Relace Raises $23 Million to Build the Rails for Software On Demand

Relace, LLMs for Code Generation, led by Preston Zhou, Eitan Borgnia, and Julian Thomassie, has secured a $23 million Series A led by Andreessen Horowitz (a16z), with participation from Matrix Partners and Y Combinator.

The rise of AI code agents marks a turning point in software creation: apps, dashboards, and workflows can now be generated instantly, with no need for traditional development cycles. As autonomous code generation accelerates, the volume of code produced will surpass anything seen before, demanding new systems purpose-built for machine-driven creation. Relace is building the foundation for this new era: compact, specialized language models running on optimized infrastructure that make code agents faster, scalable, and production-ready from day one.

As the coding agent market heats up, many teams are independently recreating the same toolsets, execution environments, and source control systems. Just as databases and deployments are now outsourced in web development, agent design choices will increasingly become managed services as software-on-demand reaches non-technical users. Infrastructure must be co-optimized with models for speed and performance. Relace is already delivering measurable gains for companies like Lovable, Magic Patterns, and Orchids, surfacing relevant code context in 1–2 seconds and merging file edits at over 10,000 tokens per second.

As software generation becomes faster and cheaper, codegen will permeate every industry. Designers will translate mock-ups directly into code, marketers will generate graphics via code, and professionals will create presentations by chatting with AI agents. Just as phone cameras enabled Instagram, smaller, faster coding agents will redefine what’s possible. Platforms like Gizmo already showcase creative web apps generated with GPT-5/Gemini Flash, and on-device models could make code generation nearly free. Moore’s Law continues to surprise, and these advances will unlock entirely new possibilities in software creation.