FINALYLLYLYY BUILDING AN AI AGENT!!!… (Well… Not Quite - BUT - FINALLY an AGENTIC WHISPERER)
Disclaimer: Everything below is a mix of what I observed and heard during the event. The goal isn’t to pinpoint "who exactly said what," but to share (usually) an outsider's view and overall perspective on these industries. I’m not here to act as a definitive firsthand source—readers should do their own research. I hope this inspires you to attend events, explore new industries, and hear what leaders are presenting. These notes combine my observations with thoughts on how things could run smoother and how ideas connect (IMO). I’m not an expert, you know? Just hanging out in the room with them. Enjoy!Topics: AI Agents, Tech, Automation, Artificial Intelligence, Entrepreneurship, Productivity, Innovation
GAHHH I’ve been up, down, left, right - all over this city trying to get someone to help me understand how to build an agent! Why? Cause its apparently the next-best thing, and not just for superficial reasons. You can build your own employees/tools apparently to help you with tasks. So, I’ve heard of this for years - I mean, its a brand new tech, but I feel like I heard of it right as it was coming out… yet, no one can seem to figure out how to explain how to build them. Just like 3 blogs ago they said “zero to agent in 3 hours” and we didnt do that. So… will this online/virtual event FINALLY meet my needs?
Why Attend: I want to learn more tools and tech so that I can optimize my business and beyond!! Duh. I mean, I want to attend a lecture that helps me jump ahead in my understanding of building agents and get to work. I feel like I get it… like what you can do with agent, but I don’t - you know? So, its friday night and let’s build an agent.
Photo Collage and Commentary:
Notes from the Event:
This event they’re off to a start with a mistake. They were supposed to give us 2,500 credits but then it was 2,200 instead and he said I made a mistake and clicked the wrong link, but I”m like … no. I literally jsut scanned your QR code you have here. So then a bunch of people said the have 2,200 too and he was like, oh. It’s my mistake.
LOL but one girl got 2700 credits .
LOL. No fair! They need to send us more.
And he’s saying if you’re interested you can access mulrun computer with open claw?? What. How does that even work.
I asked him to please explain it later
They’re building a personal AI agent with the primary goal of helping you make workflows and save them as knowledge to share them with the people around you to run similar workflows as well.
Now we’re looking at hospitality and how small/medium businesses can benefit from AI agents.
Like if you were a hotel manager of a chain and you care about your reviews. So you may ask the agent to go scrape all the reviews of 2 stars and less over the past 30 days and providing actionable items and publish it all on an HTML webpage to publish and use internally.
Also a hospitality use case. Managing a hotel chain, monitor the prices of your competitors.
People are asking him to stop and explain his company. So he did but I didn’t quite understand. Just that they’re an agent.
One thing it can do is help with finance, or e-commerce. And e-commerce has tons of manual labor, if you have a store with 100-200 products, you can upload them and have photos and then use the agent to move onto a different platform. Maybe IG to YouTube shorts. Reformat.
Lots of people are asking questions in the chat and its like, he jumped right into his product without explaining it at all. So it’s a bit chaotic.
He said mule run has API that gives you access to video gen or clank or midjourney.fusion or 1 image generation model.
Now someone asked the difference between his product and open claw. Open claw has the OS for AI agents. But there are problems with Open Claw because of security, performance, token optimization. It’s awesome but not commercially viable as a product to shelf. So for them to provide the product, tehy have to patch a lot of things up, it’s what they’re doing.
Okay. So? I’m confused by this. I feel like… why are we using this app instead of another? Whats the value propostion of his website/this app, over another? I literally asked him that.
He said the main purpose is not to fill gaps and do this.
he said they have two agentic features. One is “task” the other is “computer” and “task is giving you a dedicated VM always on to do time-triggered tasks” but under the task feature, we developed everything ourselves. It’s an AI agent that is completely done by us and the performance of us is better for a token optimization perspective. But it’s not always on or time triggered.
