I Went to the Biggest Agentic AI Conference... Here's What I Learned

[music] All right guys, here in the Hilton in Midtown, Manhattan for the AI agent conference. Definitely some very interesting companies I'm seeing on the rise at this thing. Going to try to talk to as many interesting people as possible and let's see how it goes. All right, so what are we doing here? All right, we are at the Bright Data Hexagon. Bright Data has recently partnered with Battle Bots and they are helping host a worldwide robot tournament. And to celebrate that, we've got a mini Battle Box arena here at the Bright Data booth. And we're about to fight robots. You're about to fight robots, actually. Let's see how I do. I've I've only seen this on TV, never done it. All right. Drivers, are you ready? Here we go. Three, two, one. Fight robots, fight. We've got the big Death Roll robot. The minute it's 10 seconds left, Scorpio slides up on Death Roll. I'm not doing too hot, guys. All right, I I got him in a corner. All right, here with Peter Day from Super Set. What exactly do you guys have going on? So at SuperSet we're a startup studio. Um we exist just to skew the odds for startups. So everything we do is about giving founders an unfair advantage in market. So it's everything from providing capital to space to supporting services and and everything else that it goes into kind of building a great company. Okay, so who are your ideal customers that should be looking out for SuperSet right now? People who have this kind of irrational need to change the world. Um you know, SuperSet was founded by all four general partners. We've all started companies. We've all got that chip on our shoulder that forces us to do things that's slightly insane. Um we're looking for more people like us who are going to, you know, think they can change the world. So basically everything you guys needed when you had your startups you said, "Why don't we create a company that will provide all those needs basically?" That that's exactly it. You know, all the distractions, the annoying things that we don't think are really about finding product market fit, that's what we try and take away from founders. Um and then we give them everything else they need to to really so they can spend every waking hour thinking about product solution fit, product market fit. Wow, awesome. We're going to see Peter speak a little later. But uh SuperSet, if you have a startup, check these guys out. I'm here with Elan, the CEO of Plural I. What do we got here? Yeah. Hey, nice to meet you. I'm Elan and we are building an evaluation observability platform for AI agents to enable and allow companies that are building agents to turn their agent from impressive demo into a really reliable and impactful AI agent. Okay, so who are your target customers here? So we mainly focus on on companies that are building AI agents where the cost of error is really high. So mostly our customers come from the finance working with leading banks, also big enterprise which have a customer-facing agents where every error and failure points can have a serious brand damage or really can create for them high risk. So these are our focus. Awesome. Check them out. Plural I. Okay, this guy's uh AI agent just got me this coffee. His name is Rajesh for Fetch.ai. What do you guys do exactly? Yes, we are building the infrastructure for agents at scale and these agents represent people, businesses and services. And people can access all these agents from our app ASI one and from ASI one basically they interacted with our agent to get you know all coffees and this is a business that has a deployed agent on agent verse and now people are getting coffee by just interacting with the agent instead of standing in a line. They're just talking on ASI one and getting a coffee. Okay, so who are your main customers like ideally going to be? Yes, it could be from the consumer side ASI one will get all the consumer side customers and then on the agent or side we get we have Fetch business where small and medium businesses can host their agents that can represent the businesses and also individual developers who can monetize their AI agents. Wow, awesome. Fetch.ai check them out guys. Thank you. All right, here with Alex. He's the COO of you.com. I've been seeing this company a lot of places. Want to tell us a little bit about what you guys do? Absolutely. Web search APIs for AI. So you know how your models hallucinate? They don't when they have the right data and that's what we provide. So fresh, accurate, fast intelligence from the web right to your API. Where are you guys at in your progression? What's kind of like the next moves for you.com? Yes, we were one of the early innovators in this space. Our founder was one of the first to think about putting web search with an LLM and marrying the two concepts together. So we've been at this for a while. We got to run web index. We built out a number of data sources and we're all about providing fresh, accurate answers and information to agents and users worldwide. We are about 120 people, mostly Bay Area, a little bit in New York, and some other hot spots. Um we're a series C company. Like many, we're a unicorn. Uh we had explosive growth in the last couple years, and really going for more. Wow. Awesome. Check these guys out. you.com. I'm here with Kevin. He is a senior solutions engineer with Fiddler AI. Just gave a speech on observability that we watched. Will you talk a little about Fiddler and and what about some modern observability um things you have going on right now? Yeah, so uh Fiddler, you know, the talk was about observability, and that is a really hot topic right now, especially with agents, because, you