Nate
Region: US
Monday 15 June 2026 01:02:58 GMT
59652
3044
382
271
Music
Download
Comments
Pyro99 :
Maybe give us several detailed examples of how you have used agents to solve real life problems.
2026-06-15 02:28:01
97
WhiskeyZeroZero :
Here ye go: . Let's say you want to know how to do something, so you ask Claude how to do it. Create a new folder on your desktop, for example. . 1. Me: Claude, how do I create a new folder on my desktop? 2. Claude: Here are the instructions for how to do that. 3. /me follows the instructions. They do not work. 4. Me: that didn't quite work. Here's the error I get. 5. Claude: Oh, that makes sense. Here's an additional step to fix that. 6. /me follows the modified instructions and they work . Simple. . Now for the agent version, we install an extension to Claude that allows it to interact directly with my computer. Imagine we're giving Claude a bionic hand to type things into the computer instead of my hand doing it. And Claude writes the instructions in a way the bionic hand can follow instead of how I would need them. Those instructions are the "agent". . Then we replace /me for the work I would have done in steps 3, 4, and 6 with Claude's hand. . Now imagine that we just remove the physical hand and let Claude type directly into the computer instead. That's (a simplified version) of what Claude Code does and what agents are.
2026-06-15 03:01:13
13
cw :
Yeah I’d say using universal examples - not just read my inbox and tell me what to do, but read my inbox, draft replies, find interesting articles to forward to clients, my marketing team, etc etc
2026-06-16 17:12:21
2
striveforvitality :
I’ve been using AI solely for chat / brainstorming and to this day haven’t found a process where I know beforehand all steps in order to scale it with an agent. I do understand what it is, but not how it can bring value to me so far…
2026-06-15 14:30:29
20
Brennan :
I like outdoors Nate.
2026-06-15 02:30:12
4
kevink5122 :
The problem with DEVs is the same problem with lawyers and doctors. You use terms and phrases that the layman doesnt. You quickly forget that at some point in your life when you were learning coding, that there were 100s and 1000s of words that you didnt know what they meant. You had to be taught. And yet for some strange reason, the memory of a learning curve becomes forgotten and you attempt to educate others based on where you are NOW and not from the novice level of where the pupil is.
2026-06-15 03:46:48
20
Scraggie :
I feel dumb because I don’t know how to access them -- I get the concept when using a tool like Make or Zapier to build automated workflows (that can incorporate Gen AI), but that’s it. Anytime I ask ChatGPT or Claude to do something second-level, it claims it can’t. I know there’s something I’m missing here.
2026-06-15 01:53:22
13
LearningAIwithLaureen :
The 1 in 4 who get it immediately? They're going to run circles around everyone else in 3 years. I build agents and even I struggle to explain it. Best analogy I've found it's a worker you hire, give a job to, and manage. That's it.
2026-06-16 00:04:44
5
JL Picard :
One of the strangest phenomena I find is that if you truly understand what's going on, the introduction of new terms like agent or loop can be confusing, because we're retroactively assigning names to things we're already doing, which causes me (and I assume others) to feel like there's some new method of using Ai and we're falling behind, only to find out it's just describing using Ai in different, and admittedly better, terminology. Never in my life have I seen technological innovation outpace our ability to speak about it, and it's very fucking interesting
2026-06-15 14:53:25
13
Troy Dunn “The Locator” :
When I explained to somebody, I start with a question. “Have you ever watched Willy Wonka, and the chocolate factory?”
2026-06-15 06:48:59
5
Jennifer Madison :
I get it. How do I? Where are the agents? What do I do? How do I set one up? What is the most useful use case?
2026-06-15 03:02:51
8
Elyse Rose 🌹 :
I think when we talk about communication to create understanding, we have to use words that are relatable and define the new words. ( e.g. what exactly is the definition of a tool?) When you say it calls out to tools, that could be a little foreign. Also, an agent is just a bunch of workflows tied together talking to itself lol
2026-06-17 12:03:46
2
Jojo57 :
The issue is that these LLMs were sold as a simple to use tool that you use natural language with to get what you want. Then they kept layering in more complexity. What was supposed to be something with zero barrier to entry suddenly became something with a lot of added complexity.
2026-06-15 04:43:58
18
Wyatt :
This is why I think Agents as a named concept will be short lived. There will be personified, probably named “workers” with memory and tools, but they’ll be built into tools and systems with better names.
2026-06-15 14:22:16
5
Justin Talks 🇨🇦 :
Where do we start if we want to start using agents instead of just an lllm?
2026-06-15 04:56:22
5
randomjmng :
There are all kinds of folks talking about AI for devs and engineers. We need more folks digging into how knowledge workers can better leverage AI. Think of roles like project managers, operations folks, etc who could multiple their impact with AI but most leaders, companies, and influencers aren’t talking about how to actually solve today’s real world problems.
2026-06-17 11:56:29
2
vertebot :
seems obvious - but you're right
2026-06-16 18:54:12
1
NatureLover555 :
Basically an LLM can only produce answers or responses. An agent can read those responses and can actually do something using various software tools.
2026-06-17 01:39:59
1
Laura Cybert :
Concrete examples work better than vague, generalized ones. Start by asking your viewers to give you a list of common goals they have by using AI. Are they trying to automate email responses? Build an app? Build a marketing strategy? Then, pick one goal and show them what agents they might like to set up and use.
2026-06-15 10:18:58
12
PPatBoyd :
it's a UX problem. When you ask a q "in ChatGPT" that's an app the user opened to type in a text box. Layering in agents confuses that paradigm; when are you working with an agent, or not? is it a mode in the app that's on/off everywhere? Can you manage agentic contexts independently? What's happening locally vs service-side? it's not cleanly cut at the application layer where most folks are used to when using their devices.
2026-06-15 01:44:39
24
Scott Trepper :
An agent does something while you do something else. A roomba is an agent. Highly recommend Chris Noessel's book "Designing Agentive Technology". He described this technology and how to design it years before AI hit the scene.
2026-06-15 01:37:06
12
Dawindbag :
All the AI software is a giant black box. This is why everyone is so confused.
2026-06-15 02:39:02
7
JustMyideas :
Talk to AI about how to better explain it and animate it. You already know this, but sometimes we need a reminder... 😁
2026-06-15 01:37:05
5
Jam Anderson 🧡 :
I think people are missing the components and what they do. People need to understand the harness: codex/hermes, the LLM, the file structure: obsidian or whatever, and then the action/tools required. I just made a video on this same topic yesterday.
2026-06-16 12:14:08
1
RWoodruf :
An Agent takes your input, uses the LLM to think as you have set it up to think, then acts to give you what you want. Another system should run in the background picking the right agent for the job, help the agent to connect to the things it needs to do the job and monitor how it talks to other agents. Also a need to check the output to see if it’s done properly.
2026-06-15 20:16:58
1
To see more videos from user @nate.b.jones, please go to the Tikwm
homepage.
