Great talk, Alex. Mine is the complete opposite. No technical talking.

Yeah, so I'm one of the founders of Sageflow. My CTO could not be here today.

But a bit about my background before we go in. I also come from a world of finance, investment banking, before that, defense consulting. And I came into the AI world about three years ago.

I left IB, and an AI company is like, we're going to hire you. to go do growth and go to market. I'm like, okay. So I started doing that. But around the same time, you know, ChatGPT came out, I started using a lot of AI tooling in my workflows.

And then last year, I started building agents in the financial services space. And when I was doing that, I'm no developer, so I ran into a lot of trouble. I started interacting with a lot of agentic frameworks out there. For those of you that are devs, you're probably familiar with LandGraph, Autogen, a lot of the traditional frameworks out there, but they're very complicated.

So if you're a banker, a marketer, you don't know Python, you're like stuck. Like you have to get a devs, you have to hit up like devs and you have to hit up your friends at Google and be like, come teach me how to use this. And so around that time, I wrapped up my last project and I pulled in one of my friends who was working in long-term memory for agents in LLM. So he's deep in the machine learning world. I'm like, bro, there's like maybe 30 million devs, but there's like 8 billion non-devs. There's got to be tools out there for people that are non-technical.

And that's actually what we're building now. We're building a platform where users, like everyday users, citizen developers, can actually go on our platform to create, store, and actually monetize and make money on their agents. And that's the biggest gap in the market right now is that

A lot of the AI gains are going to the top companies in the market, but there's a lot of people around the world that are going to be building agents. And I feel like it needs to be distributed to them too. So I'm very bullish on this market.

So I mentioned earlier why I'm building this. Well, three reasons. Their technical barriers to building agents are very, very high.

On one end, like I said, there's the frameworks. On the other hand, there's extremely simple action-to-action workflow builders like Zapier and like Wordware. And they're not actually agentic. They're like deterministic tasks, and they don't do the job.

Again, they're developer-centric. Great tools for devs out there. But if you're non-dev, you're kind of stuck. Like, you don't know where to go. And lastly, the tools are extremely complex.

And so you're looping together and you're piecemealing different tools. And there isn't a good platform where you can integrate your agents into real-world applications, where it's multimodal, where you're actually able to visually edit the agents. beyond just the coding agents. And I think that's why the coding agents took off very well, because they actually presented their agents in a way where UX and UI is so amazing that consumers were able to adopt it so readily. And that's what's actually missing in a lot of the horizontal platforms right now.

So again, what are we actually building? On our platform, again, we're very early stage, but you can actually go on Sageflow today and sign up for access to our beta list. But the vision is that you can build your agent.

You can actually rent out your agent to businesses and users now. So what does that mean? In the past few months that I've been talking to customers like B2B businesses and actually other users, a lot of businesses that are small businesses in America that don't have enterprise budgets, they actually don't want to build their own agents. So we've had a lot of customers come to us, can I just go on your platform and rent out an agent? I'm like, what?

They're like, yeah, we want to rent out agents that others have built. So you as a user could build agents and potentially rent them out. You could share your agents like how GitHub, you're able to share your repos. You can share your agents with other users. They can clone or edit it if you choose to have that mode on.

Lastly, you can manage your agent. So let's say like you run the agent and you wanna share it with your team. Well, you have the ability to manage it on our platform. And we're building this again for everyday users because there's not a lot of good tooling out there for them.

Again, what is our technical edge? I had a lot of things, but I just wanted to fit the core ones here today. Visual programming and editing.

You can effortlessly build tools with our intuitive tools, but again, one thing I really, really got concerned about when I was talking to customers is data. So in our platform, you actually retain full control over your data, accessing, transforming, and optimizing it. So one example, if you're coming from the finance world, you know compliance is a huge issue.

So let's say you build an agent in a compliance team, but you also want to share that agent's memory with another team. Well, you have the ability to control that data and lock down that memory for that agent if you choose to do so. So the data doesn't lead from one department to another. You as a user control that.

everyday integrations. Something very interesting we found the last few months, a lot of the businesses coming to us, we had a funeral home company come to us, I know, very morbid, a trucking company come to us, and a few other businesses come to us, and they're asking us, I wanna actually use agents beyond the desktop. We wanna bring it into the real world. So I think that's where a lot of adoption is going to happen when you start building agents that are actually integrating with your real-world applications, like Siri, like Discord, like mobile, like iOS shortcuts, and many, many more use cases.

