Learning Library

← Back to Library

Turn Your Current Role into an AI Job

Key Points

  • The crucial mindset shift is to ask how you can turn your existing role into an AI‑enhanced one rather than hunting for a separate “AI job.”
  • In 2025 AI moved from being a superficial chat/assistant layer to becoming a core infrastructure layer that underpins everyday workflows.
  • A standardized agent architecture emerged—defining agents as goal‑driven loops with context gathering, reasoning, action, and observation—and introduced maturity models and design principles for multi‑agent systems.
  • Security moved from a theoretical concern to an operational reality, as “shadow IT” and personal AI tools exposed organizations to risks that must now be governed.
  • The practical path forward for 2026 is to leverage your company’s AI infrastructure and the new agent frameworks to redesign how work gets done in your current position.

Full Transcript

# Turn Your Current Role into an AI Job **Source:** [https://www.youtube.com/watch?v=gtkRAXQf49k](https://www.youtube.com/watch?v=gtkRAXQf49k) **Duration:** 00:13:29 ## Summary - The crucial mindset shift is to ask how you can turn your existing role into an AI‑enhanced one rather than hunting for a separate “AI job.” - In 2025 AI moved from being a superficial chat/assistant layer to becoming a core infrastructure layer that underpins everyday workflows. - A standardized agent architecture emerged—defining agents as goal‑driven loops with context gathering, reasoning, action, and observation—and introduced maturity models and design principles for multi‑agent systems. - Security moved from a theoretical concern to an operational reality, as “shadow IT” and personal AI tools exposed organizations to risks that must now be governed. - The practical path forward for 2026 is to leverage your company’s AI infrastructure and the new agent frameworks to redesign how work gets done in your current position. ## Sections - [00:00:00](https://www.youtube.com/watch?v=gtkRAXQf49k&t=0s) **Make Your Current Job AI‑Native** - The speaker urges professionals to stop seeking separate AI positions and instead transform their existing roles by leveraging AI as an infrastructure layer—a shift that occurred in 2025—and outlines the new mental models and practical steps needed to become an AI‑native worker by 2026. - [00:03:56](https://www.youtube.com/watch?v=gtkRAXQf49k&t=236s) **AI Agents: 2025 Lessons** - The speaker explains how 2025 clarified that AI agents deliver strong ROI when confined to repetitive, verifiable tasks, highlighting the need for IT partnership, new deployment skills, and careful safety considerations. - [00:07:36](https://www.youtube.com/watch?v=gtkRAXQf49k&t=456s) **AI Agent Governance Essentials** - The speaker stresses the need to understand where AI agents store logs and metrics, how they communicate via protocols, enforce security roles, and integrate governance as a core operating system for credible, auditable AI adoption. - [00:10:42](https://www.youtube.com/watch?v=gtkRAXQf49k&t=642s) **Designing Human‑AI Collaborative Workflows** - The speaker urges professionals to transition from repetitive tasks to strategic, people‑centric roles by mapping their existing workflows and prototyping AI‑augmented processes, positioning themselves as designers and supervisors of AI agents. ## Full Transcript
0:00My inbox and my DMs are full of people 0:02saying, "Can I get an AI job? How do I 0:04get an AI job?" And that is the wrong 0:06question, people. The right question is, 0:09"How do I turn my current job into an AI 0:13job?" I'm dead serious. And I'm going to 0:15talk about it here. Your goal in 2026 is 0:18going to be much more specific than a 0:20dream of another job. It's going to be 0:23not changing careers, not becoming a 0:25prompt engineer, but how can you change 0:28the way work actually gets done in your 0:31current role using the AI infrastructure 0:34your company is already rolling out. I 0:38am telling you for 95% of us that is the 0:41way AI is going to come. And we don't 0:43talk about it. We talk about changing 0:45jobs all the time, but like that's a 0:47tiny sliver of the world. For so much of 0:50us, it is not about that. I'm actually 0:53going to focus on what changed in AI in 0:562025 underneath the hype, the new mental 0:59models that you need to understand what 1:01matters in 2026, particularly around AI 1:04agents, and a practical path to making 1:07your existing job an AI native job. So, 1:10what actually changed in 2025? like 1:12underneath the hood, underneath all the 1:14hype, stepping back, the first thing you 