WEBVTT 00:00:00.000 --> 00:00:02.419 When a big player like OpenAI drops a new tool, 00:00:02.500 --> 00:00:06.379 especially one they call Canva for Agents. No. 00:00:06.639 --> 00:00:09.140 Well, the internet tends to jump straight to 00:00:09.140 --> 00:00:11.179 destruction mode. Oh, yeah. It's always, what's 00:00:11.179 --> 00:00:13.439 it going to kill? As soon as AgentKit landed, 00:00:13.619 --> 00:00:17.399 October 6, 2025, you saw it everywhere. RIP and 00:00:17.399 --> 00:00:20.219 8M. Just instant hyperbole. But that whole Tool 00:00:20.219 --> 00:00:23.239 Wars thing. It often misses the point, doesn't 00:00:23.239 --> 00:00:24.859 it? Especially when you're talking about serious 00:00:24.859 --> 00:00:27.699 business automation. Does the hype around easy 00:00:27.699 --> 00:00:29.760 visual building actually hold up for complex 00:00:29.760 --> 00:00:32.179 needs? Exactly. And that's what this deep dive 00:00:32.179 --> 00:00:34.539 is all about. We're doing a really rigorous head 00:00:34.539 --> 00:00:37.659 -to -head comparison. AgentKit versus NAN. We 00:00:37.659 --> 00:00:39.299 want to cut through that noise and figure out 00:00:39.299 --> 00:00:41.820 for you when you'd use which one. We'll look 00:00:41.820 --> 00:00:44.399 at six key areas. Okay, so let's set the stage. 00:00:44.740 --> 00:00:46.799 AgentKit. That's the new one. Low -code, drag 00:00:46.799 --> 00:00:49.810 -and -drop, visual. Built specifically for conversational 00:00:49.810 --> 00:00:52.890 agents. And importantly, it lives entirely within 00:00:52.890 --> 00:00:55.030 the open AI world. It's about accessibility. 00:00:55.509 --> 00:00:57.850 Right. And then there's NAN. Been around six 00:00:57.850 --> 00:01:00.390 years. The established powerhouse. Open source. 00:01:00.469 --> 00:01:03.850 Huge number of integrations. Over 500. It really 00:01:03.850 --> 00:01:07.310 targets developers, power users, people building 00:01:07.310 --> 00:01:10.810 custom, truly autonomous systems. That one's 00:01:10.810 --> 00:01:13.150 built for total control. So our starting hypothesis, 00:01:13.450 --> 00:01:15.450 just going into the testing, was pretty straightforward. 00:01:16.569 --> 00:01:19.150 AgentKit isn't going to kill an AM. They seem 00:01:19.150 --> 00:01:20.950 built for fundamentally different users, different 00:01:20.950 --> 00:01:23.090 jobs. Let's see if that holds up. Let's do it. 00:01:23.170 --> 00:01:25.569 First up, ease of use, or what we're calling 00:01:25.569 --> 00:01:27.829 the startup experience. Basically, how much friction 00:01:27.829 --> 00:01:29.790 is there just getting your first agent up and 00:01:29.790 --> 00:01:32.909 running? AgentKit. It felt incredibly beginner 00:01:32.909 --> 00:01:35.329 -friendly. We gave it an 8 out of 10. The interface 00:01:35.329 --> 00:01:39.209 is just clean, minimal, basic logic like IFL, 00:01:39.250 --> 00:01:41.379 transforming data. Pretty straightforward. Yeah, 00:01:41.420 --> 00:01:43.180 the key is speed. You can actually build your 00:01:43.180 --> 00:01:45.239 first agent, maybe doing a simple web search 00:01:45.239 --> 00:01:47.700 in like less than two minutes. Seriously, no 00:01:47.700 --> 00:01:50.400 API keys needed right away, no complex setup. 00:01:50.560 --> 00:01:52.680 You just start building. Okay, contrast that 00:01:52.680 --> 00:01:55.299 with N8n. We scored it a 6 out of 10 here. It 00:01:55.299 --> 00:01:57.659 is definitely powerful, a power user's dream 00:01:57.659 --> 00:01:59.879 maybe, but wow, there's a learning curve. You 00:01:59.879 --> 00:02:01.840 open it up and it's just hundreds of objects. 