xAI launched Grok Connectors on May 6, 2026 — a set of native integrations that link Grok to the apps where real work happens. The launch covers GitHub, Notion, Linear, Google Workspace (Gmail, Drive, Calendar, Docs), Microsoft 365 (Outlook, OneDrive, SharePoint), plus a “Bring Your Own MCP” option that lets developers wire up custom internal tools through the Model Context Protocol. Grok Connectors lands first in Grok Web with iOS and Android rollouts following. The release positions xAI to compete more directly with Anthropic’s Claude integrations and Microsoft’s Copilot ecosystem on the workflow-AI dimension — where the value of the assistant depends on what it can reach, not just how smart it is.
What’s actually new
Grok Connectors operationalize a pattern that the AI industry has been converging on since late 2025. Instead of users copy-pasting context into a chat, the assistant reads from and writes to the user’s own tools. Anthropic launched Claude integrations with Asana, Atlassian, Slack, and more last year. Microsoft Copilot has integrations baked into Office 365. ChatGPT has connectors for various third-party apps. Grok’s launch brings xAI’s assistant into the same workflow-AI bracket.
The supported connectors at launch:
- GitHub — summarize pull requests, review code changes, manage repositories, search code.
- Notion — search and update databases, wikis, pages within the user’s workspace.
- Linear — integrate tasks, sprints, and issue tracking; create or update issues from chat.
- Google Workspace — Gmail (read, draft, send), Drive (search and access files), Calendar (read events, schedule), Docs/Sheets/Slides (read and edit).
- Microsoft 365 — Outlook, OneDrive, SharePoint with similar capabilities.
The strategic move is the “Bring Your Own MCP” feature. Model Context Protocol — the open standard Anthropic introduced for AI tool connections — is now a way for any developer to wire up custom Grok integrations. Internal CRMs, proprietary databases, internal documentation systems, anything with an MCP server can be a Grok connector. This positions Grok as a hub for enterprise workflows that go beyond Big Tech’s standard apps.
The differentiation versus competitors comes down to a few specific choices. Claude has the deepest workflow integration in 2026, with Anthropic-blessed partnerships and tight MCP support. ChatGPT has the widest consumer reach. Copilot has the Microsoft ecosystem advantage. Grok positions itself as the X-native AI plus enterprise-flexible MCP option, with the additional pitch of being where Elon Musk’s circles already work. The competitive math depends on which workflow gravity wells matter to each customer.
For users, the immediate gain is reducing context-switching. The classic AI-assisted workflow used to be: read context in app A, copy it to AI chat, get response, paste back into app B. Grok Connectors compresses that flow. Ask Grok to summarize the latest PR comments on the auth module; Grok reads GitHub directly. Ask Grok to create a Linear issue for a discovered bug; Grok creates it. Ask Grok to schedule a meeting with the design team for next Tuesday; Grok reads availability from Calendar and proposes a slot.
Why it matters
- Workflow AI is the 2026 race. Raw model intelligence has converged across frontier labs. The differentiator is integration — what apps the AI can reach. Grok Connectors brings xAI into the race.
- MCP becomes a de facto standard. By adopting Bring Your Own MCP, Grok validates Anthropic’s protocol as the open standard for AI tool integration. Network effects favor MCP further.
- Enterprise positioning matters. Microsoft 365 and Google Workspace are the two enterprise productivity giants. Grok supporting both signals serious enterprise intent.
- Developer workflows are well-served. GitHub + Linear + Notion is the modern developer stack. Grok now reads and writes across all three.
- It reframes Grok’s perception. Grok was associated with X/Twitter integration and Musk-adjacent commentary. Connectors reposition it as a serious workflow tool.
- Pressure on Claude and ChatGPT. Both have their own connector stories; Grok’s launch keeps pressure on them to expand and improve integrations.
How to use it today
Grok Connectors are live as of May 6, 2026, rolled out first to Grok Web. Practical engagement steps:
- Access Grok and verify your subscription tier supports Connectors. Connectors require a paid Grok subscription tier; check the current pricing on x.ai.
# Grok access: https://grok.com/ # Or via the X mobile app # Subscription tiers (verify current at x.ai): # - SuperGrok / SuperGrok Heavy (consumer) # - Grok for Business / Enterprise (team) # Most Connectors are paid-tier features. - Connect your first integration. The Connectors panel in Grok Web shows available integrations.
# Steps: # 1. Open Grok Web # 2. Navigate to Settings or Integrations panel # 3. Click "Add Connector" # 4. Select the service (GitHub, Notion, Google, etc.) # 5. Authorize via OAuth (you'll be redirected to the service's login) # 6. Confirm permissions; return to Grok # After connection: # Grok can read from and (where permitted) write to the connected service. - Try a simple read-only query first. Build confidence before letting Grok take actions.
# Examples for each connector: # GitHub: "Summarize the open pull requests in my main repository." # Notion: "Find recent notes in my Notion workspace about Q2 planning." # Linear: "Show me my open tasks in the current sprint." # Google Calendar: "What's on my calendar for tomorrow?" # Outlook: "Are there any unread emails from my manager this week?" - Test write operations carefully. Connectors that can modify data require attention.
# Write-capable operations to test (cautiously): # Create a Linear issue: "Create a Linear issue titled 'Investigate flaky integration test' in the Backend project, assigned to me, priority Medium." # Draft a Gmail message: "Draft an email to alice@example.com about the rescheduled meeting on Thursday. Don't send yet — show me the draft." # Update a Notion page: "Add a section called 'Action Items' to my project planning page in Notion. Don't change anything else." # Always review write actions before they're committed. - Explore Bring Your Own MCP for custom tools. If you have internal systems that need Grok integration, MCP is the path.
