In this short monologue, Jason M. Lemkin positions himself as a "presentist" rather than a futurist and argues that we are on the cusp of a "post‑software" era. He cites Amjad’s vision of programming directly in English and describes his own experience on Replit, where he can build functional applications with a single sentence to an AI agent, bypassing traditional specifications and prompts. Lemkin contends that the line between software and conversational interaction will blur as agents autonomously decide whether to call APIs, retrieve data, or spin up full applications—often without the user recognizing that software was created at all. He predicts that this invisible, agent‑driven software layer will become evident within the next 18 to 24 months, fundamentally changing how we think about building and using software.
Chapters
I'm not a futurist, I'm a presentist, but I genuinely think we will enter not the democratization of software, which is what REPLs are enabling, but post-software. Yeah, I think that we're just entering the era. You know, Amjad talks, I think Amjad talks specifically about programming in English. Doesn't he say that? And, you know, a couple months ago I would've been like, whatever, that's the marketing of the CEO. I think it's kind of true today, programming in English. I mean, literally, I, in Replit, I don't need to design a prompt anymore. I don't need to have a spec. Now, I think most people should have a spec that are new to it, but I've done enough with the agent and I've done enough of my apps, honestly, with one sentence. I, I know it's not best practice. I, I can do a lot of stuff now if, if, if I have the context in the agent. Okay. So we're just entering the area of programming in English, but to most of the world, it's really weird. Okay. This is honestly what I think deep in this community. Is the line between software and non-software will blur. Yeah. Like our lives will be prompts and no, because just like agents, you know, prompt, prompting's becoming just English. And then, but what's the line between software and just talking to an agent? Like we won't know, like the agent may spin up and create software for real and we won't even think about it as software anymore. The agent will make a decision. Should I plug into an API? Should I just answer it? Should I just grab some data? Should I grab it from Databricks or Claude? Or, you know what, I'll create an application. You won't even know I built it. It'll be an invisible application. But that's what I think. I don't mean— I'm not a futurist, I'm a presentist, but I genuinely think we will enter not the democratization of software, which is what RepliZenabling, but post-software. Like, we will be, we will be in the ether of software, but I don't know exactly what it is going to look like. Right. But the rate of change is so fast. You can see that happening in 18 to 24 months. Right. The software is just invisibly created behind us by agents.
