Pod Insights The Agile Daily Standup - AgileDad Agile is Dead. Welcome to the Era of “Agent-Driven” Development About this episode Agile is Dead. Welcome to the Era of “Agent-Driven” Development
Agile was invented for humans — biological entities that need coffee, sleep, and two weeks to understand a complex requirement. But we are no longer just building software with humans. We are building it with agents. Agents that don’t sleep, don’t need standups, and can iterate in seconds, not days.
Welcome to Agent-Driven Development (ADD). The era of the 2-week sprint is over. The era of the Continuous Loop is here.
- [website] https://www.agiledad.com/ https://www.agiledad.com/
- [instagram] https://www.instagram.com/agile_coach/ https://www.instagram.com/agile_coach/
- [facebook] https://www.facebook.com/RealAgileDad/ https://www.facebook.com/RealAgileDad/
- [Linkedin] https://www.linkedin.com/in/leehenson/ https://www.linkedin.com/in/leehenson/
Transcript 0:15 Speaker 00 All right, everyone. It's been a hot minute since I've had one of these, but I finally got one of them. So excited. So let's talk about it. Are you ready? Here it comes. Agile is dead. Ah, right. I'm telling you right now, what's so interesting about these to me is that at one point I was getting these hot and heavy all the time. Then it appeared it veered off for a little while and now I got a new one and I'm excited about this one. Actually, I gotta be honest. This is a 2026 rendition of the Agile is Dead post. And I want to start by saying I didn't do it. Somebody else made this post.
0:55 Speaker 00 And I'm both excited and heavily concerned at the same time, because I don't know exactly what direction they're going to take it. So someone asked me to review the article. They sent me the link, of course, with a smiley emoji. And we're going to run with it and see where it takes us. So here we go. So it starts off with, we all know the feeling. Okay. It's Tuesday. You're in a Zoom stand up. You explain that the off module is stuck in review. You estimate it will be done by Friday. The project manager nods, moves the Jira ticket to in progress and marks the burn down chart. Meanwhile, in your ID, an AI agent has just written 50 variations of that same module in the time it took you to unmute your microphone.
1:39 Speaker 00 Houston, we have a velocity problem. Okay, let's pause. There's no question in my mind that AI can do things a ton faster. And in many cases with higher quality and in many cases just overall better. But the question becomes, is that really what we want or need? So let's continue and see where it goes before I go down that path. Oh, boy. Here we go. In the words Sabrina Carpenter. Oh, boy.
2:09 Speaker 00 It says, "Agile was invented for humans, biological entities that need coffee, sleep, Mountain Dew, in two weeks to understand a complex requirement. But we are no longer just building software with humans. We are building with agents, agents that don't need sleep. They don't need stand-ups. They can iterate in seconds, not days. And we no longer require a Mountain Dew IV drip." Let's pause.
2:38 Speaker 00 Yes, you're right. There's nothing wrong with that sentence.
2:44 Speaker 00 But there's everything wrong with that sentence. Let's talk about it. So they continue, "Welcome to agent-driven development. The era of two-week sprints is over. The era of the continuous loop is here."
3:01 Speaker 00 Agile was designed to solve a coordination problem. Hmm. How do you organize 50 humans to build a cathedral without talking over each other? You break it into small chunks and have daily syncs. But in 2026, the bottleneck isn't writing the code. The code is free. The bottleneck is the overhead of the human coordination. Okay.
3:25 Speaker 00 I see where this article is heading. Let's have a conversation. So the first thing, Agile was invented for humans. Let's pause. Agile was invented for everyone, not just for humans. And yes, while humans are biological entities, and that portion is absolutely 100% true. And humans have needs. I've got news for you. We have a need for humans. And I think the part that you're missing is just because a computer can generate the code faster using agentic AI than a human can. No disputing that. It still doesn't mean that we need human, we don't need humans for that level of coordination, validation, quality refinement, integration, implementation.
4:09 Speaker 00 I think you see where I'm going with this. I think that the fact that you're calling out humans as disposable in a way is the most concerning piece of this. I believe that if we look at this from a perspective of at one point, humans could do certain things and we need the humans to do everything. And now we can augment what we need the humans for. We're going to have a much better experience with how the agent driven development, you know, world that you're talking about interacts with the human piece. So I have to be careful about that. The second thing you bring up is this velocity mismatch. Let's talk about that. So I don't know that I've ever had to organize 50 humans to build a cathedral before.
4:53 Speaker 00 But what I can say is I've organized many hundred humans building a Walmart before. And what I can tell you is I think, you know, not saying Walmart's a cathedral, but I think you get where I'm going with this. And this was many, many years ago. And what I can tell you is it went off without a hitch and it was easy. Part of what most people don't understand is they look at AI and they look at whether it's code development or prompting, it doesn't matter which one it is. When they look at AI, they look at it as a single threaded discussion where I'm having
5:33 Speaker 00 a conversation or a chat with one other individual who's a lot smarter than I am. That's going to do the heavy lifting for me so that all I need to do is ask the right questions. And I think that the mistake that you're making, that the pit that you're falling into is that code development was originally this way as well. Now code development is different. The way that we do things is different. I don't, I don't disagree with that. But what I do disagree with is this context that AI doesn't need chain of thought prompting and that we don't need to break AI prompts into smaller pieces.
6:17 Speaker 00 I think we're right back where we started is where I'm going with this. It's just, yes, AI can do it faster, but if it's not correct and if it doesn't break things down and it doesn't process the information a certain way, you're going to have to have a human interact to take care of those things. So. They give you a chart here called the bath of 2025 versus the math of 2026. And I love that they broke it down this way because I think you're going to find some great irony here. I may even attach this image because it's like chef's kiss for me. So here it is. It says, the first phase that you need to go through is ideation. Ideation in the agile human era was where the product manager writes tickets.
