About NovaSynth by Noveum
NovaSynth is a pre-production testing platform for voice agents that simulates realistic callers at scale. It lets teams build custom personas and scenarios-covering accents, interruptions, noise, and network conditions-then runs actual voice calls against their agent to surface failures. The platform scores interactions across 30+ audio dimensions and 100+ transcript dimensions, and includes a component called NovaPilot that recommends fixes which can be backtested before shipping.
Review
Voice agent testing often stops at happy-path scripts and basic guardrails. NovaSynth addresses the gap by making the caller itself a configurable test variable. Teams can simulate the interrupters, second-guessers, and hard-to-understand callers that human testers struggle to recreate consistently, then see exactly where their agent breaks down.
Key Features
- Caller persona and scenario builder: Define callers by accent, mood, behavior, interruptions, background noise, network conditions, and intent. These variables can be mixed and matched to create different test combinations.
- Real voice interaction simulation: Calls run through actual audio and telephony paths, capturing voice-specific issues like latency, audio dropouts, and interruptions that transcript-only testing misses.
- Multi-dimensional scoring: Each call is evaluated across 30+ audio scorers and 100+ transcript scorers, plus configurable business KPIs. The platform surfaces specific failures rather than aggregate pass/fail metrics.
- NovaPilot fix recommendations: When failures are detected, NovaPilot suggests concrete changes-system prompt adjustments, model swaps, and other configuration updates-and lets teams backtest those fixes against existing call traces.
- API-based regression testing: Personas and scenarios can be defined as tests and triggered via API, allowing integration into CI/CD pipelines for automated regression testing as voice agents are updated.
Pricing and Value
NovaSynth launched with a free trial available. The makers have not published detailed pricing tiers or subscription models on the product page. Interested teams can start a trial or book a demo to access the platform, but the long-term pricing structure is not yet defined publicly.
Pros
- Tests voice agents against edge-case caller behaviors-interruptions, mid-call mind changes, background noise-that are hard to stage manually.
- Scores on both audio quality and transcript dimensions, catching problems that text-only evaluation would miss.
- NovaPilot doesn't just flag failures; it recommends specific, backtestable fixes like system prompt changes and model swaps.
- API support allows the same persona-scenario combinations to run as automated regression tests in CI/CD workflows.
- Caller personas and scenarios are modular-any persona can be paired with any scenario as the agent evolves.
Cons
- Pricing information is absent from the launch materials; teams evaluating budget fit will need to contact Noveum directly.
- The platform is new (launched this week), so there's no public track record of long-term reliability or how scoring holds up across diverse production traffic.
- NovaSynth is not well suited for teams that only need basic chatbot testing-while chatbot evaluation exists via an HTTP endpoint, the core design centers on voice interactions and telephony paths.
NovaSynth fits teams that are actively deploying voice agents and need to move beyond scripted, happy-path testing before production. It's particularly relevant for QA engineers and voice agent developers who want to automate edge-case caller simulations and get fix recommendations tied to specific failure patterns. Organizations still in early prototyping or working exclusively with text-based chatbots may find the voice-first feature set more than they need.
Open 'NovaSynth by Noveum' Website
Your membership also unlocks:
