WEBVTT NOTE Sentence-level transcript of https://www.youtube.com/watch?v=kIlVNGZZr8o NOTE One cue per sentence. Cue ids are the line anchors on /transcripts/kIlVNGZZr8o.html. A cue ends where the next begins, or 2 s after its last word. s1 00:00:01.309 --> 00:00:03.309 [music] s2 00:00:12.560 --> 00:00:13.520 Hi everybody. s3 00:00:13.520 --> 00:00:16.160 It's really really great to be here on stage at AIE New York. s4 00:00:16.160 --> 00:00:17.480 This is kind of awesome. s5 00:00:17.480 --> 00:00:21.360 Um today we're going to talk a little bit about the Agentech SecMaster. s6 00:00:21.360 --> 00:00:31.880 But before I dive into things like introductions, I want to talk about something that's been on my mind recently, especially as I was preparing for this talk, which is finance has gotten kind of weird, right? s7 00:00:31.880 --> 00:00:39.880 Like when I was growing up, finance mostly meant public markets, you know, trading floors, ringing bells, things like that. s8 00:00:39.880 --> 00:00:43.080 And that's not really what happens anymore. s9 00:00:43.080 --> 00:00:47.800 There has been a huge influx of capital investment into private assets. s10 00:00:47.800 --> 00:00:50.160 And some of these numbers are kind of staggering to me. s11 00:00:50.160 --> 00:00:58.480 Apparently last year, there were $240 billion of secondaries volume in 2025, which is up 50% from 2024. s12 00:00:58.480 --> 00:01:00.800 I'm sure in 2026 it's even more than that. s13 00:01:00.800 --> 00:01:08.960 And people are now forecasting that we're going to do $32 trillion of alternative assets under management by 2030, which is only 3 years away. s14 00:01:08.960 --> 00:01:10.400 Now, why is that happening? s15 00:01:10.400 --> 00:01:12.400 Why is this the big story in the finance world? s16 00:01:12.400 --> 00:01:19.880 Well, some of this is because AI seems to increasingly be the only growth area in the broader economy. s17 00:01:19.880 --> 00:01:25.920 So more and more people are holding on to and investing into all of the things that feed the AI world. s18 00:01:25.920 --> 00:01:29.920 That's land, energy, data centers, GPU clusters. s19 00:01:29.920 --> 00:01:31.720 And of course secondaries, right? s20 00:01:31.720 --> 00:01:35.000 Everybody is holding on to OpenAI and Anthropic secondaries. s21 00:01:35.000 --> 00:01:35.920 They're everywhere. s22 00:01:35.920 --> 00:01:39.160 All of these private AI companies, everybody's trying to get a hold of them. s23 00:01:39.160 --> 00:01:41.240 But it's not just an AI story. s24 00:01:41.240 --> 00:01:46.600 Some of this is also a regulation story and a story about seeking alpha. s25 00:01:46.600 --> 00:01:50.040 High growth companies are finding less and less reasons to go public. s26 00:01:50.040 --> 00:01:57.040 They would rather just raise a series F. More control for the founders, less paperwork to do with the government. s27 00:01:57.040 --> 00:02:07.652 And if there are these massive well-funded firms that are taking over and squeezing the juice out of public markets, well, you're just going to start looking for edge somewhere else. s28 00:02:07.652 --> 00:02:10.972 As with all things, I think Matt Levine said it best. s29 00:02:10.972 --> 00:02:12.572 He actually wrote this a year ago. s30 00:02:12.572 --> 00:02:20.412 Private investment firms have become huge, and the legal distinctions between public and private markets have become way less important. s31 00:02:20.412 --> 00:02:26.812 If an increasing share of the economy is in private assets, investors aren't able to just index. s32 00:02:26.812 --> 00:02:33.212 You can't get the market return by buying every public stock if half of the market is actually in private companies. s33 00:02:33.212 --> 00:02:34.852 So, you end up with this cycle. s34 00:02:34.852 --> 00:02:42.292 There's more investment in private assets, which means more people have to invest in private assets just to make sure that they're getting the returns from the wider market. s35 00:02:42.292 --> 00:02:46.412 Now, one thing that's really frustrating is dealing with private assets. s36 00:02:46.412 --> 00:02:48.532 It's actually