Factory Droid
Factory Droid for engineering agents: planning, implementation, model selection, current plans and governance with clearly separated data rules.
Provider sources checkedWhat does Factory Droid do?
Who is this tool for?
- Developers with recurring maintenance tasks.
- Engineering organizations centrally managing agent permissions and budgets.
Strengths
- Planning, implementation and review are described as a coherent workflow.
- Model choice and organizational controls can be managed together.
- Desktop, CLI and SDK serve different integration levels.
Limitations and requirements
- Governance capabilities differ between individual and organizational plans.
- Model access and execution environment affect cost and privacy.
- Automated reviews do not replace complete security or functional acceptance.
Who is behind the product?
- Company
- The San Francisco AI Factory Inc.
- Headquarters
- San Francisco, United States.
- Company founded
- 2023, by Matan Grinberg and Eno Reyes.
- Product launch
- Not clearly established by the reviewed sources.
A current official company announcement establishes founding and headquarters. The founding year is not presented as the launch date of today's Droid version. The former factory.ai domain now redirects to factory.com, where product and legal pages were reviewed.
Suggested ways to use it
Suggested workflows or verification steps. These are not hands-on tests; for unverified products they are not a recommendation to use the service.
Suggested: investigate a CI failure
- Provide the failed run and relevant recent changes.
- Request a cause and bounded patch with test evidence.
- Review the PR for side effects and inappropriate test changes.
Goal: A traceable repair proposal.
Suggested: team review pilot
- Choose a repository and approved models.
- Have Droid review a representative sample of changes.
- Evaluate useful findings, false alarms and review time.
Goal: A defensible decision on further adoption.
Pricing, allowances and data
What should you check before choosing a plan?
Free requires your own inference access. Current pricing lists Pro at USD 20 per person/month, Plus at USD 100 and Max at USD 200 with larger usage tiers. Teams lists a USD 60 monthly base plus USD 40 per person. Business and Enterprise are quoted individually. Own model access can bring additional costs or separate limits.
What should you check about data and permissions?
The policy covers code, prompts, sessions and connected credentials. Processing is primarily in the US; third parties and BYOK connections have their own terms. Enterprise agreements can govern retention, isolation and model routing. Zero Data Retention listed on higher tiers is not generalized to free standard accounts. The provider requires separate agreements for regulated data.
External reviews and available evidence
No GatwAI hands-on test. This assessment is a source review without an independent original test of the current version.
Reported positives
- Plan and permission differences can be assessed in a bounded organizational pilot.
Reported cautions
- Suggested workflows were not executed. Provider examples and selected testimonials are not independent performance evidence.
Product development and notable changes
Claude Marketplace availability announced
Factory describes an additional procurement route for eligible Anthropic enterprise customers; this does not replace contract review.
Source for this change ↗
Which alternative fits which task?
OpenHands →
For a more open platform for your own agent orchestration.
OpenAI Codex →
For local and hosted engineering tasks in OpenAI's offering.
Frequently asked questions
Is Free entirely costless?
The platform tier has no base fee but requires your own inference access with its costs and limits.
Does Zero Data Retention apply to every plan?
Not universally. Current pricing and the actual agreement must match the account tier used.
