Spix

Introduction to Spix

Spix is a communications platform designed specifically for autonomous AI agents. Unlike traditional communication tools built for human users, Spix provides infrastructure for agents to initiate and manage voice calls, SMS messages, and email interactions programmatically. It targets developers building agent frameworks—including those using OpenClaw, Claude Code, and Codex—who require reliable, scriptable, and low-latency communication capabilities without GUI dependencies.

The platform eliminates the operational complexity typically associated with telephony and messaging infrastructure. By abstracting carrier compliance, protocol handling, and media processing, Spix enables agents to interact directly with external systems and people through standardized interfaces. Its architecture prioritizes automation, determinism, and integration into existing developer workflows.

Key Takeaways

• CLI-first design: A single binary executable supports all communication channels; no dashboard or visual builder required
• Flags-first interface: All operations accept non-interactive command-line flags, enabling full scriptability and CI/CD integration
• Built-in AI voice call orchestration: Integrates speech-to-text (STT), text-to-speech (TTS), and LLM coordination with sub-300ms pipeline latency
• Credit-based usage billing: Customers pay only for consumed resources (e.g., call minutes, SMS units, email events)
• Structured machine-readable output: JSON format available for all commands via --json, supporting automated parsing and monitoring
• Compliance-ready infrastructure: Includes 10DLC registration, answering machine detection, DKIM/SPF/DMARC enforcement, and automatic opt-out handling
• MCP-native and REST API support: Designed for interoperability with agent orchestration standards and custom integrations
• Safe development workflow: Supports --dry-run mode and sandbox environments for testing without incurring usage costs

How Spix Works

Spix operates through three primary stages: installation, provisioning, and communication. First, developers install the CLI binary using a curl-based installer script that detects the host environment and places the binary in the system path. Second, they provision communication resources—such as phone numbers—using commands like spix provision number --country US, which allocates a number with specified capabilities (e.g., Voice, SMS) and returns structured metadata.

Third, agents invoke communication actions directly via CLI commands: spixcall make --campaign outbound-sales --to +15551234567 initiates an outbound call, spixsms send --to +15551234567 --body "Agent task complete." queues an SMS, and spixinbox read --limit 1 --json retrieves inbound messages in JSON format. All operations return deterministic, parseable output—including identifiers like call_id or msg_id, cost metrics, and timestamps—enabling agents to log, route, and act on responses autonomously.

Core Benefits and Applications

Spix enables practical real-world agent applications where human interaction is unavoidable. Use cases include appointment scheduling via voice calls with answering machine detection, lead follow-up via compliant A2P SMS campaigns, and customer support resolution through threaded email handling with attachment support. Its infrastructure handles MIME parsing, thread history management, domain-based routing, and multi-provider email failover—tasks that would otherwise interrupt agent cognitive loops.

The platform supports production deployment across tiers: the Sandbox plan allows development and testing with limited resources; the Agent plan supports small-scale deployments with concurrent call limits and usage credits; and the Operator plan provides higher resource quotas, SLA-backed support, and enterprise-grade compliance features. Pricing is usage-based with credit rollover on paid plans, and there is no per-seat licensing model.

All plans include custom domain support for email, delivery receipts for SMS, and call recording/transcription for voice.