really, really hard. s37 00:02:48.532 --> 00:02:50.132 How do you manage private assets? s38 00:02:50.132 --> 00:02:53.652 How do you keep track of things like how many shares you own or what your strike price is? s39 00:02:53.652 --> 00:02:58.372 I mean, I'm a startup founder, and I find that frustrating just for my own little bit of stock. s40 00:02:58.372 --> 00:03:02.732 And if you invest in a plot of land next to a hydro plant, things get really weird. s41 00:03:02.732 --> 00:03:07.172 Where is all of the information about how much you paid and what your rights are? s42 00:03:07.172 --> 00:03:09.812 Well, unfortunately, it's in PDFs. s43 00:03:09.812 --> 00:03:13.252 It's in a box folder or in an Outlook email. s44 00:03:13.252 --> 00:03:14.612 It's in a Zoom call transcript. s45 00:03:14.612 --> 00:03:21.052 Sometimes, it's on a napkin with a hastily scribbled signature from a bar overlooking Columbus Circle. s46 00:03:21.052 --> 00:03:30.812 Welcome to the nightmare world that is managing private asset data, something that more and more people in this room and in New York in general have been dealing with over the last couple of years. s47 00:03:30.812 --> 00:03:32.172 My name is Amol Kapoor. s48 00:03:32.172 --> 00:03:34.692 I am the CEO of a company called Noriagentech. s49 00:03:34.692 --> 00:03:43.852 We build cloud agents, and today we're going to talk about how you can use cloud agents and AI in general to make it easier to manage, record, and audit a company's investments. s50 00:03:43.852 --> 00:03:46.852 In particular, a company's security master database. s51 00:03:46.852 --> 00:03:48.852 Now, my background is actually in AI. s52 00:03:48.852 --> 00:03:51.372 I've been an AI engineer for about 12 years. s53 00:03:51.372 --> 00:03:53.172 Uh four and a half of those years were at Google. s54 00:03:53.172 --> 00:03:57.652 So, funny enough, I am not the best person to actually talk about what a sec master is. s55 00:03:57.652 --> 00:04:09.614 But, I'm very lucky to be joined by my CTO and co-founder Cliff, who actually has spent some time in back office finance here to talk to us a little bit about what a sec master is before we dive into how AI can solve those problems. s56 00:04:19.254 --> 00:04:19.454 Great. s57 00:04:19.454 --> 00:04:24.894 So, I think it's helpful before we before we talk about where this is a hard problem, I think we should get a lay of the land. s58 00:04:24.894 --> 00:04:31.454 We can talk about a simpler example with public equities, then we can tackle some of the complexity that comes with private assets. s59 00:04:31.454 --> 00:04:35.334 So, every asset manager right now currently runs on a sec master. s60 00:04:35.334 --> 00:04:37.814 So, let's define that. s61 00:04:37.814 --> 00:04:42.614 Asset managers need to know what exists in their investment universe or in their tradeable universe. s62 00:04:42.614 --> 00:04:48.334 This is a critical piece of back office, brings together engineering, operations, and data. s63 00:04:48.334 --> 00:04:55.454 You can think of this as just a master database of all securities or instruments that represent the assets they're holding. s64 00:04:55.494 --> 00:05:03.254 For the engineers, you can think about this like any master database, it's the source of truth, and usually the starting point for most of the business critical analysis. s65 00:05:03.254 --> 00:05:11.734 An investment that looks like risk management, P&L accounting, regulatory compliance, all these functions that are critical for asset managers. s66 00:05:11.734 --> 00:05:16.574 So, say two different sources of data disagree about particular values. s67 00:05:16.574 --> 00:05:22.174 Ideally, that conflict doesn't get flagged before it reaches the sec master. s68 00:05:22.174 --> 00:05:33.054 You can go back to the sources and pinpoint either the divergence in the particular value, or sometimes it's a logic problem and you can examine how each one maps to that particular security. s69 00:05:33.054 --> 00:05:40.894 So, when done correctly, upstream problems are fixed before they contaminate all the business critical analysis we just talked about. s70 