What it does
8 features
Factory · Platform with coding agents (Droids)
Product and access
Factory is the provider and the platform, Droid is the agent and the command-line app. The domain factory.ai redirects to factory.com.
Core features
8 features from provider sources · checked on 5 November 2026. Availability and limits are directly below.
Droid in many places
According to the website, Droid runs in the terminal (interactive or headless), in desktop, web and mobile apps, in Slack and Teams, with Jira and Linear, and in CI pipelines.
Source 1Automations
Droid starts on a schedule, on Slack messages, GitHub events or webhooks. Guided setups exist, such as triage and code review.
Source 11Missions for larger projects
You plan with Droid and approve the plan; an orchestration layer then runs the execution. You can pause it and intervene.
Source 12Droid Computers and Remote Control
Droid Computers are persistent environments hosted by Factory. With Remote Control, Droid works on a computer you manage yourself.
Source 13Autonomy levels and command rules
The autonomy levels are off, low, medium and high; the default is off with manual approval. Commands can be allowed, asked about or blocked by rule.
Source 9Models and your own keys
The documentation names models from Anthropic, OpenAI, Google, xAI and open models. You set up your own endpoints and keys through Custom Models (BYOK).
Source 8
Show all 8 features
Customisable harness
AGENTS.md, connectors, MCP, subagents, skills, custom slash commands, plugins and hooks extend Droid.
Source 2Enterprise controls and EU deployment
Admins control models, MCP, retention and sync. For Factory Managed there is a separate EU deployment for enterprise customers.
Source 5
Usable AI models
Open choice
You choose from Anthropic, OpenAI, Google and xAI models plus open models, or let the Factory Router decide. In the CLI and desktop app you can also add your own models with your own key and any base URL, including OpenRouter, Ollama and LM Studio. According to Factory, only the official Anthropic and OpenAI APIs are fully tested, and organizations can restrict custom models.
Model families named
- Anthropic (Claude)
- OpenAI (GPT)
- Google (Gemini)
- xAI (Grok)
- Mistral
- Z.ai (GLM)
- Moonshot (Kimi)
- Alibaba (Qwen)
- DeepSeek
- MiniMax
- Nvidia (Nemotron)
- Other models can be connected freely
- Model choice
- Automatic or your choice
- Your own API key
- Yes
- Router or custom endpoint
- Yes
- Local models
- Yes
- Depends on
- Admin approval
The rating covers the generally available main access. Documented · Sources accessed on
Does it fit my task?
2 practical tasks · 6 requirements and limits
A good fit if …
- you want to trigger coding tasks from the terminal, Slack or CI and repeat them through automations
- you want to plan larger projects, approve them and have an orchestration layer carry them out
- you need admin controls such as model policies or an EU deployment for your company
Less suitable if …
- you are looking for a free plan
- you need EU data residency without an enterprise contract
- you want to inspect and self-host the agent’s source code
Editorial assessment based on provider documentation. Not a product test.
Practical tasks
Editorial examples based on documented features. No product tests performed.
Recurring code reviews
- Your input
- A GitHub repository and an automation using the code review setup.
- Possible result
- Droid runs on repository events such as pull requests and returns feedback.
- What to check
- Spot-check the feedback and keep the automation’s permissions small.
Feature with plan and approval
- Your input
- A project with AGENTS.md and a clearly described feature.
- Possible result
- Droid drafts a plan, carries it out after your approval and tests the change.
- What to check
- Read the plan before approving, check the diff and tests, keep autonomy low.
Set up a recurring code review task as an automation.
Your possible prompt
Read the project rules. Investigate [error], propose a small change, and run the relevant checks. Show me the diff and any remaining uncertainties.
Adjust the task, data and permissions.
Requirements and limits
Requirement
Paid plan needed
The pricing page lists no free plan; the lowest plan is Pro at 20 US dollars a month.
Source 3Requirement
Missions need testable projects
For Missions, the documentation expects the best results with a repository at Agent Readiness level 4 or above and, in existing projects, a scriptable way to test the app like a user.
Source 12Limitation
Command rules are not a sandbox
According to the documentation, command rules do not cover every program behaviour. It recommends pairing them with sandboxing, managed hooks and least-privilege credentials.
Source 7Limitation
EU deployment only through sales
The EU deployment is an enterprise feature within Factory Managed, for customers with European data residency requirements; the organisation is provisioned through sales.
Source 5Not yet confirmed
Plan coverage of automations and Remote Control
The automations and Remote Control pages do not say which plans include them.
Source 11Not yet confirmed
Training on customer data
The data flows page does not say whether customer data is used for model training.
Source 6
Can I use it?
With requirements
Europe & regional availability
Deployment, data flow and enterprise pages and the pricing page reviewed. The trust centre returned no usable content.
Reviewed scope: Platform with cloud services; code runs on your machine or in Factory environments
Reviewed on:
With requirements
EU deployment for enterprise customers
Factory Managed has a separate EU deployment that, per the page, is available to enterprise customers with European data residency requirements. Session content is stored there in a European database and model requests go only to EU-region endpoints. Organisation records, user profiles and billing stay in the global (US) deployment. Sales provisions the EU organisation.
