Speko
Speko provides a provider-neutral routing platform for voice AI workloads, letting teams direct speech-to-text, language model, and text-to-speech requests across multiple third-party providers based on latency, quality, or cost objectives. The platform includes a hosted Router API, an open-source Gateway that runs alongside LiveKit or Pipecat with customer-held credentials, and a separate Agents API for building and running voice agents with telephony. Pricing starts at $0.09 per minute on Speko infrastructure or a 5% routing fee on top of provider rates.
- Artificial Intelligence
- Developer Tools
- AI Agents
- Software Only
Funding
Raised to date
$500KRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
Founders
Founder details are not available yet.
Product
Problem
Voice AI applications often depend on a single speech or language model provider, locking teams into one vendor's pricing, latency, and quality trade-offs. Managing multiple providers manually adds operational overhead and forces developers to build custom failover and routing logic, slowing deployment and increasing maintenance burden.
Solution
Speko provides a provider-neutral data plane for routing speech-to-text, language model, text-to-speech, and real-time voice agent workloads across a catalog of third-party providers. The hosted Router exposes typed operations for model discovery, speech transcription, speech synthesis, and LLM responses, automatically selecting an eligible route by latency, quality, cost, or a balanced objective, while reporting the chosen route in structured metadata. For teams that prefer direct integration, the open-source Speko Gateway runs beside LiveKit or Pipecat, keeps bring-your-own-key credentials in the customer process, and exposes one local streaming protocol across all supported providers. Both surfaces support pinning a specific provider and model when full control is required, and the platform reports routing decisions and latency for each request. The separate hosted Agents API manages agents, sessions, phone calls, and telephony, enabling teams to build, test, and run complete voice agents in one console.
Target Audience
Primary users are application development teams building voice agents, live transcription products, or speech-enabled interfaces that need multi-provider routing flexibility, along with teams that want to keep provider credentials in their own infrastructure.
Features
- Provider-neutral routing across ASR, LLM, and TTS models with latency, quality, cost, or balanced optimization objectives
- Typed operations for model discovery, speech transcription, speech synthesis, and LLM responses with structured route metadata in every response
- Open-source Gateway runtime that runs alongside LiveKit or Pipecat, keeping BYOK provider credentials in the customer process
- Single local streaming protocol exposed across all providers, usable with provider-direct routes, Speko-managed routes, or both
- Support for both streaming and real-time transcription workloads, including high-speed ASR variants like Qwen3-ASR Fast
- Separate hosted Agents API for managing agents, sessions, phone calls, and telephony, with workflow design, simulated caller testing, and per-call scoring
- Benchmark data for speech models, language by language, published at benchmarks.speko.ai