00:05:40.894 --> 00:05:46.814 So, I like to think about this as an organizational primitive for how you handle data to enable all the consumers of data at your company. s71 00:05:46.814 --> 00:05:50.694 This layer separates the data producers from the data consumers. s72 00:05:50.694 --> 00:06:02.814 So, whether you're working on a complex medallion data lake house that has many different tables, or it's just a single denormalized source of truth, this is the bottleneck where you can enforce the standards and the visibility s73 00:06:02.814 --> 00:06:05.294 and the quality checks. s74 00:06:06.655 --> 00:06:16.775 So, depending on where you invest, you need to take this simple concept and match it up with how everyone else in the world who invests in this particular asset class, s75 00:06:16.775 --> 00:06:24.095 potentially speaking different languages languages, settling in different currencies, and operating in different regulatory regimes. s76 00:06:25.055 --> 00:06:29.535 So, for public assets, you benefit from a couple different points of centralization. s77 00:06:29.535 --> 00:06:35.855 For the needs of exchange and transactions, you have data vendors who will provide different identifiers. s78 00:06:35.855 --> 00:06:39.975 So, these might be like Bloomberg figgies or LSE RICS. s79 00:06:39.975 --> 00:06:41.735 Then there's also regulatory needs. s80 00:06:41.735 --> 00:06:48.255 And though that point of centralization causes numbering agencies to offer a different set of identifiers, like ISINs or CUSIPs. s81 00:06:48.255 --> 00:06:53.935 So, you're able to use this information to map anyone's view of a particular asset into your own view. s82 00:06:53.935 --> 00:07:02.575 And that's the only way you can understand your positions, like modeling risk or reporting holdings, joining alternative data, creating financial models. s83 00:07:03.255 --> 00:07:05.175 So, let's take a very simple example. s84 00:07:05.175 --> 00:07:08.295 You hold some mega-cap publicly traded stock like Google. s85 00:07:08.295 --> 00:07:11.495 So, already if you know the tickers, is this GOOG or is this GOOGL? s86 00:07:11.495 --> 00:07:15.095 Because there are two different share classes traded on the same exchange. s87 00:07:15.095 --> 00:07:20.495 Let's say you need to join in some data that exists only at the level of the legal entity. s88 00:07:20.495 --> 00:07:24.375 Well, if you're indexing based off of ISIN, now you can't join that anymore. s89 00:07:24.375 --> 00:07:34.255 Or maybe you only trade one share class and you only hold one, but you need to trade it across multiple different exchanges because you have different funds for the needs of multiple s90 00:07:34.255 --> 00:07:36.895 multiple funds or multiple investors. s91 00:07:36.895 --> 00:07:39.815 So, already this is getting pretty complex. s92 00:07:40.055 --> 00:07:44.455 I like to think about this kind of like a daytime library as a comparison for the engineers. s93 00:07:44.455 --> 00:07:48.215 It seems very simple when someone solved the problem well. s94 00:07:48.215 --> 00:07:55.735 But the moment you're either, you know, working with a new language that hasn't implemented good daytime yet, or you just need to make some modifications and peel back how it works inside, s95 00:07:55.735 --> 00:07:59.175 it's all edge cases and complex ontologies. s96 00:07:59.175 --> 00:08:01.855 So, this is what that Google example might look like in practice. s97 00:08:01.855 --> 00:08:08.857 Here's a denormalized table with many rows for that one equity we want to uh able to now analyze. s98 00:08:08.857 --> 00:08:18.017 So, if some data comes in, even if it only has a subset of identifiers or exists at a different granularity than your Sec Master, the code can still map it. s99 00:08:18.017 --> 00:08:20.497 So, here's where it gets interesting. s100 00:08:20.497 --> 00:08:26.697 If the code if the data coming in doesn't have regular identifiers, how do you map it? s101 00:08:26.697 --> 00:08:30.377 It's actually pretty clear how AI helps in the first case, right? s102 00:08:30.377 --> 00:08:32.977 It's very easy to reason about this data with