He asked if that answered my question better but I said not really… but I’ll keep listening. I’m just confused what is going on on this website and what is it and what is it doing and whats even going on and whats its similarity/competitor?
HE said he’ll keep going through examples and then invite me onto the mic and let me ask questions.
He said they’re using his product to make games.
It can make a Minecraft game.
They made a game thats like a temple run and it runs through obstacles.
Also you can use it to make kids games. Parents with kids who want to make games for their kids. Build edutaining games where the player has to solve math probles as fast as possible and catch the items before they drop.
love that!
Lastly, it can integrate the majority of wildly popular image and video generation.
It can turn an image into a short film.
One guy signed up thinking this will help him automate his life or make things work.
- These are all case studies without a framework.
- he just spent time talking to Claude during this
He says that mulerun is a startup. But the user needs to build automation… you guys understand the context and adapt to changes and handle multitask steps?
Okay so at least some of the audience is on the same boat as me and confused. Trying to understand it.
He said N8N and zapir and other ones are good at helping users make AI agents with node based systems. And this is really good… but for them they took a different approach and the orchestration is natural language prompting.
They’re using AI to simplify hte automation workflow.
AUDIENCE: What are your pros and cons?
Not to be in the position of explainign everything, but back up. This isn’t relevant to me. Are you an established company? Is this a startup trying to get legs? Whats happening?
He said the second question he’ll take first. Quite established and started 6 months ago. Didn’t want to build our own agent. Wanted an AI agent marketplace. You can see reminants of that. Our goal was to help agent builders distribute and commercialize agents already built. In order to do that, we had to make our platform compatible with competitors.
The barrier to use an agent effectively to increase productivity went straight down. The quality and capability of agents able to automate went straight up. SO then they capitalized this opportunity and built their own agent which is the purpose of this workshop today.
This agent is used to help everyone automate complex multistep workflows in person and professional life.
So whats the difference in you guys versus the others?
Other platforms run DETERMINISTIC workflows - node based.
However, if you’re trying to get context-complex workflows, want to use an AI agent leveraging a base model to do a lot of orchestrations.
He said one of these use cases is a building VM. Virtual browser.
Log into your own liknedin to reach a bunch of people and the ways you want the agent to do.
The audience guy asks again:
Are you guys fully in revenue or is this an MVP.
He said yes, tehy’re revenue generating.
They’ve had 200k users
This feature is just released in February this year. Overwhelming positive and partnering with partners like (this group_ and other communities for online and offline events. In tokyo, Japan, Paris, west and east coast. So far really good. All use cases are real user use cases across different verticals.
How much compute would a 3 min video use ?
I didnt undrstand his answers
He said they’re not similar to lovable cause lovable builds apps for users.
They’re just an AI agent tool, helping you automate certain things in your life. In his examples are HTML released to the web.
If you want to build a full on web app, use lovable.
Okay he said if you go to “tasks” you can find tasks you want to try.
I want to show this all to AI quickly and ask it waht something like this can do for me to help me.
He said this website is pretty safe, their app. You can delete everything and wipe it out. But when you share tasks with peopel, be mindful of what you’ve shared. And if you API key was shared with others unless you’re sure of it.
Now he’s saying there will be voting at the end. But I dont get this. I feel like.. maybe it’s super amazing or maybe it’s confusing.
I wish someone woudl simplify all of this.
He is saying we’re all going through this SME paradigm - we’re all subject matter experts and selling our time for money.
With AI agents, it’s quickly evolving from SME as a service to a product. Soon, people can leverage your expertise and knowledge as a product.
This shift will not stop and it depends on how open everyone is towards this paradigm shift and what tools we should use to embrace this.
He said I asked a good questions. Does anyone want to share how they feel so far - is everyone equally as lost?
Someone said it sounds like lovable but he says that lovable is primarily building apps, while agents automate workflows.
