I’ve noticed from one technical leader who I follow that mentioned that it doesn’t make sense to start off with an agent.
Do we agree here?
Can you outline what an ‘agent’ means here, versus the alternative?
My first assumption is:
- Manual Start - Begin by sketching, jumping to visual UI, or coding the structure.
- General AI - Leverage a general tool to help with outlining the structure.
- Agentic AI - Starting with something like Lovable that will build using boilerplate patterns.
I’d say in this context it depends on familiarity with the design patterns and software solution to be built.
1 Like
Agent in the sense of an executional element. Basically, think before you do!
Agreed that it’s contextual!
I think this is like, a normal sentiment! You have to do the brain work, leg work of understanding before sending an agent off because without that context, you will struggle to create quality?
IDK! Just a thought.
1 Like
I would agree. I’m finding that an agent (aka a tool) is just an extension to people’s own knowledge.
The stronger the foundation, the more powerful the tool. I think the key here is to use the tools to build up our mental foundations, rather than trying to remove thinking altogether
Agree. To be honest, I have yet to find a case where I need agents, or where they would be valuable. But I am still looking.
3 Likes
Interesting- is this coming from a designer perspective?
Agents make sense for pulling in and synthesizing research, helping create PRDs, pull in relevant context, and tying larger processes together. It feels like it could help designers expand their work and connect it deeper into the business (doing some synthesis, coding, etc).
I still need context here. Something I struggle with in this conversation is:
- What does it look like to use an agent?
- Have I ever encountered one?
- How is it difference from me prompting AI?
- Is all AI an agent?
Can you explain it to me like I’m 2?
Ah, simply put, an agent simply has more autonomy, and can run longer than other AI tools.
ChatGPT and Claude are both close to agents, but Claude’s Cowork is much closer to an agent, where it can create a list of tasks and tackle them 1 by 1 over an extended period of time.
Whenever I think about an agent, I think about this thingy from Wall-E. To me, if Agents had a true form, this would be it. hahahaha
2 Likes
Agreed! Actually, now that I think about it, an agent should also be able to keep context, store memories, and run for even longer periods of time than what I was mentioning before.
If we look at the definition a traditional agent, an agent is any person or entity authorized to act on behalf of another party, who is known as the principal .
So I like the idea of a mental foundation, because operating within this relationship without a shared goal or context makes the relationship weaker.
In this regard, since agents are so new, it makes sense that people need more context before the need for an agent arises. An example… if you’re not driving a car, why do you need an insurance agent?
1 Like
Very interesting- the term makes even more sense now, and I think that’s what people are trying to shoot for.
From IBM:
An artificial intelligence (AI) agent is a system that autonomously performs tasks by designing workflows with available tools.
AI agents can encompass a wide range of functions beyond natural language processing including decision-making, problem-solving, interacting with external environments and performing actions.
AI agents solve complex tasks across enterprise applications, including software design, IT automation, code generation and conversational assistance. They use the advanced natural language processing techniques of large language models (LLMs) to comprehend and respond to user inputs step-by-step and determine when to call on external tools.
So, what this makes me think of as a shorthand, is that an agent is an AI (LLM) empowered to access your computer and network to accomplish specific workflows. This means it can access multiple software tools outside its core programming to accomplish a task.
1 Like
This seems to align closer to my perspective as well. Curious when we’ll have this definition of an agent in a dictionary?