Lastly, my co-founder comes from a research background where he has worked on very interesting long-term memory frameworks for agents. Basically, if you're bringing down the simple terms, what does long-term memory do for you? It helps you allow the agents to remember, adapt to new situations, and learn over time. And what does that do for you?

It lowers your token cost and your compute cost. Your token usage goes down, and you end up burning less tokens when you're running that agent over again. Why am I talking about this today? Well, let's just put in numbers. The Asian market is expected to go from 5 billion to almost 140 billion.

And I feel like everyday users should be able to capitalize on this. Yeah, so... And then why should you care?

So in the past three weeks, this is what's come out. OpenAI just came out three days ago. They're gonna start renting out their agents, but their pricing points are insane.

They're charging 20 grand a month for the PhD level agent. The master's level agent from people I know at OpenAI, they're gonna start charging $10,000 an agent. The ones below that, $5,000 an agent. Small businesses cannot afford that, but that's what they're pricing it at.

On the other hand, we have another company that came out. They're pricing their agent at $1.63 a day. So the pricing is all over the place, and you get to decide your pricing as a user on our platform. But that's like the future you're looking at.

You could, I personally think that there's going to be many, many, many agents built and many, many millions of users are going to be actually renting them out and we're going to have full on agent to agent economy. I've had agent, an agent to company reach out to me where they're building a payment system where agents are renting out other agents. Agents are talking to other agents and that's kind of like the world we're looking at.

So before I go into the basic demo I built for today, what are some of the core features that we're building, again, cross-platform integration across devices, including mobile? The trucking company I mentioned earlier, they came to me and they're like, we want voice AI. I'm like, OK, but why? Well, if my truck driver is driving across Canada to New York and the truck breaks down, he got to call this woman in the middle of the night to go to the workshop. I don't want to do that anymore.

Instead, 1we want to build a voice AI agent that could call the local workshop. Well, that's through mobile, like you're interacting with that agent on the phone. Community driven.

We're very bullish on building something similar to what GitHub did like 15 years ago, meaning community-driven platforms are the future. When you open up the platform where people can share, clone, rent out their agents, upvote their agents, it actually drives the platform organically and it actually gives more visibility on the agent. The biggest problem from our customers has been that we've had customers come to us and they're like, I don't want to pay for 10 different YC vertical agent companies. That's literally what customers have told me. Because they don't know what's good and what's not.

I've had customers like, I tried out 10 different AI SDR agents. I dropped all of them in three months, all 10 of them. I'm like, why? They're like, we were paying for each one a separate SaaS fee, and I don't want to do that in the future. So if you're a small business who's not knowledgeable about this space, you don't have spend to hire an AI dev internally.

You don't have the budget to go to a consulting company. A platform like this is very essential where people have like evaluated agents. There's been upvotes on it. There's visibility on it. There's eval on it.

And, you know, that's all community driven. Again, memory store framework. We're very, very bullish on long term memory. There's only one other team out of Stanford that has done this. They're called MemGPT, the Leda team.

And we're the other ones tackling this. Again, it helps reduce token usage, enabling reliable swarms. And basically, what's a swarm? You have six different agents orchestrating amongst themselves. or more.

Agent rentals and monetization. Again, the reason I'm building this is so everyday people can make money on this. Enterprises have the budget.

Like, they'll go to Mark Benioff. His entry point for agent force is $50,000 a year without cloud spend and on-prem solutions. So yes, he has his enterprise clients, but majority of American businesses don't have that kind of like spend. And so we want to build a platform we can rent out agents to local businesses. And there are a lot of local businesses that want to start automating their workflows, but they just don't have the capacity, the budget or the time.

And that's why we're going down this route. So let's see if the internet works.

Okay, so it did not pull my demo. Let's see if I can pull it up. Well, I guess I'll have to, it did not pull my demo.

Yeah, so I don't know where my demo went. I'll just pull it up in a second. This is basically the basic website.

And basically what's going to happen is if you sign up for our platform, you'll be able to like integrate your agents, build them. And yeah, one second. I don't know where the demo went.

Yeah, so while I bring up the demo, these are some of the use cases we've seen over the past few weeks. We've seen had a customer come to us to automate customer support. And it was actually very interesting.