00:02:02.079 --> 00:02:05.859 That same... Simple web search. In A &A, you've 00:02:05.859 --> 00:02:07.900 got to select specific models that have custom 00:02:07.900 --> 00:02:11.099 HTTP requests. Definitely need your API keys 00:02:11.099 --> 00:02:13.659 handy right from the start. It asks for commitment 00:02:13.659 --> 00:02:16.479 up front, you know, but then it gives you enormous 00:02:16.479 --> 00:02:19.360 power back later. So for the learner just starting 00:02:19.360 --> 00:02:21.960 out, what's the main takeaway on that startup 00:02:21.960 --> 00:02:24.979 experience? Is it speed versus depth? Agent Kit 00:02:24.979 --> 00:02:28.099 gives you that fast access, that quick win. Ain't 00:02:28.099 --> 00:02:30.439 Ain't Ain't makes you invest time before unlocking 00:02:30.439 --> 00:02:33.240 its real power. Okay, makes sense. So if getting 00:02:33.240 --> 00:02:35.819 started is one thing, how does Agent actually 00:02:35.819 --> 00:02:38.979 start? Let's talk triggers. Yeah, and this is 00:02:38.979 --> 00:02:41.319 where you see a real philosophical split. Agent 00:02:41.319 --> 00:02:43.560 Kit scored a 5 out of 10 from us. Why? It's basically 00:02:43.560 --> 00:02:45.759 limited to just the start node. Meaning? Meaning 00:02:45.759 --> 00:02:48.479 interactions are almost always chat -based. You, 00:02:48.659 --> 00:02:50.979 a human, have to talk to it to kick things off. 00:02:51.080 --> 00:02:53.639 Right. It forces that human intervention for 00:02:53.639 --> 00:02:55.639 pretty much everything. No scheduled triggers, 00:02:55.860 --> 00:02:58.419 no reacting to things happening in other apps. 00:02:58.560 --> 00:03:01.639 Nope. No timed triggers, no listening for events 00:03:01.639 --> 00:03:04.879 in Slack or Gmail, no real background capability. 00:03:05.280 --> 00:03:07.610 It needs you to chat with it. That feels like 00:03:07.610 --> 00:03:09.689 a pretty big limitation for genuine automation, 00:03:09.930 --> 00:03:11.610 doesn't it, for systems that should just run 00:03:11.610 --> 00:03:13.849 themselves? Totally. If a business needs something 00:03:13.849 --> 00:03:16.990 to, say, check inventory levels every night at 00:03:16.990 --> 00:03:20.169 3 a .m. and send an alert if stock is low, AgentKit 00:03:20.169 --> 00:03:21.930 just can't do that silently. It needs a human 00:03:21.930 --> 00:03:24.069 prompt. So if a business needs automation that 00:03:24.069 --> 00:03:26.750 runs without someone typing into a chat, what's 00:03:26.750 --> 00:03:29.150 AgentKit's strategic problem there? Its design 00:03:29.150 --> 00:03:31.569 inherently forces human interaction, limiting 00:03:31.569 --> 00:03:33.870 its use for those silent background processes. 00:03:34.250 --> 00:03:37.439 Okay. And A &AN on triggers. Oh, A &AN is trigger 00:03:37.439 --> 00:03:40.379 paradise. Solid 10 out of 10. Hundreds of native 00:03:40.379 --> 00:03:43.879 triggers built in. Gmail, Slack, webhooks from 00:03:43.879 --> 00:03:46.219 anywhere, databases. Time -based stuff too, like 00:03:46.219 --> 00:03:48.810 run every hour. Yep. Time -based external events. 00:03:49.030 --> 00:03:51.610 It lets you build genuinely autonomous, scalable 00:03:51.610 --> 00:03:54.169 workflows that just hum along quietly in the 00:03:54.169 --> 00:03:56.069 background. So it's the difference between like 00:03:56.069 --> 00:03:58.590 a chatbot assistant you have to poke versus a 00:03:58.590 --> 00:04:01.270 dedicated background system doing work silently. 00:04:01.550 --> 00:04:03.789 Exactly that. A virtual assistant versus a silent 00:04:03.789 --> 00:04:05.330 engineer. All right. Moving on from starting 00:04:05.330 --> 00:04:07.069 the workflow, let's get into the core stuff. 