# MCP integration: # 1. Build (or deploy) an MCP server exposing your tool # 2. Register it in Grok's connector settings # 3. Authenticate the connection # 4. Use it in chat like built-in connectors # Anthropic's MCP docs (the protocol is open): https://modelcontextprotocol.io/ # Reference MCP server implementations on GitHub for many services. # Build custom for proprietary systems. - Set up permissions thoughtfully. Each Connector’s scope determines what Grok can access.
# For each Connector, check: # - What data does Grok read? # - What actions can Grok take? # - Can permissions be scoped down (read-only, specific repos, etc.)? # Principle of least privilege: # - Start with minimum permissions needed # - Expand only when specific workflows require more # - Periodically audit connector permissions - Compare to your current workflow tool. If you already use Claude integrations or Microsoft Copilot, evaluate where Grok Connectors fit.
# Decision questions: # - Where does most of my work happen? # - Which connector quality matters most to my workflow? # - Do I need Grok's specific capabilities (X integration, etc.)? # - Is consolidation valuable, or are multiple AI assistants OK? - Watch for the mobile rollout. Grok Connectors are launching Web-first, with iOS and Android following.
- Provide feedback to xAI. Early users shape the product. Specific feedback on integration quality, missing features, and edge cases helps the team prioritize.
How it compares
Grok Connectors enters a market with established workflow AI options. Comparison:
| Assistant | Workflow integrations | MCP support | Strengths |
|---|---|---|---|
| Grok Connectors | GitHub, Notion, Linear, Google, Microsoft + custom MCP | Yes (BYO MCP) | X-native, enterprise focus emerging, custom MCP |
| Claude | Asana, Atlassian, Slack, GitHub, many MCP servers | Yes (Anthropic’s protocol) | Deepest MCP ecosystem, mature partnerships |
| ChatGPT | GitHub, Notion, Google Workspace, various via plugins/actions | Partial (custom GPTs) | Widest consumer reach, app ecosystem |
| Microsoft Copilot | Deep Microsoft 365 + some third-party | Limited | Microsoft ecosystem depth, enterprise standardization |
| Google Gemini | Deep Google Workspace + some third-party | Limited | Google ecosystem depth, Workspace bundling |
| Cursor / specialized devtools | Code-focused integrations | Yes (MCP server support) | IDE integration depth for coding |
What distinguishes Grok Connectors: the inclusion of both Google and Microsoft ecosystems plus developer tools (GitHub, Linear, Notion) in one assistant, with MCP for everything else. The risks: relatively new to the workflow AI space; integration depth versus Claude or Copilot in their respective strongest areas is unclear yet; depends on xAI’s continuing investment.
What’s next
Signals to watch over the next three months. Mobile rollout: iOS and Android Connector support is essential for the assistant-on-the-go use case. Integration depth: does Grok handle complex workflows in each connected service, or just surface-level operations? Enterprise positioning: how aggressively does xAI go after enterprise customers? Specific case studies will be telling. MCP ecosystem growth: more third-party MCP servers expand Grok’s reach without xAI building each integration.
The longer-term picture. Workflow AI is consolidating around a pattern: AI assistant + open protocol (MCP) + ecosystem of connectors. Every major AI assistant supports this pattern by some name. The competitive differentiation moves to: model intelligence, connector quality, enterprise features (security, governance, audit), and pricing.
For users today, the practical question is which AI assistant to lean into. Most knowledge workers will use at least two (Microsoft Copilot for Office, plus another for general AI). Grok joining the workflow AI race gives users another option. The right choice depends on existing tool stack and specific use cases.
Frequently Asked Questions
Do I need to be on a paid Grok subscription to use Connectors?
Most Connectors require a paid Grok subscription tier. Check current pricing at x.ai. The lowest tier that includes Connectors varies; SuperGrok and Grok for Business both should include Connector access at the time of launch.
How do Grok Connectors handle data privacy?
Connectors authorize via OAuth, so Grok accesses data through your authenticated session — the same permissions you have are what Grok has. xAI’s data handling for Connector-accessed content follows their privacy terms. For sensitive enterprise use, review xAI’s privacy documentation and consider an Enterprise tier with stronger contractual protections.
Can I use Grok Connectors with my company’s internal tools?
Yes, via Bring Your Own MCP. If your internal tools expose an MCP server (or you build one), Grok can integrate. This is the same protocol Anthropic introduced for Claude, so internal MCP servers built for Claude work with Grok too.
How does this compare to Claude’s connectors?
Claude has been in workflow AI longer and has deeper integrations with Asana, Atlassian, and several other partnerships. Grok has broader Microsoft + Google ecosystem support in the initial launch. MCP support exists on both. Choose based on which integrations matter most for your work.
Will Connectors come to mobile?
Yes — iOS and Android Connector support was announced as part of the rollout, following Web. Timeline specifics weren’t disclosed at launch; expect within weeks to months.
Can Grok Connectors make changes I didn’t intend?
Connectors with write permissions can modify data. Grok generally asks for confirmation before destructive actions, but you should review what Grok proposes to do before approving. Don’t grant Connector permissions broader than necessary. For high-stakes operations, prefer read-only mode and manual action.
What if I use multiple AI assistants?
Many users will. Microsoft Copilot for Office work, Claude for deep technical work, Grok for X-integrated work and emerging workflows. The right combination depends on your tools and tasks. Workflow AI isn’t winner-take-all; different assistants excel at different things.