1:16need to recognize is that AI moved from 1:19a chat interface into being an 1:21infrastructure layer this year. So for 1:23the last two years, for most of us, the 1:25experience of AI has been it's a chat 1:27box, it's a writing assistant, maybe it 1:29does some code completion. That is now 1:31the most superficial layer of AI. 1:34Underneath the surface, three big shifts 1:36happened in 2025 that changed the game 1:40on AI. Number one is that architecture 1:42started to get standardized. Google's 1:44recent introduction to AI agents paper 1:46is just the latest example of this. The 1:49larger perspective if you step back is 1:51that we have started to get a clear 1:53industry definition around an agent as a 1:56loop. An agent has a goal, gathers 1:59context, it reasons, it acts, it 2:01observes. And we have patterns now for 2:04multi- aent systems that include planner 2:06agents, retriever agents, executor 2:08agents, etc. We also have a the 2:12beginning of an industry model for agent 2:15maturity from simple tool calling all 2:17the way up to self-improving systems 2:18which nobody has or almost nobody has. 2:21And finally, we have design principles 2:23around how we think around issues like a 2:25budgetary authority for agents, 2:27boundaries for agents, security identity 2:29for agents. Still evolving, but it's 2:31starting to come into place. The the 2:33reason you need to care about this is 2:36that until we had that architecture, 2:38agents were mostly theoretical or they 2:40were point solutions to problems. 2:42Because of the work done in 2025, 2:45because that architecture is more 2:46standardized, we are now set up to do 2:48much more interesting things, much more 2:51comprehensive work with agents in 2026. 2:53The second big piece in 2025 is that 2:56security is no longer a hypothetical. 2:582025 was a year of shadow IT. bring your 3:02own AI to work. Maybe security won't 3:05check. Maybe your chief information 3:06security officer won't notice you 3:08brought your personal chat GPT. That is 3:10increasingly going to be out of bounds, 3:13caught and not allowed. And the reason I 3:16say that is because these CISOs, 3:19information officers have had a year to 3:22get their teams in gear to approve a 3:24bunch of tools like Cloud Code, like 3:26Chad GBT, like lovable. And so 3:28increasingly the tools that are allowed 3:30are inside the fences now. And the 3:33critical thing that you need to be aware 3:34of is that the security focus is now 3:37moving into that agent space. And so 3:40more and more the real meaningful shifts 3:42are going to be done in partnership with 3:46your security teams at work. It's not 3:48going to be just the marketing team 3:50setting up their individual little tool 3:52and hoping and praying nobody notices. 3:54more and more that's going to require 3:56your partnership with the rest of the IT 3:58department and that is something I will 4:00absolutely get into but it's it's a 4:02skill we need to develop that most of us 4:04haven't had to use before because 4:06frankly the ability to deploy technical 4:08agents to do this work is brand new. The 4:11third major change in 2025 is that 4:14enterprises learned where AI agents 4:16actually work. This is probably the 4:18biggest one. I can't underline this one 4:20enough. across hundreds of deployments. 4:22The pattern is annoyingly consistent. 4:25Agents are reliable and deliver really 4:28good ROI on work tasks when they are 4:31bounded in scope, when they are 4:32objectively verifiable, when they are 4:35repetitive, and when they have clearly 4:37defined inputs and outputs. So you can 4:40think back office operations, triage 4:42operations, claims, lead qualification, 4:45document checks, basic compliance, 4:47customer support flows. It is not invent 4:50our product strategy the AI agent. It is 4:53hey can you execute this same process we 4:55do 10,000 times a week and please don't 4:57get bored. That's where AI agents are 4:59going. So 2025 gave us a lot of clarity 5:02and that shapes how we prepare ourselves 5:04in our roles for AI agents and yes it 5:07will touch all of us. So it gave us 5:08clarity on what agents are how they 5:10operate at scale when where they're safe 5:13where they're useful and where they're 5:14dangerous if you're sloppy. This all 5:17lays the foundation for what comes next. 5:20If you're looking ahead to 2026, these 5:22are the three mental models that you 5:24need to survive in your career as we 5:28start to have AI agents more and more in 5:30the workplace. Number one, AI is a 5:33collaborator on structured work. It is 5:35not a magic brain. So, I'm going to say 5:37it again, LLMs are pattern machines. 