know, the um agents that customers are using are really complex. They can have hundreds of different spans, individual actions agents are taking, different tool calls or MCP calls. So, being able to monitor that effectively in terms of is the agent producing good responses? Are they good quality outputs? Um are they safe, right? Is it leaking PII, or are they making uh safe decisions, or are they leaking secrets? Um all of this is part of, you know, Fiddler's uh mission to be the control plane for uh agents, as well as predictive models. Our roots are in machine learning uh and predictive models, as well. So, we have uh that observability layer that we uh unify into an executive dashboard capability. So, you see uh a lot of opportunities in observability coming forward with AI? Yeah, absolutely. So, you know, from customers that we talk to, uh a lot of them are still releasing or thinking about release these releasing these agents into production. So, it's really critical as they're releasing that to have the confidence that, "Hey, this agent's doing well. It's not doing anything malicious, and it's safe." Um in order for them to move forward. And that's really important for, you know, governance teams, as well, and that's part of what Fiddler does. Uh as part of the control plane is have that evidence layer to show that you're compliant with regulations. Yeah. Awesome. Great stuff. Check them out. Fiddler AI. Okay, we're here with Ariel. He is the chief product officer over at Bright Data. These guys are absolutely blowing up. They're all over this conference. When you tell us a little bit about the product space when it comes to the agentic AI arena? Yeah, well, agentic AI is interesting because it really depends on the quality of the data that you feed it into the into the agent. In fact, this very little that can be improved in terms of the intelligence of the agent. Most of the mistakes that we see come from bad data. So, what we call garbage in, garbage out. And web agents are only useful I mean, AI agents are only useful when they operate with fresh data from the internet. And as you know, the internet was built for humans. It's not something that's structured or easy to work with for computers. So, what we do in Bright Data and the products that we develop are actually taking all that messy data and structuring it and making it easy to ingest and token economical for AI agents or You have this thing for us. So, we're so early in this space now. What are some of the biggest challenges? Obviously, it's the messy data, but what specifically? So, many of the websites that we're all using are websites that agents will still try to go to. But, many of those websites will misinterpret AI agents, which are actually taking actions on behalf of humans, as bots. So, they will put up all sorts of defenses and things that make make life difficult for those bots. Maybe captures, maybe misinformation, things like that. And what we do in Bright Data is we actually offer kind of uninterrupted access to these to these AI agents so that they have the clear you know, reliable view of the web data as it should be. Okay. So, you're saying basically the agents try to go to a website and it's getting captured every time and it's causing a lot of issues right now? Yeah, because I I I've decided as a human that I want my agent to do ABC for me. But, that agent could be misinterpreted by the website to be a malicious bot that's trying to overload the system or do all sorts of malicious things. That's clearly not the intent. So, the tools that we provide in Bright Data make sure that the bot will actually see the website without any interruptions without adding in a CAPTCHA and it will see the same data that a real human would see if it would if they were to go to that website. Wow, incredible. Well, check them out. Bright Data, these guys are at the cutting edge of all the agentic AI stuff. Thanks a lot, Or. I'm here with Alexandra. She is a developer advocate for Databricks. See Databricks all over the place. Tell us a little bit about what exactly what you guys do. Databricks is data and AI platform and what we do, we support end-to-end workflow with your data starting from data ingestion up until agentic workflows. As all this conference is about, it's all about agentic workflows. Yeah, and then obviously a lot of people know what a developer is. What is a developer advocate? What How is that different? Developer advocate is a person who actually talks to developers and who is actually tries to see where they struggle, what can be done better, understand their experience and try to help them to make it better and uh how our platform can suit developers better. Who should consider a developer advocate job versus a developer job? Like what type of person? Developer advocate uh it's a very social job, you know? Like you need to talk to a lot of people. You need to uh um you need to know your stuff. So, you need to be technical enough to understand it. But I think like the biggest skill is essentially compassion. Like to be able to talk and to understand what they do, how they do, and like a general curiosity. I think it's not like a technical skills that what they do matter. They matter a lot. But like it's not about only technical skills, but also about like the whole social component of it. Great. Well, check them out. Databricks, obviously they're a big dog in the space and uh this is the type of uh job you want, developer advocate, check it out, too. I'm here with Gunja. He is the Chief Revenue Officer at Bright Data. He just uh moderated a panel on trust in uh web agents. So, uh why don't you tell us a little bit about what you guys are up to right now? Bright