Factory

Factory builds Droids, autonomous software-development agents that take an entire task — a migration, a refactor, a code review, an incident triage — rather than autocompleting the next line. The same agent runs in a desktop app, a CLI, the browser and Factory-managed cloud machines, against whichever frontier model you choose, with enterprise controls for regulated engineering organisations.

Factory sells an agent-native software development platform built around Droids: autonomous coding agents that take on a complete task — a refactor, a framework migration, a code review, an incident triage — rather than autocompleting the next line in an editor. The deliberate design choice is surface- and model-independence. The same Droid runs in the Factory desktop app, in the Droid CLI, in a browser at app.factory.ai, and on Factory-managed cloud machines called Droid Computers, and it can be pointed at different frontier models rather than being welded to a single vendor's. Droids connect to repositories and the systems around them — GitHub, issue trackers and incident-management tooling — so an agent can read a ticket, open a branch, and hand back a reviewable pull request, and the documented automation targets are triage, code review, QA, documentation and incident response. Factory's pitch to enterprises is orchestration, policy and auditability rather than raw model quality: Business and Enterprise tiers add SSO, SAML and SCIM, zero data retention, audit logging, customer-managed encryption keys, data-residency guarantees, dedicated compute with partitioned inference, and on-premise deployment. On benchmarks, Factory's entries took first and third place on Terminal-Bench at 58.8% and 52.5% task success, and its earlier Code Droid posted a state-of-the-art SWE-bench result in 2024. Founded in San Francisco in 2023, the company raised a $50 million Series B in September 2025 from NEA, Sequoia Capital, NVIDIA and J.P. Morgan after a $15 million Sequoia-led Series A, with angels including Frank Slootman, Nikesh Arora and Aaron Levie. Named users span Blackstone, Adyen, Wipro, Comarch, Groq, Chainguard, Podium, MongoDB, EY, Bayer, Zapier and Clari.

Category

Developer Tools

Pricing

Subscription, Contact for pricing

Target Market

CTOs, VPs of Engineering, Enterprise Developers, Platform and DevEx Teams

Deployment

Cloud-first, Hybrid, API-based

Founded

2023

Headquarters

San Francisco, CA, United States

• ✓
  Droids (autonomous task agents)

  Agents that own an entire unit of work end to end rather than suggesting the next line of code.

• ✓
  Surface independence

  The same Droid behaves identically in the desktop app, CLI, browser and Factory-managed cloud machines, so workflows do not fork.

• ✓
  Droid Computers

  Factory-managed cloud machines that run agents in the background, freeing the developer's local environment during long tasks.

• ✓
  Model independence

  Agents can be pointed at different frontier models, so a model change does not require replacing the tooling around it.

• ✓
  Repository and workflow integrations

  Connects to GitHub, issue trackers and incident-management tools so agents work inside existing engineering process.

• ✓
  Enterprise identity and data controls

  SSO, SAML and SCIM, zero data retention, audit logging, customer-managed encryption keys and data-residency guarantees.

• ✓
  On-premise and dedicated compute

  Enterprise tier offers on-premise deployment and partitioned inference for organisations that cannot use shared capacity.

• ✓
  Agent-readiness dashboard

  Usage statistics and billing tracking that show where agents are actually being used across a team.

• •
  Large-scale codebase migration

  Run agents across repositories to move a framework or language version, with each change surfacing as a reviewable pull request.

• •
  Automated code review pass

  A review Droid inspects incoming pull requests and leaves findings before a human reviewer spends time on them.

• •
  Incident triage and response

  An agent pulls context from logs, tickets and source, drafts a diagnosis, and shortens the path to a fix during an outage.

• •
  Backlog burn-down of mechanical work

  Delegate dependency bumps, test coverage gaps and documentation debt that engineers routinely deprioritise indefinitely.

• •
  Regulated-industry AI rollout

  Deploy agents under zero data retention, audit logging and data residency so security review can actually approve the tool.

Individual tiers are published and flat — $20, $100 and $200 per month — but they are differentiated by opaque 'usage limits' expressed only as multiples of Pro (roughly 5x at Plus, 10x at Max), so the real cost of a heavy agent workload is not knowable before you run it. There is no free tier and no advertised trial. Team pricing is seat-based and entirely contact-sales: Business caps at 150 seats, Enterprise is unlimited, and everything a security review will demand — on-premise deployment, customer-managed encryption keys, data residency, sub-organisations — sits behind the Enterprise quote.

This page was written from 6 sources, 4 on domains other than factory.ai.

1. 1.factory.ai — factory.aivendor
2. 2.factory.ai — pricingvendor
3. 3.docs.factory.ai — docs.factory.ai
4. 4.implicator.ai — factorys 50m bet on agent native coding meets a noisy market
5. 5.thesaasnews.com — factory raises 50 million in series b
6. 6.hn.algolia.com — search