00:04:07.270 --> 00:04:09.650 What tools can these agents actually use? What 00:04:09.650 --> 00:04:11.889 can they connect to? Okay. AgentKit scored a 00:04:11.889 --> 00:04:14.949 5 out of 10 on tools. The built -in set is, well, 00:04:15.030 --> 00:04:17.819 limited. You get web search. something called 00:04:17.819 --> 00:04:20.220 a client tool to send data back to the chat interface. 00:04:20.500 --> 00:04:23.360 And then these things called MCP servers. Hang 00:04:23.360 --> 00:04:25.579 on. MCP servers. Can you break that down? What 00:04:25.579 --> 00:04:28.230 is that in plain English? Sure. Think of MCP 00:04:28.230 --> 00:04:30.709 servers as AgentKit's pre -built connectors for 00:04:30.709 --> 00:04:33.290 a few common services, like basic hooks into 00:04:33.290 --> 00:04:35.970 Gmail or Google Drive. They're easy to set up, 00:04:36.029 --> 00:04:38.850 which is nice, but the selection is really narrow 00:04:38.850 --> 00:04:40.829 compared to what else is out there. Gotcha. And 00:04:40.829 --> 00:04:44.069 AN. AN is the integration universe here. Easy 00:04:44.069 --> 00:04:47.629 10 out of 10. Yes, it has over 500 native integrations 00:04:47.629 --> 00:04:50.990 built right in, but the real superpower, it's 00:04:50.990 --> 00:04:54.189 the HTTP request tool. Which means? Which means 00:04:54.189 --> 00:04:57.209 if any service out there has an API, Any service 00:04:57.209 --> 00:04:59.730 at all. Any ant can talk to it. Period. Full 00:04:59.730 --> 00:05:02.569 stop. Wow. Okay. And what's really cool, architecture 00:05:02.569 --> 00:05:05.709 -wise, is modularity. NEN lets you build and 00:05:05.709 --> 00:05:08.550 reuse sub -workflows. Imagine building little 00:05:08.550 --> 00:05:11.290 specialized microagents, one for analyzing sentiment, 00:05:11.629 --> 00:05:14.850 one for parsing invoices, one for summarizing 00:05:14.850 --> 00:05:16.790 emails, and then having your main agent call 00:05:16.790 --> 00:05:19.689 these as needed. Yeah. I still wrestle with prompt 00:05:19.689 --> 00:05:21.829 drift and tool selection myself sometimes when 00:05:21.829 --> 00:05:24.449 I'm trying to chain multiple complex AI tasks 00:05:24.449 --> 00:05:26.930 together. So the idea of building one really 00:05:26.930 --> 00:05:29.769 solid, reliable sub -workflow for, say, data 00:05:29.769 --> 00:05:31.709 extraction and then just reusing that everywhere, 00:05:31.949 --> 00:05:35.329 that's... incredibly appealing, reduces so much 00:05:35.329 --> 00:05:37.670 variability. That modular approach using sub 00:05:37.670 --> 00:05:40.149 -workflows is NEN's killer advantage for complex 00:05:40.149 --> 00:05:43.089 specialized business processes, consistency and 00:05:43.089 --> 00:05:45.189 reliability. So beyond just the sheer number 00:05:45.189 --> 00:05:47.329 of tools, what's NEN the key advantage for those 00:05:47.329 --> 00:05:49.810 really specific business needs? Its architecture 00:05:49.810 --> 00:05:52.930 lets you build modular, reusable agent systems 00:05:52.930 --> 00:05:55.410 using those sub -workflows. Okay, let's shift 00:05:55.410 --> 00:05:57.610 to the brain behind the operation. Model support. 00:05:57.910 --> 00:06:00.949 Which AI engines can you actually use? With AgentKin, 00:06:00.970 --> 00:06:03.300 it feels like the OpenAI Exclusive Club. We gave 00:06:03.300 --> 00:06:05.459 it a 6 out of 10. Yeah. The integration with 00:06:05.459 --> 00:06:08.439 OpenAI models is excellent. Really deep. Easy 00:06:08.439 --> 00:06:11.519 toggles for GPT -4 versus others. Reasoning effort. 00:06:11.639 --> 00:06:14.860 How chatty it is. But, and it's a big but, you 00:06:14.860 --> 00:06:17.420 are completely locked into the OpenAI ecosystem. 