Use Orq.ai as the model control layer for LiveKit Agents. Route LLM calls through one OpenAI‑compatible endpoint, capture traces, monitor cost, and manage fallback behavior without rebuilding your LiveKit agent code, media routing, or telephony stack.
What is LiveKit?
LiveKit Agents is a Python and TypeScript framework for real‑time voice agents, handling audio streaming, turn‑taking, telephony (SIP), and routing between your agent code and multiple model providers. LiveKit focuses on voice, media, and session orchestration; Orq.ai centralizes model routing, cost controls, and cross‑provider observability.
Why use Orq with LiveKit
Trace voice agent behavior end to end
Trace LiveKit agent sessions end to end, including audio turns, transcripts, tool calls, LLM responses, TTS output, and errors in one place. You get synchronized visibility into what the caller said, what went into the model, and how the agent responded. All without stitching together logs across multiple services.
Model flexibility without rewiring agents
Test new models, add alternative providers, or assign different model tiers to different LiveKit agent flows from Orq.ai, while keeping your LiveKit Agents code and telephony setup stable. Your agent still “thinks” it’s talking to a single LLM client; Orq handles which underlying model is called.
Evaluate real production calls
Use real session data like audio, transcripts, traces, and logs to build datasets, compare prompt or model changes, and move from subjective tuning to measurable quality checks. Failed or slow calls can be turned into eval inputs before you roll out changes across your voice agent fleet.
Control spend and access centrally
Track token usage and spend per agent, phone number, team, and workflow so you can see which routes drive LLM cost on top of minutes. Add budgets, rate limits, and approved‑model lists at the Orq platform layer instead of enforcing governance separately inside each LiveKit project.
How the integration works
Step 1
Point LiveKit’s LLM plugins at Orq.ai’s router
Configure the LLM plugin or HTTP client your LiveKit agent uses so its model calls go to Orq.ai’s base URL and key instead of calling each provider directly. This gives Orq the context it needs to apply routing rules, capture usage, and enforce fallback behavior behind your voice pipeline.
Step 2
Enable tracing from LiveKit to Orq
Keep LiveKit’s own Agent Observability (audio + transcript + traces + logs) while exporting OpenTelemetry traces from LLM calls to Orq’s OTLP endpoint. LiveKit continues to show voice‑centric insights; Orq records LLM interactions, latency, and cost metrics, so you see both session‑level and model‑level traces.
Step 3
Define routes, fallbacks, and policies in Orq
Create routes for key LiveKit workflows (for example inbound support lines, outbound campaigns, or multimodal agents) and assign them model tiers, fallback chains, and region/data policies. LiveKit sends LLM requests to the configured Orq routes, and Orq applies routing before invoking OpenAI, Gemini, Claude, or other models.
Step 4
Monitor, evaluate, and tune
Once connected, use Orq’s dashboards to watch LLM latency, errors, and cost for LiveKit agents, and run evals or experiments on their traces. You can adjust routes, models, or prompt configurations centrally where supported, while LiveKit continues to manage streaming audio, telephony, and turn‑taking.
Use Cases
Multi‑agent products with real observability
Trace which LiveKit agent, LLM call, tool, and TTS output contributed to a failed or frustrating call, combining LiveKit’s Agent Observability (audio + transcripts + traces) with Orq’s router‑level LLM traces.
Cost‑aware internal tools
Route routine IVR‑style flows and simple QA to lower‑cost models while keeping complex troubleshooting, compliance‑sensitive calls, or VIP lines on your strongest routes configured in Orq, all still running through LiveKit telephony.
Eval‑driven agent improvements
Reuse failed calls, long silences, barge‑in edge cases, and LiveKit trace trees as eval inputs before shipping prompt, tool, or model changes, turning production call telemetry into a continuous improvement loop.
Safer experimentation across providers
Test a new model or provider on a small share of LiveKit traffic (for example 5% of calls), then promote or roll back based on traces and evals, instead of editing each agent definition or LLM plugin config by hand.
FAQ
Do I have to change my LiveKit code to use Orq.ai?
In many cases, you can repoint the LLM client or plugin your LiveKit agent uses (base URL, key, and model identifier) to Orq’s router and, optionally, export traces to Orq, keeping your voice, telephony, and media logic the same. Audio, SIP, and turn‑taking remain in LiveKit.orq+1
Does Orq.ai replace LiveKit’s own runtime or framework?
No. LiveKit remains your platform for voice, telephony, and synchronized audio/transcript observability. Orq.ai sits alongside it as the control plane for models, multi‑provider routing, LLM cost tracking, and additional evaluation. You still design and run voice agents with LiveKit at the core, but you use Orq to see what their LLM calls cost, which models they should call, and how behavior compares across providers under the same call workflow.live
Can I keep using my existing LLM providers with LiveKit if I move to Orq.ai?
Yes. You can bring your existing provider keys into Orq.ai and route LiveKit traffic through them, alongside any new models you add later. That way you centralize access, routing, and tracking without losing the LLM and LiveKit setup you already rely on. Bring production controls to LiveKit apps
