Jason Lemkin on AI Productivity Gains
SaaStr's Jason Lemkin, amplifying a thread by Andrej Karpathy, made a stark claim about the impact of AI agents on engineering velocity. He argued that top startups have seen massive productivity gains since December, stating, "If your startup hasn't gotten wildly more productive since December, something is wrong."
Andrej Karpathy pinpointed December 2025 as a critical turning point for AI's impact on programming, arguing that AI coding agents crossed a crucial reliability threshold, making them capable of handling complex, multi-step tasks autonomously. This "phase change" is a shift from manually writing code to orchestrating AI agents that can now take a high-level prompt and independently execute a project that might have previously taken a full weekend, completing it in as little as 30 minutes. This leap in AI capability is being driven by a new generation of sophisticated tools. AI-native code editors like Cursor and agent-based platforms such as Devin are becoming central to developer workflows. These tools, along with evolved assistants like GitHub Copilot's agent mode and Anthropic's Claude Code, can now understand entire codebases, write and refactor code across multiple files, and even debug their own work. Startups are reporting significant, measurable gains from adopting these AI tools. One analysis of 100 AI-assisted startups found they were shipping new products 1.8 times faster on average. The biggest productivity boosts were seen in coding and implementation, testing and quality assurance, and documentation. Some companies have reported development time for repetitive coding tasks being cut by as much as 50%. This shift is fundamentally altering the career path for engineers at startups. The role is evolving from a focus on writing code to one of system design, prompt engineering, and acting as a reviewer of AI-generated work. As AI automates more routine coding, the lines between individual contributor (IC) and management tracks are blurring, with a greater emphasis on skills like technical strategy and overseeing AI systems. This creates opportunities for senior ICs to have a greater strategic impact, while leadership roles will increasingly require a deep understanding of how to leverage AI for team productivity.