6:59 Speaker 00 In an agent driven era, the large language model generates five implementation strategies instantly with all the tickets.
7:09 Speaker 00 Question, do you honestly believe that no one's going to review the five implementation strategies? Do you think you're just going to create five strategies and tell AI, okay, pick the one you think it's going to be best and let's roll? It doesn't work that way, right? Sure, the LM can generate five implementation strategies, not arguing that point, but it's still going to require human oversight. Now, this is the part where it's disappointing. He skipped discovery altogether, which I'm going to add is very much telling in this case. We'll talk about that in a second and went straight to coding. In the past, senior dev would write features and take two days. Now an agent can create that same feature in 30 seconds.
7:54 Speaker 00 That part is probably the most accurate part, but the fact that you skip discovery altogether is just, it's so telling about the way that you see a project flowing. Instead of taking time in discovery to make sure we've identified the targeted personas correctly, that we have laid out the work in such a way that we have some mapping, that we have a clear vision and strategy, what we're trying to do. You skipped all of that and went straight to coding, then straight to review and straight to testing. And so for example, in a review, you say that in the human era, we had to have a real code review with comments and fixes. Now AI can self critique. It can do self critiquing.
8:37 Speaker 00 It can do agent based reviews and all the fixes in 10 seconds. In testing QA would test things in the past and could take up to a day. The agent can generate, run its own unit test and test itself in five seconds. So the total cycle time for a sprint for a specific item in the human era was four days, where in the AI era, it's 45 seconds.
9:01 Speaker 00 Let's pause.
9:05 Speaker 00 I think if you're looking at, you know, I could, I could paint a wall in six seconds. I could do that. Just grab the bucket of paint, splash it on a wall, the wall's painted, right? Does it mean it was done correctly? Does it mean that other people are going to walk in and say, now that's a well painted wall? Is it something that I would want to use to paint all the walls of my home or to paint my car? I think what you're trying to do is justify speed of deployment and speed of delivery. Saying that if we just crank stuff out, somebody's gonna want it. And that is what I am calling the bigger problem of the 2026. You've got the math right. You've just got the problem wrong. Okay?
9:50 Speaker 00 So this person goes on to say agent driven development is not just using Copilot. It's a paradigm shift where agents on their, you know, operate in their own life cycle of the task. And humans move into the role of goal setters and safety netters. Oof.
10:06 Speaker 00 That might be the most harsh. Okay. So I'm going to challenge you right now, friend. I want you to go look up Copilot's terms and conditions. And specifically where it says that Copilot should be used for entertainment purposes only and should not be used for anything that is serious in nature.
10:24 Speaker 00 I think if you were just doing this for fun, or who knows? Maybe you're being snarky and at the end you're gonna say, just kidding. But right now my skin is crawling with this article. And I also went and looked on Indeed, well, I didn't really, but you can imagine. And I didn't see any jobs for goal setters or safety netters. So what you're saying is people are just out of work at this point? Just move the fuck. And then they go in and talk about the three core pillars of agent-driven design. And that would be like, you know, outcome-based goals, continuous automation loops, observability over status updates. And they give examples of this autonomy loops. I think this person
11:08 Speaker 00 is serious. They say step one should be the goal, and that does require human input. We're going to implement a dark mode toggle switch. Yay. Then they have the orchestrator, which is the leader who approves the use of AI to do this. Then the loop is autonomous. The quality gate or safety net is autonomous. And you replace all the rituals with the agent doing everything for you. You replace stand-ups with a live dashboard. So you don't have to worry about anything. Replace backlog refinement with limited, limited interactive prompt engineering. Wow. So now you're saying that, oh, you know, they can even fix its own prompts. I mean,
11:52 Speaker 00 don't get me wrong. I'm all about, or he's not that. Replace sprint retrospective with loop tuning. And the role of the developer? Well, there is no developer. This person does want to say developers aren't fired, but the job description is changing from I write the code for the login button to I designed a safety gate that ensures that AI didn't break the login button it created.
12:19 Speaker 00 All right.
12:22 Speaker 00 If this person is just trying to say, speed is the new standard. I don't disagree that fast is important. But I do disagree that you are pushing everything to AI haphazardly and without regard. You have not mentioned one time in here, you know, meeting, meeting a standard of high impact or meeting a standard of value. In fact, the words impact or value have not appeared in this entire article that was sent to me, which is quite sad. So what I'm trying to say is while I feel like you have some factual statistical data inside of the article, I don't think that it was put together in such a way that it's
13:06 Speaker 00 meaningful for people who can actually use it. I think we need to rediscover. And I think, you know, I'm begging that you take some time. And if, if you need to come to one of my AI classes to help you figure this out, I'm going to personally invite you to join me at an AI class. And yeah, we'll help you get this figured out. That's going to do it for today. Agile isn't dead yet, still finally or ever. And if you have a topic you want to discuss or an article you want me to review, feel free to reach out. I'm happy to do that for you to accommodate. You can do that. Learn more at that.com. We'd love to hear from you. And as always, we encourage you to stay healthy, stay well, and stay agile, my friends. Until next time, do take care. We'll see you soon. Bye now.
Previous Simplicity in Agile: The Art of Maximizing Work Not Done Next 5 Hidden Costs of Sprint Planning And the Agile Approach That Actually Works Page updated November 6, 2026 · transcript and summary are machine-generated
