Democratizing AI Customer Engagement
Historically, customer engagement platforms were built for dashboards.
Users learned segmentation systems, campaign builders, analytics interfaces, channel configuration, delivery tooling, and platform-specific workflows. The software was powerful, but operating it required expertise.
AI changes that.
Instead of learning the mechanics of a platform, users can simply dictate the outcome they want and have the software execute on their behalf.
This shift is occurring across software. Developers are using natural language from the command line. AI agents are completing multi-step workflows across tools. Chat interfaces are replacing complex navigation patterns. Operational work is becoming increasingly automated.
We believe customer engagement software will evolve the same way.
Today, we’re introducing two major steps in that direction:
- The OneSignal MCP Server (available now in Open Beta)
- OneSignal AI (available today through an early access program)
Together, they make OneSignal accessible through AI-native workflows — whether inside our platform or through external AI assistants like Claude, Cursor, and ChatGPT.
Introducing the OneSignal MCP Server
The OneSignal MCP Server allows AI assistants like Claude, Cursor, and ChatGPT to securely interact with OneSignal on a user’s behalf.
Rather than switching between dashboards, documentation, APIs, and external tools, users can work directly through their preferred AI environment using natural language.
This transforms AI assistants from passive advisors into active operators inside OneSignal.
For example, an agent can:
- Retrieve campaign analytics
- Inspect segments
- Manage templates
- Look up users and subscriptions
- Export data
- Trigger messaging workflows
More importantly, MCP enables cross-platform orchestration. An external AI agent is no longer limited to operating inside OneSignal alone. It can coordinate workflows across multiple systems simultaneously.
For example:
- Pulling data from a warehouse
- Analyzing results
- Creating a segment in OneSignal
- Generating campaign copy
- Scheduling a send
All from a single conversational workflow.
Our MCP server currently supports 31 tools across the following platform features:
- Messaging (3 tools)
- Users and subscriptions (12 tools)
- Templates (4 tools)
- Segments (4 tools)
- Live Activities (2 tools)
- Exports (2 tools)
- Analytics and utility (4 tools)
You can read more about the MCP and how to set it up in our documentation.
Introducing OneSignal AI
While the MCP server is designed for external AI ecosystems and agentic workflows, OneSignal AI brings AI directly into the OneSignal platform itself.
OneSignal AI is an embedded assistant available within the OneSignal UI that allows customers to interact with the platform conversationally.
Users can ask questions like:
- “How did our push notifications perform last week?”
- “Which segments are growing the fastest?”
- “Create a highly engaged email audience.”
- “Draft a push campaign for users who abandoned checkout.”
OneSignal AI performs actions directly within the platform, reducing the need to manually navigate workflows or learn platform-specific implementation details.
How is the MCP server different from OneSignal AI?
The MCP server is how external agents like Claude, ChatGPT, Cursor can access and use OneSignal. OneSignal AI enables chat based queries and automations directly in the OneSignal platform.
The MCP server is more about making OneSignal dynamically readable and actionable to external agents. This unlocks a powerful use case for the MCP, which is the ability for users to command agents to orchestrate cross-platform tasks across many MCP servers (not limited just to OneSignal’s).
OneSignal AI, on the other hand, offers a chat interface to query and use the product, within the OneSignal UI. It dramatically reduces the learning curve for using OneSignal because you can ask questions and perform actions with natural language abstracting many of the idiosyncrasies of the product. Ultimately, it’s most powerful for automating work and reducing the labor overhead of using our product.
Why we're building this
More and more of our customers are incorporating AI into their daily workflows. We think that trend is only accelerating, and we want OneSignal to work naturally inside those workflows.
Whether that’s becoming fluent in agentic workflows arbitrated from your command line or a preference for interacting with software via natural language, we’re excited to use these breakthrough technologies to improve outcomes from our software and reduce the work our customers need to do to achieve those outcomes.
Next steps
It’s important to note that this are both early stage releases. We're actively developing and iterating based on real usage and feedback. If you're someone who's already building with AI tools day-to-day, we'd love for you to try it out and tell us what works and what doesn't.