structure. s103 00:08:32.977 --> 00:08:40.730 The coding agents are increasingly good at writing code and if you have the source of kind of gold standard ontology about the business critical data for your company, s104 00:08:40.730 --> 00:08:41.057 [snorts] s105 00:08:41.057 --> 00:08:43.537 AI can enable new use cases very easily. s106 00:08:43.537 --> 00:08:52.257 But with unstructured, potentially not identified data sources, you have a huge burden of manual verification and manual mapping. s107 00:08:52.257 --> 00:08:57.337 So, this is where the happy path breaks down when private assets come into the Sec Master. s108 00:08:58.697 --> 00:09:08.377 So, when you're buying the Google stock, you know what you're paying, you know what you're getting, and going into the transaction, everyone has a shared language of different identifiers that they can speak. s109 00:09:08.377 --> 00:09:18.377 Let's imagine you're going to lend some money to a company building a power plant near some data center developments or you're investing in land near the power plant for potential future s110 00:09:18.377 --> 00:09:20.177 uh development itself. s111 00:09:20.177 --> 00:09:23.457 There's no identifiers everyone shares going into that deal. s112 00:09:23.457 --> 00:09:27.497 There's no one that regulates the entire structure of information you might get out the other end. s113 00:09:27.497 --> 00:09:34.497 And there's no one data vendor that has a single API where they've thought through the ontology of these entities in particular. s114 00:09:34.497 --> 00:09:41.057 Everything from the repayment schedule to the collateralization, all of this lives in one-off agreements or amendments. s115 00:09:41.057 --> 00:09:42.657 It's not fungible. s116 00:09:42.657 --> 00:09:47.137 The asset is captured and defined entirely by the documentation. s117 00:09:47.177 --> 00:09:57.777 And this is interesting because now the data scales almost one-to-one with the assets and the burden to scale the the Sec Master itself is constantly increasing. s118 00:09:59.257 --> 00:09:59.537 Great. s119 00:09:59.537 --> 00:10:02.057 So, I'll pick up back here. s120 00:10:02.057 --> 00:10:03.817 Now, Cliff makes a really good point. s121 00:10:03.817 --> 00:10:06.537 The private assets are an absolute pain to manage. s122 00:10:06.537 --> 00:10:08.937 They just don't fit into the Sec Master all that well. s123 00:10:08.937 --> 00:10:12.898 But all that information still needs to actually be collated in some way. s124 00:10:12.898 --> 00:10:17.738 It's not like your audit requirements and your cleaning requirements and your standardization requirements disappear. s125 00:10:17.738 --> 00:10:21.418 If anything, it's actually a lot worse now because it's just so much less standardized. s126 00:10:21.418 --> 00:10:26.738 You can't rely on the people who are doing your risk modeling to know whether or not that data is ground truth. s127 00:10:26.738 --> 00:10:30.778 You certainly don't want them to be the ones going through and munging through all your documents. s128 00:10:30.778 --> 00:10:38.618 So, somehow the data that lives in that PDF somewhere, that information needs to make its way back into your SecMaster. s129 00:10:38.618 --> 00:10:42.578 And right now, that is just an extremely high labor task. s130 00:10:42.578 --> 00:10:47.018 I mean, generally it just takes a lot of people hours and a lot of domain expertise. s131 00:10:47.018 --> 00:10:49.338 I'm also going to really quickly move this mouse out of the way. s132 00:10:49.338 --> 00:10:51.018 It's been bothering me. s133 00:10:51.018 --> 00:10:52.618 All right. s134 00:10:52.618 --> 00:10:55.218 So, I want to ground this in a specific example. s135 00:10:55.218 --> 00:11:01.338 We work closely with a Fortune 200 asset manager here in New York, one of the benefits of being New York local. s136 00:11:01.338 --> 00:11:10.138 They recently started owning a bunch of private assets, who hasn't, and it was causing a lot of problems for their data engineering teams, the people who were maintaining their SecMaster. s137 00:11:10.138 --> 00:11:15.978 Now, when any assets in their SecMaster