00:06:17.759 --> 00:06:20.420 No Anthropic Quad. No Google Gemini. No open 00:06:20.420 --> 00:06:22.500 source models running locally. None of that. 00:06:22.600 --> 00:06:25.839 No Azure. No AWS Bedrock. No Cohere. Definitely 00:06:25.839 --> 00:06:28.620 no local models on your own hardware. It's OpenAI 00:06:28.620 --> 00:06:31.660 or nothing. And NAN. NAN is the model democracy. 00:06:32.240 --> 00:06:34.480 A clear 10 out of 10. You get genuine flexibility. 00:06:35.139 --> 00:06:38.139 Anthropic, Azure, OpenAI, AWS Bedrock, Cohere, 00:06:38.259 --> 00:06:41.180 anything via OpenRouter, plus local models using 00:06:41.180 --> 00:06:43.220 things like Alama. And you get control over the 00:06:43.220 --> 00:06:45.300 settings too, like temperature, top brand. Full 00:06:45.300 --> 00:06:47.540 control, which is crucial not just for performance, 00:06:47.600 --> 00:06:49.519 but for cost optimization. Right. So if cost 00:06:49.519 --> 00:06:51.720 saving is a major strategic goal, why does that 00:06:51.720 --> 00:06:54.019 model democracy matter so much? It allows you 00:06:54.019 --> 00:06:56.420 to route different tasks to the most cost effective 00:06:56.420 --> 00:06:58.800 and task appropriate model provider dynamically. 00:06:59.459 --> 00:07:02.089 Save money where you can. mid -roll placeholder 00:07:02.089 --> 00:07:05.829 sponsor content goes here. All right. So far, 00:07:05.970 --> 00:07:08.670 it feels like NAN is dominating on the pure back 00:07:08.670 --> 00:07:10.810 -end logic, the integration power, the model 00:07:10.810 --> 00:07:13.490 choice. But let's switch gears a bit. What good 00:07:13.490 --> 00:07:16.750 is all that complex logic if the user experience, 00:07:16.910 --> 00:07:19.910 the interaction is clunky? Let's talk UI and 00:07:19.910 --> 00:07:23.009 chat components. Ah, and this is where AgentKit 00:07:23.009 --> 00:07:26.149 really makes its case. A strong 9 out of 10 here. 00:07:26.310 --> 00:07:28.819 It's the front -end champion. Its integration 00:07:28.819 --> 00:07:31.620 with ChatKit means you can create really slick, 00:07:31.720 --> 00:07:34.240 professional -looking, branded website chat widgets 00:07:34.240 --> 00:07:36.220 and embeddable components right out of the box. 00:07:36.399 --> 00:07:38.160 We saw HubSpot talking about this, actually. 00:07:38.160 --> 00:07:40.459 They said ChatKit saved them literally weeks 00:07:40.459 --> 00:07:42.500 of custom front -end development work trying 00:07:42.500 --> 00:07:44.439 to build a similar interface themselves. That's 00:07:44.439 --> 00:07:47.160 huge. Weeks of dev time saved. Absolutely. So 00:07:47.160 --> 00:07:49.139 where's the clearest, most immediate time -saving 00:07:49.139 --> 00:07:51.620 advantage with AgentKit? Building polished, customer 00:07:51.620 --> 00:07:54.740 -facing chat interfaces takes basically zero 00:07:54.740 --> 00:07:57.660 extra development time. Compare that to ANAN, 00:07:57.800 --> 00:08:00.000 which we called the back -end beast. It scored 00:08:00.000 --> 00:08:03.480 a 5 out of 10 on UI. Its focus is overwhelmingly 00:08:03.480 --> 00:08:07.060 on the workflow logic. Yes, you can build basic 00:08:07.060 --> 00:08:09.480 chat interfaces, but they're hard to customize, 00:08:09.620 --> 00:08:12.279 hard to brand. For anything polished, you're 00:08:12.279 --> 00:08:14.500 looking at building a completely custom front 00:08:14.500 --> 00:08:17.459 -end using other tools and connecting it via 00:08:17.459 --> 00:08:19.939 API. So it's like Agent Kit gives you the finished, 00:08:20.000 --> 00:08:22.920 polished car. Well, NAN gives you a powerful 00:08:22.920 --> 00:08:24.819 engine and says, okay, now build the car body 00:08:24.819 --> 00:08:27.220 around it. That's a great analogy. Yeah. Okay. 00:08:27.259 --> 00:08:30.459 Final category, deployment and control. This 00:08:30.459 --> 00:08:33.080 gets into bigger enterprise concerns. Where does 00:08:33.080 --> 00:08:35.620 this thing run? Who controls the data? Who manages 00:08:35.620 --> 00:08:37.639 the infrastructure? Right. AgentKit gets a 7 00:08:37.639 --> 00:08:39.759 out of 10. It's a managed service. It runs only 00:08:39.759 --> 00:08:42.779 in OpenAI's cloud, the upside. Zero infrastructure 00:08:42.779 --> 00:08:44.799 for you to manage. Super convenient. The downside. 