1 00:00:00,06 --> 00:00:01,07 - [Instructor] Now that your agent works 2 00:00:01,07 --> 00:00:03,03 and you know how to observe it, 3 00:00:03,03 --> 00:00:06,07 it's time to make sure it behaves responsibly. 4 00:00:06,07 --> 00:00:09,04 This is where governance and compliance come in. 5 00:00:09,04 --> 00:00:12,08 They define who controls what, how decisions are logged, 6 00:00:12,08 --> 00:00:15,07 and which rules your agent must follow. 7 00:00:15,07 --> 00:00:18,02 Why should you care about governance from day one? 8 00:00:18,02 --> 00:00:20,04 Well, governance makes sure that your agent 9 00:00:20,04 --> 00:00:22,02 doesn't go off script. 10 00:00:22,02 --> 00:00:24,05 It's essentially about accountability, 11 00:00:24,05 --> 00:00:27,08 knowing who configured what, what data it can touch, 12 00:00:27,08 --> 00:00:31,02 and how its actions are recorded for later review. 13 00:00:31,02 --> 00:00:32,05 When we say governance, 14 00:00:32,05 --> 00:00:35,04 we mean the policies, processes, and ownership 15 00:00:35,04 --> 00:00:38,06 that keep your system safe and accountable. 16 00:00:38,06 --> 00:00:41,00 Every AI agent needs clear ownership, 17 00:00:41,00 --> 00:00:43,06 someone responsible for maintaining it, 18 00:00:43,06 --> 00:00:46,07 reviewing its actions, and improving changes, 19 00:00:46,07 --> 00:00:49,01 and being held accountable for the outcomes. 20 00:00:49,01 --> 00:00:52,00 You can't just say the AI did that. 21 00:00:52,00 --> 00:00:54,09 Governance also defines who can modify prompts, 22 00:00:54,09 --> 00:00:56,04 who adds or removes tools, 23 00:00:56,04 --> 00:00:59,04 and how those changes are documented. 24 00:00:59,04 --> 00:01:02,09 Without ownership, no policy or audit log really matters 25 00:01:02,09 --> 00:01:06,04 because no one is accountable for following them. 26 00:01:06,04 --> 00:01:08,09 Let's contrast that to compliance. 27 00:01:08,09 --> 00:01:10,05 Compliance means your agent 28 00:01:10,05 --> 00:01:13,09 respects internal and external rules, privacy laws, 29 00:01:13,09 --> 00:01:15,00 retention policies, 30 00:01:15,00 --> 00:01:18,03 and any standard your organization must follow. 31 00:01:18,03 --> 00:01:20,02 Even if you're just prototyping, 32 00:01:20,02 --> 00:01:22,06 it's smart to design with these in mind. 33 00:01:22,06 --> 00:01:24,07 Even if you cancel them right now, 34 00:01:24,07 --> 00:01:28,02 you should be aware of what's coming down the road. 35 00:01:28,02 --> 00:01:31,01 N8n already includes some key building blocks 36 00:01:31,01 --> 00:01:32,00 for governance. 37 00:01:32,00 --> 00:01:34,09 For example, you can manage who accesses what, 38 00:01:34,09 --> 00:01:37,03 keep credentialed separate from logic, 39 00:01:37,03 --> 00:01:41,07 and use version history to track prompt or workflow changes. 40 00:01:41,07 --> 00:01:44,06 The execution log then serves as an audit trail, 41 00:01:44,06 --> 00:01:48,04 showing exactly what the agent did and when. 42 00:01:48,04 --> 00:01:51,05 Good governance also means environment separation, 43 00:01:51,05 --> 00:01:55,03 develop safely, and then promote to production, 44 00:01:55,03 --> 00:01:57,06 review workflows before they go live, 45 00:01:57,06 --> 00:02:00,03 keep sensitive API keys out of reach, 46 00:02:00,03 --> 00:02:02,01 and treat prompts as code, 47 00:02:02,01 --> 00:02:06,07 version and approve them just like any software change. 48 00:02:06,07 --> 00:02:09,05 Most importantly, governance isn't just policy, 49 00:02:09,05 --> 00:02:11,03 it's human judgment. 50 00:02:11,03 --> 00:02:14,06 For critical actions like sending quotes, payments, 51 00:02:14,06 --> 00:02:18,07 or emails, always include a human in the loop step, 52 00:02:18,07 --> 00:02:20,09 especially at the very beginning. 53 00:02:20,09 --> 00:02:22,07 That keeps accountability clear 54 00:02:22,07 --> 00:02:27,00 while still letting the agent handle the repetitive work. 55 00:02:27,00 --> 00:02:29,08 Governance builds trust inside your organization. 56 00:02:29,08 --> 00:02:31,09 Compliance protects you. 57 00:02:31,09 --> 00:02:35,03 Together they go hand in hand and ensure that your agent 58 00:02:35,03 --> 00:02:37,09 turns into something you and your team 59 00:02:37,09 --> 00:02:40,00 can truly stand behind. 60 00:02:40,00 --> 00:02:41,08 And if you are wondering where to start, 61 00:02:41,08 --> 00:02:43,09 do two simple things today. 62 00:02:43,09 --> 00:02:46,03 First, assign clear ownership. 63 00:02:46,03 --> 00:02:48,01 Make one person or small team 64 00:02:48,01 --> 00:02:51,00 responsible for your agent's health and updates. 65 00:02:51,00 --> 00:02:53,03 And second, start an audit trail. 66 00:02:53,03 --> 00:02:54,09 Version your prompts and workflows 67 00:02:54,09 --> 00:02:57,03 so you can see what changed and when. 68 00:02:57,03 --> 00:03:01,00 Those two steps alone set you off on a great path.