The Agent-Native Cloud: 3M Users, 100K Signups/Wk, Data Centers, & Death PRs — Jake Cooper, Railway
TL;DR
Jake Cooper, founder of Railway, explains how the 'agent-native cloud' hit 3 million users and 100,000 weekly signups by betting that manual coding is obsolete, detailing their journey from a $500K/month free tier loss to bare metal infrastructure ownership.
🤖 🤖 Agent-Native Development 3 insights
Stop writing code by hand
Cooper argues developers should prompt agents to generate code and focus on reviewing and reconciling it to their standards, as writing manually is now inefficient when tools can move 'extremely extremely quickly'.
Evolution to 'words' as code
Software development has progressed from assembly to C to JavaScript to natural language, with agents representing the next dominant species that will define the next decade regardless of temporary 'inference walls'.
Architectural expertise still critical
While agents handle implementation, understanding architectural patterns matters more than ever for validating that generated code meets production standards.
📈 📈 Growth and Business Evolution 3 insights
The six-year overnight success
Railway grew from manually onboarding its first 100 users via Discord to 3 million users through alternating 'expansion' phases (adding features) and 'compaction' phases (refining the ICP and stripping non-essential features).
The half-million dollar free tier mistake
Early hypergrowth came from a free tier costing $500,000 monthly against just $50,000 in revenue, forcing the company to temporarily close free signups and rebuild for sustainable unit economics.
100K weekly signups on a lean team
Despite current growth of 100,000 users weekly, Railway maintains only 35 employees by prioritizing systems and automation over headcount expansion.
🏗️ 🏗️ Deep Infrastructure Control 3 insights
Bare metal to kernel patches
Railway owns its entire stack down to data centers and modifies the Linux kernel directly to achieve desired user experiences, treating technical depth as 'figureoutable' in service of the product.
Versioned environments over staging
The platform treats infrastructure as versioned software, enabling users to fork production environments, test changes with real data, and collapse updates without traditional staging complexity.
Progressive rollouts vs. broken pointers
Cooper critiques GitHub's model as 'broken pointers' and advocates for percentage-based deployments where updates reach low-impact users first before flowing upstream to risk-averse enterprises.
Bottom Line
Start using AI agents to write your code today while applying your architectural expertise to review and reconcile the output, as the ability to orchestrate agents will define the next decade of software development.
More from Latent Space
View all
🔬 "The Most Innovative Diffusion Research Is Happening in Drug Discovery, Not Image Generation"
Evan Fineberg and Sergey Udov of Genesis Molecular AI discuss how diffusion models have pivoted from image generation to drive breakthroughs in 3D protein structure prediction. They detail how their Pearl model applies LLM-style scaling strategies—including synthetic physics-based training data and inference-time 'thinking'—to solve the historically intractable challenge of predicting how small molecules bind to proteins.
Cooking with OpenAI’s Research Chief: AGI, o1, Evals, and Scaling Laws — Mark Chen
OpenAI Chief Research Officer Mark Chen discusses the company's research philosophy while cooking Korean tofu stew, emphasizing that scaling laws remain robust, reinforcement learning excels in objective domains, and successful research organizations balance top-down vision with bottom-up conviction.
The Agent Cloud: Databricks’ Bet on the Future of AI — Matei Zaharia and Reynold Xin
Matei Zaharia and Reynold Xin detail Databricks' open-source 'Agent Cloud' platform (Omnigen), arguing that standardized protocols and persistent infrastructure—not just better models—will determine which enterprises successfully deploy collaborative, secure AI agents at scale.
AI Security After Codex and Claude Code — Zico Kolter & Matt Fredrikson, Gray Swan
Gray Swan co-founders Zico Kolter and Matt Fredrikson explain why AI systems require a fundamentally different security approach than traditional software, highlighting how their automated red teaming system 'Shade' has begun to outperform human experts at finding model vulnerabilities. They emphasize the urgent need to treat AI agents as inherently untrusted entities capable of correlated failures across the software ecosystem.