Voice platform
VoiceRun
Code-first enterprise voice agent platform with built-in simulation, evaluation and call QA.
- Pricing at a glance
- Pilot $25K for 12 weeks; annual plans from $100K
Consultation request; VoiceRun sells scoped pilots rather than a self-serve trial.
Engineer contact-center agents like software releases
What you get
Build agents in code with simulation, regression tests, evaluations and canary releases, and choose each LLM and speech provider.
What to account for
It is sold through a $25K pilot and six-figure annual plans, needs engineers, and per-minute costs are not published.
What the platform provides
- Engineering workflow
- Agents defined in code and released through a CLI with simulation, regression tests, evaluations and canary or A/B releases.
- Providers and QA
- A choice of LLM, speech and telephony providers, with automated call scoring through VoiceScore.
Connections & compatibility
- Connects with
- OpenAI, Anthropic Claude, Gemini, Deepgram, Cartesia, Twilio, Telnyx
Pricing & terms
Published plans: a $25K, 12-week pilot with up to 250,000 minutes; Commercial at $100K per year (about 200,000 minutes a month); Enterprise at $500K per year (about 1M minutes a month); Transformation at $2M per year. Per-minute rates, and whether LLM and speech costs are included, are not published.
Before committing
In the pilot, ship a deliberately worse prompt and check that regression tests and canary release stop it before callers hear it.
- Which use case would the pilot cover, and how will containment and quality be measured?
- Are LLM, speech and telephony costs inside the plan minutes or billed separately?
- How do regression tests and canary releases catch a bad prompt or model change before callers hear it?
- Which deployment option fits our data residency needs, and who can access recordings?
Controls to confirm
VoiceRun describes a SOC 2 managed cloud, VPC and on-premise or air-gapped options on higher tiers, and customer ownership of data and fine-tuned models. Request the SOC 2 report, recording and transcript retention terms, and the subprocessors behind your chosen LLM and speech vendors. Morak has not audited these claims.
What to consider next
A different approach
Consider Cognigy for an enterprise contact-center platform with a visual builder instead of code-first agents.
A different approach
Investigate LiveKit Agents if your team wants an open-source framework it runs itself.