5:40They're very, very good at transforming 5:41text and code. They they can map messy 5:44inputs to structured outputs very well. 5:46They follow explicit instructions 5:47increasingly well and they can do the 5:50same thing a thousand or 10,000 times 5:53and never get bored. But they are not 5:55inherently good at making high stakes 5:58decisions with very ambiguous 5:59trade-offs. They don't understand your 6:01organization's politics or background 6:03well. They don't know your context 6:05unless you give it to them. and they are 6:07very very bad at respecting boundaries 6:09that you have not defined previously. 6:12And so the right question is not can AI 6:15do my job although I hear that a lot 6:17that that's wrong. That's not the right 6:19question to ask given what we know about 6:20AI agents today. Instead it is which 6:23parts of my job are repetitive are 6:26checkable are describable or verifiable 6:29and how do I turn those into workflows 6:31that AI can run or assist with? How do I 6:33begin to take charge of how AI shapes my 6:36job? And if you can't describe the work 6:38clearly, that's something that you're 6:40going to have to do. The AI just doesn't 6:42have a chance at that. The second major 6:45mental model is agents plus 6:47orchestration are becoming the new 6:49middleware. And if that sounds abstract, 6:52the key thing to understand is that 6:54middleware has always existed in our 6:56software stacks. In between backend and 6:58front end, there has always been a piece 7:00of the stack that translates. That part 7:03of the stack now got intelligent. It got 7:06intelligent because agents are 7:08increasingly going to be that 7:09middleware. All an agent is is a loop 7:11around a model. It has tools. It has 7:13some kind of state that it's working 7:14with and it has decision logic. That's 7:17it. The important part here isn't that 7:20we label this middleware. It's that we 7:22understand that this orchestration layer 7:24is going to be driving a lot of how we 7:25do productivity. And we need to take 7:27charge of what that looks like. So what 7:29tools does it allowed to use? Under what 7:32identity is it secure with what budget? 7:35What where are the logs and the metrics 7:36stored? What does it do when it doesn't 7:38know? This is the part that most people 7:40don't see or think about. But you need 7:43to think about it if you want to have a 7:46productive relationship with AI agents 7:48in your role. You need to at least 7:50understand the vocabulary, how models 7:53talk to tools and data. Maybe through 7:54model context protocol, maybe other 7:56ways. What are agentto agent protocols? 7:58How do teams of agents coordinate? And 8:00how can you talk about that at a high 8:02level even if you're not an engineer? 8:04Control panes, gateways. What are the 8:05choke points where organizations are 8:07going to enforce security policies and 8:09observe behavior? How do you ensure that 8:12the agents that are built have the right 8:14roles and permissions? I am not 8:16expecting you to implement this 8:18yourself. Most people won't. But if you 8:21want to be taken seriously, you do need 8:24to be able to talk at a high level about 8:26AI workflows in your area in these terms 8:29because that makes you translatable. 8:31That makes you accessible to people who 8:33will be building this for you and you 8:35will want that skill. The third major 8:37mental model for 2026 is governance. 8:40It's not a bolt-on. It is it's going to 8:42be the new operating system, guys. AI is 8:44becoming grown up. If your AI adoption 8:46story doesn't include security and 8:48privacy and auditability and all of that 8:49stuff that seems boring, it's not going 8:51to be taken seriously. And so you need 8:54to be providing proactive answers to in 8:56your domain, where would you allow AI to 8:58act autonomously? Where would you allow 9:00it to only draft? Where would you 9:02require a human approver? How do you 9:04shut it down safely? This is no longer 9:06just your chief information security 9:08officer's problem. It is becoming 9:10everyone's problem because AI agents 9:12will not roll out successfully if they 9:14do not know your local information and 9:17data. So where will AI actually reshape 9:19your job keeping all of that in mind? 9:21Fundamentally, you need to think of your 9:23job as a stack of workflows. Your job is 9:27going to be decomposed and you need to 9:29take charge of what that looks like. So 9:31don't think of it as doing marketing. 