changed, they had to make sure that change was correct. s138 00:11:15.978 --> 00:11:17.138 So, they had a system. s139 00:11:17.138 --> 00:11:21.258 They did a day-over-day change analysis of their table. s140 00:11:21.258 --> 00:11:31.018 If you have a table with 50,000 assets and each one of those assets has something like 200 columns, even infrequent changes are going to happen on some daily cadence. s141 00:11:31.018 --> 00:11:39.418 And every single one of those changed cells was then sent to a team of people to triage because, again, all of this data is super unstructured. s142 00:11:39.418 --> 00:11:41.698 It just lives in documents somewhere. s143 00:11:41.698 --> 00:11:52.538 So, that team was then responsible for going through docs and going through various data sources to determine if any given change was good or bad because upstream data providers s144 00:11:52.538 --> 00:11:57.298 may have fat-fingered something or may have gotten something incorrect or they may have just been conflict. s145 00:11:57.298 --> 00:12:03.618 And I want to highlight that this is not just every row, not every column, literally every single cell. s146 00:12:03.618 --> 00:12:12.698 So, if an asset changed price or if an asset changed its name or if a company went bankrupt or if it changed ownership or you renegotiated a contract or whatever, s147 00:12:12.698 --> 00:12:17.779 every single one of those changes in that investment universe would have, on average, an hour of work behind it. s148 00:12:17.779 --> 00:12:23.139 An hour of somebody somebody's time, somebody that you're paying to go and just lunge through data. s149 00:12:23.139 --> 00:12:26.619 That is hundreds of thousands of hours every single month. s150 00:12:26.619 --> 00:12:29.459 Hundreds or thousands, sorry. s151 00:12:29.459 --> 00:12:34.179 So, we're at AIE, how would an AI engineer solve this problem? s152 00:12:34.179 --> 00:12:35.259 Two words. s153 00:12:35.259 --> 00:12:37.259 Cloud agents. s154 00:12:37.259 --> 00:12:38.379 What are cloud agents? s155 00:12:38.379 --> 00:12:40.499 This is more for the finance folks in the room. s156 00:12:40.499 --> 00:12:52.139 In simplest terms, a cloud agent is a coding agent like Claude Code or ChatGPT, but instead of running on your local machine, it lives and runs in a remote environment instead of on your computer. s157 00:12:52.139 --> 00:12:54.659 Because it lives on the cloud, we can do a lot of really interesting things. s158 00:12:54.659 --> 00:13:06.339 We can uh assert a bunch of audit requirements, we can get auditability and trans um transparency, we can get observability, and more importantly, we can have it spin up and do work on a schedule without a person being involved. s159 00:13:06.339 --> 00:13:10.339 For example, when new data hits the SEC master. s160 00:13:10.339 --> 00:13:22.299 So, whenever there is a change in the investment universe for this particular asset manager, a Nori cloud agent will spin up, in this case running a locally deployed open source model under the hood, so all of the data stays on prem. s161 00:13:22.299 --> 00:13:26.779 There's no concerns about information getting out of the underlying environment. s162 00:13:26.779 --> 00:13:32.939 The agent here has explicit instructions to figure out how the source uh figure out the source of this change. s163 00:13:32.939 --> 00:13:41.819 And it also has access to a few useful tools, Box for PDFs, public APIs like Bloomberg, and even a small segment of email. s164 00:13:41.819 --> 00:13:50.179 The agent then assembles an explanation of the change, gets citations for why that change happened, and then provides a recommendation for whether to keep or discard it. s165 00:13:50.179 --> 00:13:55.459 And finally, it pings the owner of whichever cell changed in Slack for review. s166 00:13:55.459 --> 00:14:03.739 So, you have this agent that is operating basically entirely autonomously, providing tasks for a human that the human just has to say yes or no to. s167 00:14:03.739 --> 00:14:10.499 And of course, if necessary, the human can go and do further research to make sure that the information provided