While Cursor was capturing the hearts of individual developers, a much quieter and much larger game was unfolding inside enterprise organizations.
On April 16, 2026, Factory announced a $150M Series C. Valuation: $1.5B. Led by Khosla Ventures, with follow-on from Sequoia Capital and new entrants Blackstone and Insight Partners. And the customer list is striking: NVIDIA, Adobe, Morgan Stanley, EY, Palo Alto Networks, Adyen. Those names alone signal that this is not a "developer tool" fight — it is an "enterprise infrastructure" fight.
What Is This?
Factory is a startup building AI agents called "Droids." It started in 2023 when Matan Grinberg, then a physics PhD student at UC Berkeley, cold-emailed Sequoia partner Shaun Maguire. Maguire had his own physics PhD (Caltech), the two connected immediately, and Grinberg dropped out to co-found the company with machine learning engineer Eno Reyes.
Droids are not simple code-completion tools. They are agents that autonomously handle the entire software development lifecycle — code generation, test authoring, code review, documentation, and deployment. Unlike Cursor, which operates inside an IDE, Droids receive tasks via Slack, process Linear tickets, and connect all the way to CI/CD pipelines.
The flagship feature, Missions, is particularly impressive. Describe a business outcome in natural language — "migrate the payments service off legacy Python 2 onto a Kotlin microservice" — and multiple Droids plan, execute, and verify the work over hours or days, without requiring constant human intervention.
Model selection is not locked to any single provider. Droids automatically route tasks between Claude, DeepSeek, and other frontier models depending on the workload. For enterprise procurement teams that want to avoid vendor lock-in to any one AI lab, that flexibility is genuinely attractive.
What Changes?
The AI coding tools market has split clearly in 2026 into two segments: IDE assistants for individual developers and autonomous agent platforms for enterprise teams. Factory is squarely in the second category.
| Tool | Primary Target | Core Capability | Model Support | Enterprise Fit |
|---|---|---|---|---|
| Factory (Droids) | Enterprise teams | Missions, multi-Droid orchestration, CI/CD integration | Model-agnostic (Claude, DeepSeek, etc.) | ★★★★★ |
| Cursor | Individual devs / small teams | Fast in-editor code edits, Agent mode | Claude, GPT, Gemini | ★★★☆☆ |
| GitHub Copilot | All developers | Code completion, chat | GPT, Claude, Gemini | ★★★☆☆ |
| Claude Code | Claude-centric developers | Terminal-native autonomous coding | Claude only | ★★★★☆ |
| Cognition (Devin) | Enterprise | Fully autonomous software engineer | Proprietary stack | ★★★★☆ |
The reason Factory has landed customers like Morgan Stanley and EY is not simply a "better model." It is because Droids understand a company's internal systems before writing a single line of code. A Morgan Stanley Droid knows the firm's proprietary risk framework. An NVIDIA Droid understands CUDA driver internals. That accumulated context creates switching costs that a competitor cannot easily overcome by offering a slightly better model.
"The model is a commodity. The value comes from the fifty other things around it — test harnesses, permission boundaries, audit trails, model routing."
— Eno Reyes, Factory Co-founder
How to Get Started: Enterprise AI Coding Agent Adoption
Adopting enterprise AI coding agents like Factory's Droids requires a structured approach. Going straight to "hand everything over" almost always fails.
- Fix the Context First
Organize internal documentation, test coverage, CI/CD pipelines, and internal API specs. If the environment where Droids operate is messy, even the best agent will produce unreliable results. Factory calls this "paving the roads," and it is the most important step. - Start with Narrow, Verifiable Tasks
Delegate work like legacy migrations, test automation, and documentation — tasks where results are easy to verify. Handing mission-critical systems to agents from day one is risky. - Set Audit Trails and Permission Boundaries
In enterprise environments, you must be able to track what code an agent modified and with what permissions. This step is mandatory in regulated industries like finance and healthcare. - Define a Model Routing Strategy
You do not need the same model for every task. Use Claude for complex planning, DeepSeek for high-volume code generation. Optimizing cost and performance by task also helps you avoid lock-in to a single AI provider. - Expand Autonomy Incrementally
Gradually reduce the scope of human review as Droids demonstrate reliability. Factory's approach involves measuring productivity metrics monthly and adjusting autonomy levels accordingly.
Go Deeper
Factory Series C Official Announcement Written by founders Matan Grinberg and Eno Reyes. Covers the Missions architecture and next investment priorities. factory.ai
TechCrunch: Factory hits $1.5B valuation Official funding round coverage. Concisely covers Keith Rabois joining the board and the customer list. techcrunch.com
Factory Missions Architecture Deep Dive An engineering blog explaining design decisions behind how Droids handle long-horizon multi-step tasks. factory.ai
Tech Insider: Factory AI $150M Series C In-Depth A detailed analysis covering investor perspective, competitive landscape, per-customer use cases, and risks. tech-insider.org