00:08:44.980 --> 00:08:47.440 OpenAI controls everything. Your data lives there. 00:08:47.600 --> 00:08:49.899 The uptime depends on them. The pricing is set 00:08:49.899 --> 00:08:52.100 by them. You have to fundamentally trust OpenAI 00:08:52.100 --> 00:08:54.799 with your critical operational data. Okay. And 00:08:54.799 --> 00:08:59.009 NAN. NANN is the sovereignty option, a full 10 00:08:59.009 --> 00:09:02.070 out of 10. It offers just unparalleled flexibility. 00:09:02.490 --> 00:09:05.070 Use their cloud service, sure, or self -host 00:09:05.070 --> 00:09:07.230 it on your own servers, or even run it locally 00:09:07.230 --> 00:09:09.149 on your laptop. And because it's open source? 00:09:09.450 --> 00:09:12.429 You get complete data control, zero vendor lock 00:09:12.429 --> 00:09:14.210 -in. You own the whole stack if you want to. 00:09:14.429 --> 00:09:17.509 Whoa. Imagine scaling that level of control, 00:09:17.769 --> 00:09:21.110 running maybe a billion queries a month, all 00:09:21.110 --> 00:09:24.389 on your own private, completely sovereign infrastructure. 00:09:25.279 --> 00:09:27.960 That's a game changer for highly regulated industries 00:09:27.960 --> 00:09:30.220 or just for companies wanting total control. 00:09:30.340 --> 00:09:33.769 Exactly. For compliance, data revidency. It's 00:09:33.769 --> 00:09:36.529 huge. So for a large enterprise, maybe in finance 00:09:36.529 --> 00:09:38.809 or health care, what's the single biggest factor 00:09:38.809 --> 00:09:41.210 tipping the scales toward A &N in this category? 00:09:41.450 --> 00:09:43.629 Complete sovereignty and control over data and 00:09:43.629 --> 00:09:45.750 infrastructure, especially for compliance needs. 00:09:46.090 --> 00:09:47.990 OK, so the final scores came out, AgentKit 40 00:09:47.990 --> 00:09:51.149 and A &N 51. But like we said up front, the raw 00:09:51.149 --> 00:09:52.669 numbers don't really capture the whole picture, 00:09:52.710 --> 00:09:54.789 do they? Not at all. The qualitative stuff matters 00:09:54.789 --> 00:09:56.909 just as much, maybe more. Take debugging, for 00:09:56.909 --> 00:09:59.509 example. Right. AgentKit, even though it's visual, 00:09:59.649 --> 00:10:02.809 sometimes when a workflow breaks. figuring out 00:10:02.809 --> 00:10:05.049 where the data went wrong between nodes can be 00:10:05.049 --> 00:10:08.049 opaque. You end up digging through logs. Whereas 00:10:08.049 --> 00:10:10.490 NAN, because it's built with developers in mind, 00:10:10.590 --> 00:10:13.110 gives you crystal clear visualization. You click 00:10:13.110 --> 00:10:15.809 on any node, you see the exact data that came 00:10:15.809 --> 00:10:17.889 in, how it was configured, and the exact data 00:10:17.889 --> 00:10:20.350 that went out. Troubleshooting is usually way 00:10:20.350 --> 00:10:23.470 faster. Plus, there's the community aspect. NAN 00:10:23.470 --> 00:10:26.440 has six years of history. Forums, templates, 00:10:26.720 --> 00:10:29.179 tutorials, community support. Yeah, a massive 00:10:29.179 --> 00:10:31.440 head start there. AgentKit is brand new, still 00:10:31.440 --> 00:10:33.580 building that ecosystem. Which brings us back 00:10:33.580 --> 00:10:36.080 to that strategic mindset we talked about. AgentKit 00:10:36.080 --> 00:10:38.399 isn't killing NEN. They're aimed differently. 