article-3-chatgpt-deep-research-connectors

What You'll Learn

• The full list of new ChatGPT Deep Research connectors and their capabilities
• How GitHub enterprise pre-indexing works and what it means for codebase analysis
• Data privacy guarantees: what OpenAI actually stores and what it doesn't
• Rollout timing for each connector tier (Team, Plus/Pro, Enterprise/Education)
• Practical workflows for research across files, code, email, and customer data simultaneously

ChatGPT Just Became Your Personal Research Assistant

OpenAI has quietly expanded ChatGPT Deep Research with 11 new connectors, turning the model into a cross-platform research engine that can pull from Dropbox, GitHub, Google Drive, SharePoint, Box, Outlook, Gmail, Google Calendar, Linear, HubSpot, and Microsoft Teams simultaneously. A user can ask ChatGPT a complex question, and the model can now synthesize answers by searching across all of these platforms in a single chat session without ever switching windows or copying data between tools.

This is a significant product pivot. ChatGPT's original advantage was a large language model. But a large language model trapped in its own context window is limited by knowledge cutoff dates and can't reason about your actual data. The connector ecosystem changes ChatGPT from a general-knowledge engine into a personal research engine that operates on your real-time, proprietary information.

The expansion is rolling out in stages. Dropbox connector is available globally to Team users. GitHub is in beta for Plus/Pro/Team. Other connectors are on a staggered rollout schedule. Enterprise and Education tiers get special treatment: GitHub repositories are pre-indexed by OpenAI, meaning ChatGPT can analyze your entire codebase instantly without waiting for indexing.

The 11 New Connectors: Capabilities and Coverage

OpenAI's connector roadmap spans the complete data lifecycle of a knowledge worker:

GitHub Enterprise Pre-Indexing: The Technical Detail That Matters

The GitHub connector works differently for Enterprise and Education tiers versus Plus/Pro/Team. For smaller accounts, ChatGPT indexes the repository on demand, which takes time proportional to repository size. For Enterprise accounts, OpenAI pre-indexes the entire repository, meaning ChatGPT can search and analyze your codebase instantly without any waiting period.

This is significant for two reasons:

The pre-indexing process is handled entirely by OpenAI. Your code is indexed and stored on OpenAI's systems but access is authenticated through GitHub OAuth. OpenAI cannot see repositories you don't have access to, and the indexed data is encrypted in transit and at rest.

Data Privacy: What OpenAI Stores and What It Doesn't

This is the critical question for enterprise users: does OpenAI retain your data The answer is nuanced.

What OpenAI Does Store

OpenAI stores encrypted indexes of the data you query. These indexes are necessary for the connectors to function. The indexed data is deleted when you revoke connector access. Additionally, OpenAI may train future models on aggregated, anonymized patterns from connector data across all users (with opt-out available for Business/Enterprise).

What OpenAI Doesn't Store

OpenAI doesn't store your actual files. When you query Dropbox through ChatGPT, the model receives the file content during the query session, generates a response, and doesn't retain a copy of the file itself. Your email addresses, phone numbers, and customer details from HubSpot are similarly not persisted in OpenAI's systems beyond the active query session.

Access is governed by your third-party permissions. If you revoke ChatGPT's access to Dropbox, OpenAI's cached index of your Dropbox files is deleted. If you revoke Gmail access, OpenAI can no longer read your emails. The revocation process is instantaneous.

Practical Workflows That Now Become Possible

The connector ecosystem enables workflows that required manual data aggregation previously:

Rollout Timeline and Tier Availability

Not all connectors are available immediately and not to all tiers:

Action Steps Summary

How to Use This Capability

• Enable Connectors for Your Use Case : Assess which of the 11 connectors aligns with your current workflow. Start with one (Dropbox or GitHub) to test the integration.
• Connect Your Primary Data Sources : Enable access to your most frequently referenced systems first. For knowledge workers, this is typically email (Outlook/Gmail) and file storage (Dropbox/Google Drive).
• Design Your Research Workflows : Before relying on connectors, map out which questions you ask repeatedly. These are the candidates for connector-based research.
• Review Data Access Settings : Understand what ChatGPT can access. For Enterprise users with HubSpot connector enabled, ensure you're comfortable with customer data queries running through OpenAI's systems.
• Set Up Revocation Protocols : If employee turnover is frequent in your organization, establish a process for revoking ChatGPT connector access when people leave.
• Test GitHub Connector for Codebase Analysis : If you're Enterprise tier, the pre-indexed GitHub connector is immediately useful. Test it with architecture and pattern analysis queries.