's request.
Feedback loops happen whenever an AI agent finishes a task. The agent can then receive human feedback, analyze usage-related information from any of the tools, and even compare the result of its work with industry standards. Any of these types of information can help the AI agent improve its next iteration of the workflow.
I'm not aware of an easy way to implement feedback loops. However, I believe that with some experimentation, we could come up with a prototype workflow featuring positive feedback.
Thanks. It's certainly a wonderful idea, and yet-another reason for folks to invest in getting their digital API schemas in order, to allow them to become more AI-accessible going forward.
Interesting article. You point to the ability of AI agents to both discover new APIs, and to optimize workflows through feedback loops. I've been building "AI agent" workflows via n8n (who sponsor the article) for a while, and I've never seen a workflow that it able to do either of these things, although I accept that OpenAI Operator may be much smarter.
So if they don't exist today, are they perhaps visions of the future as opposed to something that is possible today?
Thanks for the comment, Chris.
API discovery can happen if an AI agent can programmatically access an API catalog and obtain the best choice for the task it needs to finish. You could create a workflow where one of the steps would be to discover which API(s) to use.
With n8n, you can already implement API discovery by connecting more than one tool to an AI agent. Behind the scenes, n8n uses LangChain, which follows the ReAct framework to determine at runtime which is the best tool to fulfill the user's request.
Feedback loops happen whenever an AI agent finishes a task. The agent can then receive human feedback, analyze usage-related information from any of the tools, and even compare the result of its work with industry standards. Any of these types of information can help the AI agent improve its next iteration of the workflow.
I'm not aware of an easy way to implement feedback loops. However, I believe that with some experimentation, we could come up with a prototype workflow featuring positive feedback.
Thanks. It's certainly a wonderful idea, and yet-another reason for folks to invest in getting their digital API schemas in order, to allow them to become more AI-accessible going forward.