News for hackers. Stuff that matters.
Re:Ask HN: Are there any real examples of AI agents doing work by jncfhnb on Saturday January 11, 2025 @11:46AM (#42667038)
Most “agents” are just starter prompts + a small set of tools that they can use to respond to things, like access to a database.

They’re workflow automation

Re:Ask HN: Are there any real examples of AI agents doing work by viraptor on Saturday January 11, 2025 @02:24PM (#42668276)
By that definition, humans doing tasks are workflow automation.
Re:Ask HN: Are there any real examples of AI agents doing work by jncfhnb on Saturday January 11, 2025 @04:43PM (#42669149)
Humans are autonomous and not task specific
Re:Ask HN: Are there any real examples of AI agents doing work by viraptor on Saturday January 11, 2025 @06:00PM (#42669653)
Most people in large work environments are very task specific and deviation from playbook is punished.
Re:Ask HN: Are there any real examples of AI agents doing work by jncfhnb on Saturday January 11, 2025 @09:34PM (#42670783)
That’s great. They’re still capable of doing other things
Re:Ask HN: Are there any real examples of AI agents doing work by viraptor on Sunday January 12, 2025 @02:45AM (#42671948)
Sure, they're capable. They can be assigned to a new task after getting instructions and tools for that task ; )
Re:Ask HN: Are there any real examples of AI agents doing work by varelaseb on Saturday January 11, 2025 @04:12PM (#42668962)
What else do you want them to be?
Re:Ask HN: Are there any real examples of AI agents doing work by jncfhnb on Saturday January 11, 2025 @04:43PM (#42669155)
I’d like it if we could trust agents to have agency but in practice they dont
Re:Ask HN: Are there any real examples of AI agents doing work by dimitri-vs on Saturday January 11, 2025 @07:23PM (#42670087)
IMO:

Workflows = automations that use LLMs (or a sequence of LLM calls) at some point. Eg. "Classify this input and respond with JSON"

Is Siri or Google Assistant an "agent"? I would say no it's basically an LLM with function calling. Eg. "What's the weather -> uses predefined weather api"

Agents would need to be able to self integrate, which is impossible without giving them full computer use or admin permissions - which creates massive security risks nobody seems to be talking about.