was correct. s168 00:14:10.579 --> 00:14:13.459 Now, you still need a human to review those changes. s169 00:14:13.459 --> 00:14:17.461 Of course, this is necessary just for accountability among other things. s170 00:14:17.461 --> 00:14:25.941 But, this kind of work dramatically cuts down the response time because most of the previous slowness was just people having to manually search through a bunch of documents. s171 00:14:25.941 --> 00:14:29.941 Agents are really, really, really good at that, and humans are pretty bad at that. s172 00:14:29.941 --> 00:14:37.901 So, if most of the time is just assembling a case, well, you can take something that used to take an hour and make it just 5 minutes of human work. s173 00:14:38.061 --> 00:14:38.901 But, wait. s174 00:14:38.901 --> 00:14:43.181 It's worth pausing just to ask, how does the agent actually get domain expertise? s175 00:14:43.181 --> 00:14:45.581 How does it know what a good change looks like? s176 00:14:45.581 --> 00:14:53.421 I'm concerned that some of the folks in the room might think, "Oh man, I have to go and build this like custom agent infrastructure that's going to do this really, really specific one-off thing." s177 00:14:53.421 --> 00:14:56.501 Well, I don't think that's actually how these things should work. s178 00:14:56.501 --> 00:15:05.941 Agents get information the same way a new analyst does, from written instructions about where documents live, which sources to trust, and how to judge a change, all in English. s179 00:15:05.941 --> 00:15:08.061 Agents are pretty good at self-modification these days. s180 00:15:08.061 --> 00:15:13.381 So, you can change how an agent behaves by just editing natural language English markdown files. s181 00:15:13.381 --> 00:15:22.821 You use a generic agent like Claude Code or Open Code or anything that's open source like Pi, and then you skin those agents with configs like skills or agent MD files. s182 00:15:22.821 --> 00:15:32.821 And most of the customization actually comes from the environment in which the agent is running in, which of course is just about what kind of configs and what kind of credentials you're dropping on the machine. s183 00:15:32.821 --> 00:15:35.821 So, I want to take a step back and zoom out a little bit. s184 00:15:35.821 --> 00:15:45.781 Everything that we've talked about so far, this agentic SecMaster concept, this is essentially a database with a bunch of cleaner fish automatons that are dynamically making sure that everything is up to date. s185 00:15:45.781 --> 00:15:51.421 They can spot problems quickly and proactively surface them and even solve them in some cases, all autonomously. s186 00:15:51.421 --> 00:15:58.341 And one of the things I want to highlight is that this technology is not limited to SecMasters or even data cleaning in general. s187 00:15:58.341 --> 00:16:06.701 It can be useful across a firm because it is a generic primitive in which we are having agents do work that was previously loaded onto humans. s188 00:16:06.701 --> 00:16:14.581 In my mind, the incredible potential of AI tools in finance is that they can make a huge range of tasks proactive instead of reactive. s189 00:16:14.581 --> 00:16:23.102 Right now, every time anything needs to happen, and a human has to spend time and energy evaluating whether or not that task is worthwhile and then sending that work to an agent. s190 00:16:23.102 --> 00:16:28.062 If you ask me, the future of enterprise AI is not use a robot on your laptop to solve a problem. s191 00:16:28.062 --> 00:16:34.622 It's having the robots come to you with tasks, advice, analysis, and recommendations. s192 00:16:35.262 --> 00:16:45.062 I think that in order to make that future a reality, you need to use flexible infrastructure that can be used for any task and make it really, really easy to modify what kind of tasks it's being used for. s193 00:16:45.062 --> 00:16:48.382 Your SecOps master agent shouldn't be a one-off unique agent. s194 00:16:48.382 --> 00:16:57.142 It should actually be running on the same cloud agent infrastructure as your audit agent or your code review agent or any of the agents that