9:32Think of it as you run campaigns, you 9:34create briefs, you analyze performance, 9:36you manage stakeholders. Those are 9:37workflows. You don't do product 9:39management. Instead, you collect 9:40requirements. You prioritize. You write 9:42specs. You coordinate launches. 9:44Workflows. Again, you don't do finance. 9:46Instead, you reconcile. You forecast. 9:48You analyze variants. You produce 9:49reports. Again, workflows. Each of these 9:52can be decomposed into triggers, what 9:54starts the work, inputs, what you look 9:56at, transformations, what do you do with 9:58it? Decisions, outputs, and checks to 10:01know if it's correct. AI slots into a 10:04structure like that. AI will handle the 10:07boring and repetitive parts of those 10:08workflows. It is up to you to figure out 10:11how that actually shapes in your role. 10:13Across industries, the same categories 10:15keep getting automated or heavily 10:17assisted. Triage tasks, routing tasks, 10:20summarization tasks, synthesis tasks, 10:23policy and rule tasks, repetitive 10:26document workflows like pulling data 10:28from forms, glue work across tools, 10:30moving information from Excel into Word 10:32or vice versa. If you look at your job 10:34honestly, for most of us, a non-trivial 10:37percentage is in one of those buckets. 10:39And that is what is going to move first. 10:42Now, the parts that stay human for a 10:44long time to come are parts around 10:46negotiation, around trust building, 10:47around politics, around deciding which 10:49problems to solve, around setting 10:51strategy, around being accountable when 10:53things go wrong. So, I don't want you to 10:55hear Nate is proposing that I AI away my 10:58job. I want you to hear that AI drains 11:01repetitive and checkable work out of 11:03your role. You should be in charge of 11:06what that looks like or someone else 11:07will do it for you. And your value is 11:09going to shift toward defining 11:10workflows, supervising them, handling 11:13exceptions, choosing what to build, 11:15touching the work that matters. And so 11:17when you think about what to do in the 11:19next few weeks as you head into 2026, if 11:22you want to get a running start, number 11:24one, map your work as if you were a 11:26systems designer. I've given you a cheat 11:28code here. Write down your workflows. 11:30Write down what triggers them, what 11:32inputs there are, what outputs there 11:33are, what decisions there are. Learn to 11:36express those workflows to the tools you 11:38already have. Try something even if it's 11:41prototypy in chat GPT enterprise or in 11:44Copilot or in Gemini to get the idea of 11:47what that workflow would look like so 11:48you can show a workable prototype when 11:51it comes along and has an AI agent 11:53initiative. I'm not saying spin up rogue 11:55infrastructure. I'm saying try and 11:57prototype something so you get a living 11:59feel for what AI agents working with you 12:01would look like and then be in the 12:03driver's seat when you have these 12:05conversations with your engineering 12:06teams. I would also encourage you to 12:09build a relationship with the people 12:11championing AI in your org. Maybe that's 12:14you cuz you watch this channel or maybe 12:17it is someone else who is responsible 12:18for the technical side. But either way, 12:21make sure that you are finding the right 12:23team who is responsible for AI 12:25initiatives in your area. That you are 12:27showing them you've done your homework 12:29and you're thinking inside the existing 12:31organizational guard rails. You're 12:32thinking about workflows. You're 12:34thinking about patterns and tools. At 12:35that point, you're no longer a random 12:38person. You're a valuable champion and 12:40an ally who speaks both languages. The 12:42messy reality of the business language 12:44and the constraints of the platform that 12:45the technical teams think about. and you 12:47are in a position to be a fluent 12:49translator of AI and drive how AI agents 12:51work with you in your role. That is a 12:53very valuable position and that is what 12:56you need to be able to do to be in the 12:59driver's seat. This is what I wish I 13:01could tell 95% of people who are not 13:03going to switch jobs in the next year 13:05for AI roles. This is what you need to 13:07know to be in charge of AI in your role. 13:10So, I hope this has been helpful. 13:11There's a lot more written up on the 13:13substack, including some prompts to help 13:14you think through this. My goal is to 13:16give you a guide so that you can 13:18meaningfully engage with your existing 13:21role and prepare for it now before we go 13:23into 2026. And AI agents are absolutely 13:26everywhere. Good luck.