Factory debuts Droids coding agents with fresh $50M backing

Most AI coding tools want you to abandon your setup and live inside their custom editor. Factory is taking the exact opposite approach.

The startup just raised a massive $50M Series B and officially unleashed Droids, its autonomous software engineering agents. The round drew big names including Sequoia, NEA, Nvidia, and J.P. Morgan, alongside angel checks from tech veterans like Frank Slootman and Aaron Levie.

Along with the cash, Factory revealed that Droids recently took the No. 1 spot on Terminal Bench, the key evaluation benchmark also used to test tools like Claude Code and Cursor.

Agents that meet you in the terminal

Founded in 2023, Factory is building around an agent-native philosophy. Instead of autocompleting a couple of lines while you type, Droids are built to handle complete engineering workflows while humans stick to high-level architecture.

The clever twist is that Droids are both model-agnostic and interface-agnostic. You do not have to migrate to a new IDE. Developers can interact with Droids from the command line, their existing editor, Slack, Linear, the browser, or custom automated scripts.

To make meaningful pull requests, an agent needs context that goes far beyond raw syntax. Droids hook directly into an organization's existing toolchain, ingesting data from GitHub, Jira, Datadog, Slack, and Google Drive.

Factory claims this allows the agents to onboard like experienced engineers and build an accurate mental model of how the software operates. From there, Droids tackle complex refactoring, documentation, pull request reviews, and even live incident resolution.

Enterprise names are already biting. Factory counts Nvidia, MongoDB, Zapier, Bayer, and Ernst & Young among its early enterprise users. Across these deployments, Factory claims feature delivery has jumped by up to 31 times, while migration and on-call triage times have plunged by 96%.

Why it matters

For engineering teams, Droids represent a far more practical entry point into agentic workflows. Forcing an entire engineering department to abandon VS Code or JetBrains for a proprietary AI fork is a tough sell. By living in the terminal, Slack, and existing toolchains, Factory removes that friction entirely. If your team is buried under maintenance tickets or migration slog, this is definitely worth testing.

For the broader industry, this moves the battle line from clever autocomplete toward fully autonomous project execution. When players like Nvidia and J.P. Morgan back a platform that is entirely model-agnostic, it sends a clear signal: the real enterprise moat is not the underlying model weights, but the contextual plumbing that ties agents to corporate infrastructure.

We think Factory's decision to avoid building another closed garden is brilliant, but it is not without risk. Context ingestion across noisy tools like Slack and Jira often leads to hallucinations in messy legacy environments. Whether Droids can sustain those claimed 31x velocity gains outside controlled enterprise environments remains the real test.