00:10:38.659 --> 00:10:41.039 We need to shift from thinking which tool wins 00:10:41.039 --> 00:10:44.100 to which tool solves this specific problem best. 00:10:44.340 --> 00:10:46.620 That's the key. your clients or your business 00:10:46.620 --> 00:10:48.879 they choose you for results for solving their 00:10:48.879 --> 00:10:50.720 problems they don't really care which tool you 00:10:50.720 --> 00:10:52.720 used under the hood as long as it works reliably 00:10:52.720 --> 00:10:55.299 and efficiently so let's make it practical when 00:10:55.299 --> 00:10:57.720 should someone choose agent kit choose agent 00:10:57.720 --> 00:11:00.299 kit if you need something fast rapid deployment 00:11:00.299 --> 00:11:03.000 simple conversational agents quick prototypes 00:11:03.000 --> 00:11:07.500 or crucially if you want that polished chat widget 00:11:07.500 --> 00:11:09.899 on your website without spending weeks on front 00:11:09.899 --> 00:11:12.639 -end development and when is n8n the right call 00:11:13.100 --> 00:11:15.860 Choose N8n when you need flexibility with AI 00:11:15.860 --> 00:11:19.080 providers, when cost optimization across different 00:11:19.080 --> 00:11:21.940 models is critical, when you need complex background 00:11:21.940 --> 00:11:24.360 automations running silently without human intervention, 00:11:24.720 --> 00:11:27.460 when data sovereignty and control are non -negotiable, 00:11:27.559 --> 00:11:30.240 or frankly, when you need to connect to almost 00:11:30.240 --> 00:11:33.059 anything out there using its powerful API capabilities. 00:11:33.580 --> 00:11:36.559 It really sounds like the future is just coexistence. 00:11:36.559 --> 00:11:38.639 There's plenty of room. Absolutely. The market 00:11:38.639 --> 00:11:40.919 for automation is massive. AgentKit is going 00:11:40.919 --> 00:11:43.309 to be huge for quick internal. tools, simple 00:11:43.309 --> 00:11:45.429 customer -facing bots, people who value speed 00:11:45.429 --> 00:11:48.549 and ease above all. N8 will continue to dominate 00:11:48.549 --> 00:11:50.970 the complex enterprise -grade custom control 00:11:50.970 --> 00:11:53.490 scenarios. They serve different needs. So the 00:11:53.490 --> 00:11:56.570 big idea to wrap up. These platforms serve different 00:11:56.570 --> 00:11:59.370 masters. AgentKit is about making agent building 00:11:59.370 --> 00:12:03.230 accessible, fast, easy. NEN is about giving you 00:12:03.230 --> 00:12:06.360 unlimited control and integration power. The 00:12:06.360 --> 00:12:08.320 most important thing for you, the listener, the 00:12:08.320 --> 00:12:11.159 learner, is to become tool agnostic. Focus on 00:12:11.159 --> 00:12:13.059 the problem first. Yeah, and if you're totally 00:12:13.059 --> 00:12:15.340 new to this whole automation game, maybe start 00:12:15.340 --> 00:12:17.539 with AgentKit. Get your feet wet, understand 00:12:17.539 --> 00:12:19.960 the concepts, get some quick wins. Then when 00:12:19.960 --> 00:12:22.019 you inevitably hit its limits for more complex 00:12:22.019 --> 00:12:24.500 tasks, graduate to NAN for that deeper power 00:12:24.500 --> 00:12:27.100 and control. Because the tools themselves, AgentKit 00:12:27.100 --> 00:12:29.500 today, something else tomorrow, they'll keep 00:12:29.500 --> 00:12:32.879 changing. But the underlying skill. The ability 00:12:32.879 --> 00:12:35.399 to logically structure a problem, break it down, 00:12:35.460 --> 00:12:38.179 and design an automated solution. That skill 00:12:38.179 --> 00:12:41.039 is permanent. Focus on building that core problem 00:12:41.039 --> 00:12:43.019 -solving framework. That's the real investment. 00:12:43.139 --> 00:12:44.480 Focus on the framework. Focus on the outcomes. 00:12:44.659 --> 00:12:45.940 Excellent point. Until next time.