Key Takeaways

  • The core ambition for builders has shifted from "learn to code" to "learn to create," driven by new AI capabilities that abstract away programming complexities.
  • Agentic AI, emerging in 2024, was the critical unlock. It allows AI to perform sequences of actions over a long horizon, unlike earlier LLMs such as GPT3 in 2021-2022.
  • Solo founders can now build multi-million dollar businesses without needing traditional coding skills, focusing instead on defining what they want to create.
  • For platforms that support these builders, staying competitive means constantly building sophisticated infrastructure around evolving AI models, predicting their next capabilities.

From Code to Creation: The Solo Founder's New Path

For years, the mantra was "learn to code." But Amjad Masad, CEO of Replit, now sees that advice as outdated. On 20VC with Harry Stebbings, Masad detailed Replit's journey, explaining a fundamental shift in what builders truly want. People don't want to become programmers; they want to bring ideas to life.

Masad points to examples like Jason Lein of SaaSter, observing a new class of solo founders. These individuals are building multi-million dollar businesses without a single developer on staff. Their secret? They don't need to write code. They need a platform that lets them create. As Masad puts it, “people that are building multi-million dollar businesses solo with no developers, they don't need to learn how to code. They need to learn how to create. They need to learn how to build.” This practical reality is already playing out, showing that the barrier to entry for turning an idea into a shipping product has dropped dramatically. It's not because everyone learned to code, but because coding itself became less necessary.

Agentic AI: The 2024 Unlock

What changed? It wasn't merely the arrival of large language models like GPT3 in 2021 or 2022. Masad identifies a specific turning point: the rise of "agentic AI" in 2024. Earlier AI models could generate text or code, but they required constant human supervision. Agentic AI, however, can execute actions across a longer sequence, making decisions and chaining operations without explicit step-by-step instructions.

“The real unlock wasn't just AI,” Masad explained. “It was AI that could do actions over long horizon. So we've had AI since 2122 with GPT3, but the unlock in 2024 was agentic AI and that was the first glimpse of it.” This ability for AI to act autonomously, to perform a series of tasks to achieve a higher-level goal, gives agency to a "learn to create" builder. Instead of writing a script, you define the desired outcome, and the agent orchestrates the steps.

The Invisible Infrastructure Race

While agentic AI makes creation simpler for the end-user, it complicates things significantly for platforms like Replit. Masad stressed that harnessing these models isn't a "plug and play" situation. To offer users the most advanced capabilities, Replit must constantly build sophisticated infrastructure around the raw AI models. "You at any given point you have to plug in a lot of holes in order to build the most advanced thing that you can build," Masad said.

This means continuously assessing model capabilities, understanding their limitations, and developing the surrounding tools, APIs, and frameworks that make them truly useful and reliable for builders. The market moves fast, and staying at the forefront demands foresight. Masad emphasized, “You have to understand what is the model capability and how much infrastructure you have to build in order to make it perform. And you have to stay ahead of that in order for your product to stay the most advanced on the market.” For founders leveraging these tools, it means picking platforms committed to this unseen, constant engineering battle.

What to Do With This

Audit your current workflows or product's onboarding process. Identify one area where users (or even your internal team) still manually write code for a common task that could be defined by its outcome. Then, experiment with an agentic AI tool—whether it’s a feature within Replit or an open-source agent framework—to see if a non-technical person could "create" that outcome without touching a line of code. The goal is to abstract away the "how" and focus on the "what."