Meta has acquired Conductor, a startup whose software lets developers coordinate several AI coding agents across separate workspaces, adding another developer-focused asset to the company’s expanding artificial intelligence business. Conductor confirmed the acquisition in a message from founder and chief executive Charlie Holtz on November 4. Meta also confirmed the deal to TechCrunch, but neither company disclosed the purchase price or detailed terms. The acquisition is notable not only for the technology Meta is bringing in, but also for the transition Conductor has announced for existing customers.
Conductor was built around a workflow that differs from the familiar pattern of asking one chatbot to write or repair a block of code. A developer can divide a project into separate tasks, assign those tasks to multiple agents and let them work in parallel environments. The aim is to reduce the time spent waiting for one sequence of prompts and edits to finish before another can begin. Each agent can concentrate on a compartmentalised part of the project, while the developer remains responsible for reviewing and combining the results.
The distinction is important because parallel work introduces coordination problems of its own. Agents may change overlapping files, make incompatible assumptions or produce code that appears complete in isolation but does not fit the wider application. Separate workspaces give developers a way to inspect changes independently before deciding what should be merged. That makes the product less like an autonomous replacement for a software team and more like a control surface for managing several AI-assisted work streams.
A multi-agent workspace moves inside Meta
Conductor said it launched its product in July 2025, when the idea of routinely delegating software work to teams of agents was still relatively new. Its public announcement did not disclose the acquisition value, the number of employees joining Meta or the precise role the product will play inside the company. Holtz said the team had plans for a new interface and expected to have something new for customers to use early next year. Those comments suggest that Meta may be interested in the underlying interaction model as much as the existing application.
The company’s announcement also makes clear that the acquisition is changing the product’s availability. New users can no longer sign up or download the app, while existing free users cannot upgrade to a paid plan. Existing Pro users can continue to use Pro features without charge during the transition. The company has described these measures as part of the move into Meta rather than as a normal product expansion, so prospective users should not treat Conductor as an open service for new deployments.
Cloud customers face a defined shutdown timetable
Conductor Cloud has a separate and faster timetable. Its cloud workspaces are scheduled to become read-only on November 13, 2026, after which users will still be able to read existing conversations and export data but will no longer be able to send new messages. The cloud service is due to shut down on November 20, when the company says it will begin deleting cloud user data. Active subscribers are due refunds, the company says’s announcement.
The startup has provided a migration route for customers who want to retain cloud workspaces locally. Users can transfer a cloud workspace and its chats to a Mac, or export an archive through the account settings. The announcement says local Conductor will continue receiving new models and critical bug fixes until early 2027, after which the existing application will begin winding down. Customers with important project history should export it before the cloud deadlines rather than assume the acquisition will preserve every hosted feature.
Meta’s developer strategy is becoming more integrated
For Meta, the acquisition fits a broader effort to make AI useful inside software development rather than limiting it to consumer chat. The company has already introduced its own coding-agent products, and Conductor provides a different way to orchestrate agents supplied by major AI laboratories. TechCrunch reported that the startup’s agents were not built as a proprietary model family; the product coordinated agents from providers including Anthropic and OpenAI. That distinction could let Meta focus on the developer interface and workflow layer instead of relying on one model alone.