Replit grew annual revenue from $2.5M to $250M in a single year (2024 to 2025), a 100x increase.

“We went from 2.5 to 250 million dollars in 1 year.”

Replit Agent's launch generated $1M in new ARR on day one and $2M on day two.

“The revenue first day made like a million dollars of ARR. Second day, 2 million dollars.”

Amjad Masad on Replit's revenue trajectory.

“We're on our way to a billion this year.”

Replit's self-improving agent loop works by autonomously analyzing traces, generating pull requests with prompt changes, and deploying them to production. It improves context and prompts rather than model weights, but still constitutes a meaningful autonomous self-improvement cycle.

“It's not improving its weights, but it's improving its contacts, which is as all running itself autonomous loop.”

An AI agent handling marketing email follow-ups for 120 people costs $254 per month, cited as evidence of agents replacing human labor at a fraction of the cost.

“Running these the perfect email you saw following up with 120 folks, it's $254 a month on there.”

Andrej Karpathy's reaction to the Replit Agent launch, describing it as an AGI milestone.

“This is the I feel the AGI moment.”

Amjad Masad declares coding skill obsolete, reframing the goal as creation.

“I no longer think you should learn how to code.”

Replit, used as an autonomous coding agent that spawns sub-agents (including one on OpenAI Codex) for complex tasks and collaborates with Claude via MCP without direct human orchestration.

“When Claude on Opus talks to Replet on sonnet if it's a big feature, Replet brings in a sub agent called the architect and the architect it turns out runs on codecs on OpenAI.”

How a multi-model agent architecture emerges automatically: when Claude (Opus) assigns a large feature task to Replit (Sonnet), Replit internally spins up a sub-agent called 'the architect' that runs on OpenAI Codex, with no explicit human orchestration.

“When Claude on Opus talks to Replet on sonnet if it's a big feature, Replet brings in a sub agent called the architect and the architect it turns out runs on codecs on OpenAI.”

Replit has built and deployed an internal self-improving agent that closes a feedback loop from production traces to prompt changes to deployment, making it an early live example of autonomous agentic infrastructure.

“Now we have a self-improving loop. This is the crazy thing that everyone needs to understand today is that we are on the precipice of unlocking Self-improving agents is like the stuff of sci-fi or nightmares depending on what kind of sci-fi you read.”