Data is the biggest web data collection company in the world. What we do is we provide reliable, trustworthy data in a structured format to everybody who's building agent AI ecosystems around the world. Yeah. Any tidbits you want to share from the uh the talk you just moderated on trust? Yeah. It's so important to see that people are looking at all possible angles when they're building an AI agent. So, web data is very fragmented, very contradictory, hard to get to, and not always something that you can easily access. So, a lot of people give up, they take shortcuts, but to build the right framework, like you heard from IBM, you heard from Glean, and to be able to have that right kind of framework for you to build your web agent so that it the results it gives you are trustworthy from an enterprise perspective is super key. Wow. Awesome. Well, Bright Data, these guys are about to blow up, and this man is uh about to blow up their revenue as their CRO. So, thanks, Gunja. Thank you. All right, coming on day two, Agent AI Conference. Had a great day one. Met some interesting people uh doing some really cool stuff in the Agent AI space. So, let's see what we find today. There's a couple more people I definitely want to interview. Uh specifically SailPoint, cuz we have done a lot of content on that. So, um hopefully going to get a hold of them, and uh try to get some of the speakers after they uh go on stage. So, we'll see how it goes. the funny thing, but it actually is a perfect analogy for where we are with AI today. We give instructions, we don't always get the right outcome, and here's the problem. AI is accelerating how fast we build, change, and apply software, so the bottleneck is shifting because it's no longer about building or coding, it's about validating the output. So, the question becomes, how can you do that at AI scale? I'm here with Nick from Leapwork, just gave a great talk on stage that we'll show you about validation in the AI space. So, Nick, tell us about what you guys are up to over at Leapwork. Yeah. So, for the last decade, Leapwork has been the core foundational component for many enterprises to do continuous validation of everything they release, whether it's packaged software, SDLC that they're releasing internally, to ensure that systems end-to-end work across user and customer journeys. What are some of the biggest challenges you guys face in this space? It's no surprise, AI and the advent of AI has changed and shifted so many things in the industry. From a software perspective, it means a lot of code is being deployed and released much faster than ever before. What that means for us is the more code being released, the more changes happening at lightning speed, traditional approaches to QA can't keep up. So, we have just created from the ground up a cloud-native, AI-native solution to address these types of challenges. Wow, awesome. Check these guys out, Leapwork. I'm here with Shawn. He is an account executive with this company, Onix. They're in the what, GRC space, security? AI security and governance space. Okay. So, what are some of the biggest challenges with cybersecurity and AI right now? Yeah, so right now we're seeing a lot of um organizations getting pressure from their board, their senior leadership to adopt AI quickly. AI is It's really fast, right? And from a security perspective, you can't just rapidly deploy agents across your entire environment, right? Without having a plan in place or strategy in place. So, that's really where we focus on. So, what we do is we provide real-time visibility and inventory of all of your AI assets for a customer's environment. Once we provide a real-time visibility and inventory of all their AI assets across their cloud, their SaaS, their endpoints, that's where we can actually part apply the guardrails around, you know, you know, the runtime protection and things like that. So, basically, what we do is we look at things like prompt injections, we look for sensitive data leakage, or you know, if AI if AI agents are actually acting not anonymously or maliciously, we can surface all of that through our platform. Wow, great. All right, well, if you're deploying AI agents and they're unsecured, talk to this man at Onyx. Sounds good. Appreciate it. Good night. Cheers. All right, here with Jimmy from SailPoint. Anyone who watches this channel knows this is one of topic I talk about a lot. These jobs are nearly impossible to fill, to find SailPoint experts. Jimmy, what are you guys up to with AI though right now? Hi, thanks thank you. Today I'm at the AI conference. We're meeting a lot of AI builders and having a great time here. So, what SailPoint doing with AI today is we focus on same exactly our bread and butter identity governance. So, now we focus on not just humans, but we also focus on the non-human and the AI agents. So, we provide the full visibility and the control able to provide revoke and provide access to these agents. What's the biggest challenge you guys have in the security space in AI right now? So, a lot So, the challenge is a lot of customers, they don't know how many AI agents that they have out there, what access that they have, and what exactly they're doing. So, where SailPoint comes in is we provide that information for you. You're able to control that. Yeah, no. In my recruiting business, SailPoint jobs are brutal to fill. There's a very major lack of talent. How can someone get started with you guys as well? You can come to our career page. Uh I think sailpoint.com/career and we have we have a list of uh open positions. All right. Get some experience in SailPoint because there's just a lack of talent right now. Thanks a lot, Jimmy. All right. Thank you.