I’m like, uh. okay. Hm. Makes sense, like lovable is “extra”
One guy wants to make a suggestion - but he says maybe there are people who are interested in using mule run… 33 people here all of different skillsets I woudl assume. Maybe we can break into smaller groups who jsut want to understand what the hell this is and use it, and then others are gungho about building a lijnkedin thing. I need to use it for my email, I know what I want to do -> build a movie… maybe interest based?
If this zoom can go into breakout rooms, we can learn about each other and work in small groups. Maybe some more level setting for different skillsets. If this is a good idea, if not, continue as you are.
He says what we can do is this… idk if zoom has breakout sessions. For those of you already knowing you want to experiment, go ahead. For the remaining, if you want to listen, I’ll give you a couple easier examples so we can get started.
Okay show he’s showing how you can evaluable YouTube videos and scrape them to gather data with this app.
I liked that “breakout room” idea - but… no go.
The data isn’t stored anywhere and/or it’s in a way that can be deleted and no one knows it?
But when you delete it, your data is wiped clean. And they have enterprise agreement and promise not to use the prompts to train the model.
They make sure the base model providers and they give us the promise.
It’s hard to get the service as an average consumer. You’re making sure there’s another level of security.
He says you’re not building an agent. They are the agent. So you can learn to use it to optimimate workflows…
We’re all familiar with ChatGPT… then let’s move from “thinking” to DOING.
Draft a document for me. A deed for my house.
Shifting from thinking to doing… to help you think and orchestrate and doing stuff.
But you also need resources, like in this example, a template - a dox template.
The brain needs a virtual computer to store this template and then maybe inshore Microsoft words so it can actually read and edit as per your instruction. Thinking to doing.
So.. here the natural language is slightly easier.
Omg suddenly his answer is done and I’m so surprised cause I didnt understand it ahahah. I thought his answer wasn’t done. Yet I was enjoying him speaking ahhahaha. Cuase it was so intelligent sounding. He went from “frusterating” to “useful” over the course of this.
He said they do have a library ( I was asking if they’r more of a library or a typewriter or both)
He said yes, at the skills hub, they have a library
Okay and then I pointed out their verbiage of this event is confusing even cause its like literally saying (I wrote them this comment): this event is called “Turn Daily Workflows into AI Agents “ so even that is a bit confusing - cause you feel like you’re supposed to learn how to “build an agent” and turn workflows into ai agents… but its actually - automate workflows WITH this AI Agent… So we’re using your agent to build tools to help us.
Then even their staff member said: “Hi Kelly, This is a really good feedback. Thank you for pointing this out, we will take that into account next time when we’re naming our events to avoid future confusion.”
Dude. Idk I dont even know if I want to stay here and build this solo. I’d love to work in a group. Idk I feel like I wanna leave now hahahah.
I asked them so many questions and I feel like it’s cool and interesting and now I understand this so much better… but also, I’m uninspired and it’s late. It’s 7:30pm. And I’m like, uh.. idk do I want to do this till 9pm? I wish we were all talking and building together.
I wanna first brainstorm with AI that already knows my company and then figure out how I can build tools on this. Yeah I dont think I’ll stay. But I am glad to finally learn of this tool. Cause it’s like exactly what I’ve been looking for.
Okay, I sent them this message:
- “Okay this answered a lot of questions for me and I hope you guys can offer more virtual events. I'd like to join your newsletter or something if you have more "non-tech people" online events. I'll go now, but this was really helpful and kinda exactly what I was hoping to get out of this event.
I don't wanna go build on my own for an hour, I dont have the focus at a moment hahaha. But I’d wanna build with more interaction/chatter, and/or just build another time when I'm more in the mood. Like the daytime 🙂
But this was super helpful and explained so much I've been trying to understand. Thank you so much! I'm excited to use your website soon.”
The leader in charge send an applause emoji and said thank you to me. lol. I’m impressed with his intelligence and ability ot stay cool in that session. He’s a cool guy hahahah omg.
LOL on my way out this girl asks if they’re HIPPA compliant and he said no, not yet
Ahahaha. Okay, I’m out.