your engineers are using for their day-to-day job. s195 00:16:57.142 --> 00:17:01.742 Each of the Each of these agents will have different instructions and tools and triggers, different context. s196 00:17:01.742 --> 00:17:04.582 That, of course, is owned by you, by the company. s197 00:17:04.582 --> 00:17:10.382 But the orchestration, the security, the integrations, all of that should actually just be the same across these different tasks. s198 00:17:10.382 --> 00:17:18.902 And really, in my mind, should just be owned by a vendor because everybody is inventing these problems or solving these problems over and over again in the same organizations. s199 00:17:19.382 --> 00:17:22.622 So, what were the things that we learned by building AI for finance? s200 00:17:22.622 --> 00:17:27.062 I want to leave with three lessons that we that we had that I want to impart on everybody here. s201 00:17:27.062 --> 00:17:31.622 First, if you ask me, AI adoption is primarily a collaboration problem. s202 00:17:31.622 --> 00:17:36.342 People want to buy solutions, engineers want to sell building blocks. s203 00:17:36.342 --> 00:17:37.822 We have to find a way to meet in the middle. s204 00:17:37.822 --> 00:17:52.142 We found the most success when we work with brilliant folks who are experts in their domain, who are able to make the leap, that intuitive leap from cloud agents to wait, this actually fits into a problem that I have right now today. s205 00:17:52.142 --> 00:17:58.222 Second, I think it's really important to aim for 95% automation and it's okay to keep a human in the loop. s206 00:17:58.222 --> 00:18:02.262 People want to use AI to do 100% of everything and this doesn't work. s207 00:18:02.262 --> 00:18:07.382 It doesn't work in software, we've seen that with software factories over and over again, but it really doesn't work in finance. s208 00:18:07.382 --> 00:18:08.742 It's okay to have a human in the loop. s209 00:18:08.742 --> 00:18:10.422 In fact, I think we should lean into it. s210 00:18:10.422 --> 00:18:18.302 There are way more problems that you can solve if you're okay with having one human spend 5 minutes looking at something than trying to have an agent do something 100% s211 00:18:18.302 --> 00:18:21.022 end-to-end every single time. s212 00:18:21.022 --> 00:18:26.103 And finally, I want to point out that the cost of accidental complexity has dropped significantly. s213 00:18:26.103 --> 00:18:34.023 Fred Brooks used the phrase accidental complexity to describe the difficulty that comes from how a system was built as opposed to the problem it solves. s214 00:18:34.023 --> 00:18:37.383 And I think in the back office, a lot of work is actually shaped like this. s215 00:18:37.383 --> 00:18:44.223 Teams record the same data in different ways, and vendors use different formats, mostly because that's just how things evolved organically. s216 00:18:44.223 --> 00:18:56.823 I think agents are really, really good at handling this kind of work, this low-complexity, high-effort work, which actually then allows people to spend more time on the problems that are essentially complex, the problems that are actually hard. s217 00:18:56.823 --> 00:18:58.263 So, that's all I got, guys. s218 00:18:58.263 --> 00:18:59.583 Thank you so much for coming. s219 00:18:59.583 --> 00:19:01.783 Uh before I go, I want to just quickly highlight two things. s220 00:19:01.783 --> 00:19:06.783 One is if you're interested in more talks like this, we run a community called Agentics NYC. s221 00:19:06.783 --> 00:19:08.943 It is New York's largest agentic engineering community. s222 00:19:08.943 --> 00:19:12.783 We run regular events on building with agents for beginners through experts. s223 00:19:12.783 --> 00:19:20.263 And second, if you are interested in cloud agents, software factories, agentic back office, SecMasters, all these different things, uh you should reach out. s224 00:19:20.263 --> 00:19:23.343 Cliff and I will be in the hallway track immediately after this. s225 00:19:23.343 --> 00:19:26.183 Uh and of course, you can check us out at noriagentic.com. s226 00:19:26.183 --> 00:19:28.463 Thank you so much.