Speko

Speko: The Voice-AI Gateway

Speko is a voice-AI gateway that provides a single API for the entire voice stack, including Speech-to-Text (STT), Large Language Models (LLMs), and Text-to-Speech (TTS). Backed by Y Combinator, the platform continuously benchmarks providers in real time based on accuracy, latency, and cost, automatically routing each call to the best-performing stack with automatic failover in over 10 languages.

Key Features

• Dynamic Provider Routing: Automatically routes calls to the best STT, LLM, and TTS providers based on continuous benchmarks for accuracy, latency, and cost.
• Unified API: Eliminates the need for multiple provider keys; one Speko API key manages the entire voice stack, with optional bring-your-own-key support.
• Fast Retrieval Engine: Features a voice-native retrieval engine that answers in sub-millisecond time (0.76 ms across 100K documents).
• Multi-Channel Deployment: Deploy voice agents to phone numbers (inbound and outbound calls over SIP), web and mobile drop-in widgets, or via API and webhooks.
• Hosted MCP Server: Offers an OAuth-protected Model Context Protocol (MCP) server requiring no local installation, compatible with tools like Claude Code and Cursor.
• Data Privacy: Users retain full ownership of their data and agents, and audio or transcripts are not used to train models.
• Official SDKs: Provides full programmatic control with official SDKs for TypeScript and Python, plus a LiveKit adapter.

Use Cases

• Lead Qualification: Automate the initial screening and qualification of inbound leads.
• Appointment Booking: Deploy agents to handle scheduling, such as booking appointments for a dental clinic.
• Receptionist: Act as a virtual receptionist to handle inbound inquiries and route information.

Getting Started

Website: https://speko.ai/

Users can describe their desired agent in plain English, test it live in the browser, and deploy it to a phone number, web widget, or API in minutes.