Factory's Droid gets sign-in Connectors and role-based Personas
- Source
- x.com
- Date
Droid now works everywhere you do. Connect @Google, @linear, @databricks, @NotionHQ, @Snowflake, @figma, @datadoghq and over 50 additional tools. Introducing Connectors and Personas.
Context
Factory, which makes the Droid coding , now lets Droid use outside apps through hosted Connectors instead of servers the user sets up. Its announcement names Google, Linear, Databricks, Notion, Snowflake, Figma and Datadog and says more than 50 other tools are supported. Factory's launch post describes the setup: pick an app in a conversation with Droid or under Settings, sign in to that app, and the connection then works across Droid's desktop, command-line, web and mobile versions.
The difference from Droid's existing route is where the integration runs. With , the Model Context Protocol, a user configures each tool server in a settings file and runs it locally or points Droid at a remote one. Factory's Connectors documentation says connector tool calls instead go through a Factory-managed service, with no connector process on the user's machine, and each connection is tied to one person and one organization rather than shared. MCP stays available for custom, private or self-hosted tools, and both can be used in the same session. Connector actions follow Droid's existing autonomy setting: the documentation treats read-only tools as low risk that can run automatically, while writes and destructive actions need confirmation unless that setting allows them.
Personas are a separate onboarding layer. At first launch a user picks a role, such as Engineering, Product, Finance or Sales, and that choice shapes the first tasks Droid suggests. In Enterprise organizations, owners and managers can attach recommended connectors to each role and decide which apps members may connect at all. A recommendation does not grant access, since each person still signs in to their own account, and switching an app off removes its tools even from sessions already running. The role examples Factory gives, like tracing a spike in checkout timeouts through monitoring tools to a bad commit, describe intended workflows rather than measured results. The announcement video was not reviewed.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
- MCP — The Model Context Protocol — an open standard that lets any AI assistant plug into any tool or data source without custom integration code.
Checking sign-in…
Loading comments…






