Spectrum

What is Spectrum?

Spectrum is an open-source framework from Photon that connects agent applications to everyday messaging and chat interfaces. Its core purpose is to let you deploy agents so users can interact with them through platforms people already use.

Spectrum supports multiple interfaces (including iMessage, Telegram, WhatsApp, Slack, Discord, and Instagram) and focuses on reliable agent execution and messaging so interactions can feel fast and responsive across different native platform behaviors.

Key Features

• Open-source agent connectivity layer: Use Spectrum to connect agents to common user-facing messaging interfaces without building a separate integration per platform from scratch.
• Low-latency agent execution and messaging: Spectrum is designed to support fast message delivery (including sub-1-second messaging on Photon’s edge network) aimed at responsive agent interactions.
• Adaptive content rendering for native platform constraints: Spectrum adapts message structure, formatting, and interaction patterns to better match what each platform supports.
• Transparent scaling support: Spectrum is positioned to handle growth from early experiments to high-volume production deployments without requiring architectural changes.
• Built-in observability and control: Provides detailed audit logs, message histories, and human-in-the-loop controls for monitoring and review of agent behavior.

How to Use Spectrum

1. Get started with the SDK: Install Spectrum via npm (the page shows npm install spectrum-ts).
2. Configure your agent to target supported interfaces: Use Spectrum to connect your agent to one or more of the supported messaging platforms (e.g., iMessage, Telegram, WhatsApp, Slack, Discord, Instagram).
3. Send platform-aware interactions: Use Spectrum’s message handling to render content appropriately for each platform’s constraints.
4. Monitor and manage behavior: Use Spectrum’s audit logs, message histories, and human review controls to operate the system reliably.

Use Cases

• Consumer-style chat experiences (multi-platform): Deploy an agent so users can message it directly through platforms like iMessage, Telegram, WhatsApp, Slack, Discord, or Instagram—without asking users to adopt a new app.
• Interactive iMessage workflows: Build an iMessage-based interaction that uses structured prompts (for example, sending a poll-style question with options) and have Spectrum adapt how the content is presented within iMessage.
• Human-in-the-loop review for sensitive tasks: When an agent’s output needs oversight, use Spectrum’s human-in-the-loop controls to flag items for review and support operator decisions using audit logs and message history.
• Scale-out from prototype to production: Start with a smaller deployment of an agent connected to one interface, then scale to larger user volumes across supported platforms while keeping the same overall integration approach.
• Operational monitoring for agent behavior: Use built-in observability (audit logs and message histories) to understand what the agent did and how messages were handled over time.

FAQ

• Is Spectrum a hosted service or a software framework? Spectrum is an open-source framework that you can install and integrate (the page shows npm installation), intended to connect agents to messaging interfaces.

• Which platforms does Spectrum support? The page lists iMessage, Telegram, WhatsApp, Slack, Discord, and Instagram, plus “other interfaces people use every day.”

• Does Spectrum change how content is sent per platform? Yes. It includes adaptive content rendering that adjusts message structure, formatting, and interaction patterns to match native platform constraints.

• Does Spectrum provide monitoring tools? Yes. It includes audit logs, message histories, and human-in-the-loop controls.

• How do I start building? The page points to getting started and installing the library via npm install spectrum-ts, then connecting your agent to supported interfaces.

Alternatives

• Messaging-platform-specific bot SDKs / APIs: Build separate integrations for each platform using their native bot or messaging APIs. This can increase platform-specific work compared to Spectrum’s unified connectivity approach.
• General-purpose automation/workflow tools with messaging integrations: Use workflow platforms that connect to messaging apps for chat-like automations. These tools may be better suited for deterministic workflows, while Spectrum is framed around agent connectivity and adaptive messaging.
• Custom message routing services: Implement your own message routing and rendering layer across platforms. This offers maximum control but shifts more engineering effort for scaling, formatting adaptation, and observability to you.
• Agent frameworks with built-in channel connectors (if available): Some agent frameworks include their own messaging integrations. The difference is where the channel-connection layer lives (Spectrum vs. the framework) and how much adaptive rendering and observability you get out of the box.