Region, reason and source
Region covered: EU/EEA
Documented reason: Only for enterprise customers with data residency requirements and through sales.
Official source · docs.factory.com · Official source · factory.com · Official source · docs.factory.com
Availability still unclear
United Kingdom and Switzerland: no statement
The reviewed pages name no availability or data region for the United Kingdom or Switzerland.
Region, reason and source
Region covered: United Kingdom and Switzerland
Documented reason: No country statement in the reviewed sources.
Official source · docs.factory.com · Official source · factory.com
Accounts & permissions
Based on provider documentation. Permissions depend on the account, integration and organization policies. Review the permissions actually requested before use.
Configuration covered: Factory account, sign-in through the browser
Sources reviewed:
Show all 3 points
Documented with conditions
Admin approval
Org admins can control models, MCP allowlists, session sync and retention through managed settings. The pricing page lists basic admin controls for Business and full admin controls for Enterprise. Enforcement of the MCP allowlist is off by default.
Evidence · Evidence
Memory and conversation continuity
2 of 6 criteria documented
Sessions can be resumed and AGENTS.md carries project knowledge; the reviewed pages describe no memory of the agent itself.
Sessions and AGENTS.md of the Droid agent
Sources accessed:
Documentation by criterion
StoragePartly documented
Storage and scope
The reviewed pages do not say where sessions are stored locally. With cloudSessionSync (default: on), CLI sessions are mirrored to the web app; when off, they stay local. Project knowledge lives in AGENTS.md files.
SelectionNot publicly established
Selection and timing
You write and commit AGENTS.md yourself. Whether Droid saves memories itself is not stated on the page.
Edit & deletePartly documented
Inspect, edit and delete
/sessions lists and resumes earlier sessions. Org admins set how long cloud sessions are kept with sessionRetentionDays (14 to 365 days). The pages do not describe deleting or editing individual sessions.
Access boundariesDocumented
Agent and channel boundaries
Droid looks for AGENTS.md from the working directory up to the git root and in .factory/, .agents/ and .agent/; personal instructions live under ~/.factory/. Nested files load when reading, editing, listing or searching in their folders.
RetrievalDocumented
Retrieval
According to the page, AGENTS.md is loaded into every session, with limits of 80,000 characters at startup and 40,000 per tool call. You bring back earlier sessions with /sessions or droid --resume <id>.
New conversationsPartly documented
Session reset and compaction
/fork copies the current session into a new one; the original stays open. The AGENTS.md page says nothing about context compaction.
Cost and effort
6 plans · Prices as of 5 November 2026
Prices and included usage
Pricing checked
The pricing page lists Pro (20 US dollars), Plus (100), Max (200), Teams (60 per team plus 40 per seat), and Business and Enterprise on request. It lists no free plan.
| Plan | Price | Included |
|---|---|---|
| Pro | 20 US dollars per month | Desktop, CLI and SDK, cloud and local background agents, usage stats, agent readiness dashboard; individual use. |
| Plus | 100 US dollars per month | Everything in Pro plus Droid Computers; about 5x Pro usage. |
| Max | 200 US dollars per month | Everything in Plus plus early access to new features; about 10x Pro usage. |
| Teams | 60 US dollars per team plus 40 per seat per month | Everything in Pro for the team, centralised billing, up to 10 seats. |
| Business | On request | Custom usage limits, SSO, SAML/SCIM, zero data retention, audit logging, admin controls. |
| Enterprise | On request | Everything in Business plus on-premise options, sub-organisations, data residency and premium support with SLAs. |
- The pricing page names no token amounts, only ratios to Pro. Models count with different multipliers.
Setup
Install Droid or download the desktop app and sign in through the browser. For automations you set up triggers and instructions in the Factory app.
Operation & oversight
Based on public provider documentation. Assessments are editorial inferences from the stated properties. Actual time requirements and reliability have not been measured.
Deployment covered: Terminal, apps and automations
Sources reviewed:
Editorial assessment
You get fine control through autonomy levels and rules, but you are responsible for isolation, since rules are not a sandbox. For automations, keep permissions small and check results.
Show all 6 points
Evidence and open questions
17 sources · 3 open points
Open points
The reviewed sources contain no reliable information on these points. A gap says nothing about the product.
Europe: Availability still unclear
Requirements and limits: not yet confirmed
Sources and evidence
Website, pricing page and documentation reviewed. Feature descriptions come from the provider; example tasks are editorial. The trust centre could not be evaluated. Hands-on tests are still pending.
Editorial team: Michael Meierhoff · sources reviewed on . Provider information is not independent evidence of performance.
Provider
Factory AI runs the platform and, per the repository README, holds the copyright; the product source is not there, the repository holds documentation and community material.
- Operator
- Factory AI
- Organization
- Provider
- Website
- factory.com
Changes to this profile 1
Development platform with Droids, agents that work on software tasks in the terminal, apps, Slack, Jira and CI.
These notes describe editorial changes. Source and pricing review dates appear in their respective sections.