Crafting announced the general availability of Crafting for Agents alongside a $5.5 million seed funding round aimed at building infrastructure for AI-driven software engineering. The round was led by Mischief with participation from WndrCo and angel investors including Andy Manoske of CoreWeave, Arash Ferdowsi, co-founder of Dropbox, Lenny Rachitsky of Lenny’s Newsletter, Gokul Rajaram, formerly of Meta and Square, and Merrill Lutsky of Graphite.
The San Francisco-based company, founded by former leaders from Google, Meta, Uber, and Discord, is focused on addressing a growing challenge for engineering organizations as AI-generated code becomes more common. According to a recent developer survey cited by the company, about 42% of committed code today is generated or assisted by AI, a share expected to increase to 65% by 2027. While autonomous agents are increasingly opening pull requests, monitoring repositories, and writing production code, validation infrastructure has not kept pace.
Crafting’s platform is designed for engineering teams at AI-first companies running complex, multi-service architectures with real dependencies and data. Early customers include Faire, Brex, Webflow, Verkada, Persona, and Instabase.
Crafting for Agents enables AI agents to operate within production-like environments that mirror real enterprise infrastructure. Agents can write code, test it against live dependencies, identify failures, iterate on fixes, and deploy changes within a unified environment shared with human engineers while maintaining enterprise security controls.
The platform includes capabilities such as Kubernetes interception, multi-service orchestration, hot-swapping services, and automatic failover across regions and cloud providers. By creating environments tailored to each company’s tech stack, Crafting aims to allow teams to validate AI-generated code safely before deployment.
Companies using the system report shipping 25% more pull requests quarter over quarter without expanding engineering teams. The company also said AI-generated code at some organizations has grown from single-digit percentages to roughly 70% of output within a year, while engineers save about 2.5 hours per week on environment setup tasks.
Crafting positions the platform as a complement to existing CI/CD pipelines rather than a replacement, enabling faster validation cycles for both human engineers and AI agents.
KEY QUOTES
“We started Crafting to alleviate many of the challenges we were experiencing ourselves as engineering leaders at leading AI-first tech companies. The companies scaling fastest right now are the ones removing bottlenecks safely and letting agents operate inside real infrastructure. With Crafting, teams can scale agent output without compromising enterprise-grade security.”
Sumeet Vaidya, Co-Founder And CEO, Crafting
“Before Crafting, every path we explored was either a point solution or something we had to stitch together ourselves. Now, thanks to Crafting, we have scaled our entire agentic stack, enabling on-demand agents with access to internal systems, MCP servers, and cloud resources in a secure and scalable way.”
Cheuk-man Kong, Senior Engineering Manager, Faire
“We’re seeing a lot of tooling in this category that puts agents in a sandbox and calls it progress. Crafting takes a different approach with a platform that allows engineers to ship validated code, not just push it to production. That’s the real unlock for ambitious engineering organizations that care about secure autonomy.”
Dustin Moring, General Partner, Mischief
“The shift to agentic engineering requires more than incremental tooling. It requires a system designed to unify human and agent engineers inside complex, enterprise architecture. Crafting is solving that problem end-to-end.”
Andy Manoske, Security Product Lead, CoreWeave
“Crafting is the most elegant approach I’ve seen in unblocking the test/validate/deploy process for large engineering orgs that value velocity without compromising quality. With the launch of Crafting for Agents, they’ve brought the same capability to AI-first engineering companies to allow agents to operate safely and effectively inside real enterprise systems.”
ChenLi Wang, General Partner, WndrCo