This was actually a marketing agency. And they came to us because they have clients where they're doing SEO marketing for their clients. On the client side, they're charging $5,000 to $10,000. They came to us.

because they want to automate that and get rid of their staffers that they're doing billable hours for in the middle. Another team that came to us was doing Reddit sentiment analysis. And what that means is I had a growth marketer at a SaaS company and they came to us and they're like, right now we're doing like local Python scripts manually to pull keywords on Reddit so we can actually plug our product in front of editors. We're like, okay.

But now they actually have built an agent that can do Reddit sentiment analysis and automate that process. So what that agent does is it has actually automated the keyword analysis on Reddit. It'll pull in that sentiment analysis. It'll rank the keywords in terms of mentions.

Then it'll take those keywords and generate 10 headlines on each keyword that performed the best. Then the next agent will go and write 10 articles, AB test 10 articles at each headline. That's what they've done with that agent. Last one is very interesting.

So I come from the world of consulting. And a lot of our work is competitor analysis in the beginning.

So we had a really boutique consultancy come to us where they wanted to actually build an agent swarm to do competitor analysis. Again, boutique agencies, again, don't have the budget for enterprise. Like Salesforce has the enterprise budget. They don't have that kind of spend. So that's what they did.

Conclusion

And I found my demo. So I'm just going to pull it up. Yeah, before I pull up the demo, again, why are we doing this? Because we see a future where millions of agents will be used, monetized by millions of people worldwide.

And we are here to provide access and level the playing field. And before I just hop over to the demo, you can actually get access to our beta list today. So you can just sign up and just add yourself to there. And a shameless plug, we also have a community on Discord. where we actually have a lot of people that are AI researchers, scientists, builders, developers.

We have a couple people actually from OpenAI in the community too. So if you're like looking to learn, even if you're like non-deaf, just come in there and you can interact with folks in there. And yeah, I'm just going to pull up the demo now.

Yeah, so we basically are going to allow our users to determine the price, and it's going to be a marketplace model. So we've had users that are telling us that they want to price their agent per use at, let's say, on a monthly basis, let's say $10 to rent that agent, and you get a certain amount of uses. And then we've had users tell us that I'm just going to rent it out per day at $2 a day. We don't set the pricing. We take a transaction fee on the agent that they rent out.

But if you're just only building and you just want to use as a business, you just want to build on platform, we're very similar to the coding agents out there like her. So we have a base plan. Do you do a review for it, right? Just to kind of make.

Yes, we definitely want to review agents internally. I don't want agents. Yes, we are actually going to have upvote system. So you know how Product Hunt and GitHub has implemented organic loops to evaluate your repos and your products.

We're actually implementing that. And I'll show you that right now. Exactly. It's community driven because then you can't lie. People are going to tell the truth.

People can monitor each other's agents and they get to vote on it. And this is just a mock demo, guys. So you see how people have upvoted here? This is, again, something I built like two days ago.

I was half asleep. But this is basically an example of basically what's going to happen. You can list your agent. People will score it and rate it.

And it will also show how many times that agent got used. So then let's say you want an agent for, let's say, content creation. Well, this agent has 4.9 stars, and it costs 150 credits. These prices are set by the users.

It's something that we're very bullish on because you have to remember, this crowd is a lot of technical people. The average Joe on the street doesn't know the difference between the API call, LM model, and an agent. So they're going to go on this platform. They're just going to rent out the top rated agent. And I think that's going to be very key for getting a lot of consumer adoption globally.

So you can also manage your agent. Let's say you've built 10 agents. You can manage it internally.

It'll actually show you how many runs you'll have per agent. And we have the memory side too. Every time you have a memory loop, it'll actually show that interaction.

We actually also display how many tokens you've run on each memory because we want to be very transparent with token usage. A couple of problems have emerged in the industry where a couple of the coding agents, people are running out of their tokens in two days. It's a repeated problem we've had with our customers.

History, so if you have workflow history, you ran an agent and it completely failed and your boss is like, why isn't it running? You can show it here. Well, this is why it didn't run.

Again, it's a lot of transparency. And then the builder, you could actually add a visual prompt. Let's say SEO, build me agent for keyword analysis.

So it's going to be visual, and you can edit it visually. Right now, this is very basic. In the future, you'll actually have a chat interface.

So you can actually prompt it with NLP internally. But yeah, very basic demo. We're not even live yet.

But feel free to sign. I'm actually talking to people every day and getting their use cases. So I would love to hear about your use cases. Thank you, and sorry for the tech hiccup.