Noveum

Noveum is an evaluation platform for production chat, voice, and workflow agents. It is best suited to teams that need to trace and score agent behavior, then validate proposed fixes through human review. Its connected workflow is a strong reason to consider it; the right plan depends on how much data you generate and how long you need to retain it.

Overview

Noveum connects three stages of production agent work: NovaTrace records activity, NovaEval scores traces, and NovaPilot tests candidate fixes before delivering validated changes as pull requests for human review. That makes it a fit for teams that want to move from diagnosing failures to checking proposed corrections in one workflow, rather than assembling those steps separately.

The platform targets teams operating chatbots, voice agents, and multi-agent workflows. Its scope is broad, but runtime guardrails are not yet a settled part of the offering: NovaGuard is marked beta or coming soon.

Key features

Trace and score agent behavior

NovaTrace captures LLM calls, tool invocations, retrieval, and agent steps, with tokens, cost, and latency on spans. NovaEval scores traces with calibrated scorers across 15+ categories, including voice and audio. Together, these tools give teams a way to inspect operational detail and evaluate behavior across more than text-only agents. Tool-call checks, safety evaluations, and regression runs are supported.

Validate fixes through pull requests

NovaPilot backtests candidate fixes on failing calls, re-simulates them end to end, and sends validated fixes as pull requests for human review. That keeps people in the approval loop while connecting failure analysis to a concrete proposed change. Teams looking for automatic runtime protection should not treat this as a substitute for NovaGuard, which remains beta or coming soon.

Integrations and deployment

Native tracing integrations cover LangChain, LangGraph, LiveKit, Pipecat, and CrewAI, with compatibility also listed for OpenAI, Anthropic, and OpenTelemetry. Open-source tracing SDKs support Python and TypeScript. Deployment choices include managed cloud, VPC, BYO ClickHouse, and on-premise, with Kubernetes and Helm options. This breadth can suit teams with different infrastructure requirements, though the choice of deployment should be weighed alongside plan limits and access controls.

The homepage lists SSO/SAML, RBAC, and audit logging; pricing specifies SSO/SAML for Scale and above. GDPR and SOC 2 Type II are in progress. Enterprise deployments include encryption in transit and at rest and fine-grained access controls.

Pricing

Noveum has a free plan, monthly subscriptions, and one-time add-on credit packs. The free plan is useful for a small evaluation or low-volume use, but its 1M spans per month, 2 GB storage, three members, and 30-day retention set firm bounds. Monthly plan credits reset each billing cycle and do not roll over; purchased add-on credits never expire.

PlanPriceAllowance and termsFreefree2.5K credits/mo, 1M spans/mo, 2 GB storage, 3 members, 30-day retentionBoost$35 one-time5,000 add-on credits; credits never expirePro$69/mo10K credits/mo, 20M spans/mo, 20 GB storage, 10 members, 90-day retention; 7-day trialGrowth$99/mo25K credits/mo, 50M spans/mo, 50 GB storage, 15 members, 180-day retention; 7-day trialAccelerate$150 one-time25,000 add-on credits; credits never expireMax$199/mo50K credits/mo, 100M spans/mo, 100 GB storage, 25 members, 1-year retention; 7-day trialScale (credit pack)$500 one-time100,000 add-on credits; credits never expireScale$599/mo200K credits/mo, 500M spans/mo, 500 GB storage, unlimited members, 2-year retentionEnterpriseCustom pricingUnlimited credits, spans, storage, and members; custom retentionPro is the entry point for paid recurring use at $69/mo, with a 7-day trial. Growth doubles Pro’s credit allowance and increases spans, storage, seats, and retention for $99/mo. Max raises those limits again, including a year of retention, while Scale is aimed at much larger teams needing unlimited members and two-year retention. The one-time Boost, Accelerate, and Scale packs add credits without expiration but do not replace the recurring plans’ span, storage, member, or retention allowances.

Growth and above include priority support; Enterprise adds dedicated customer success. Community support is the listed support level below Growth. SSO/SAML begins at Scale, so teams that require it should factor that threshold into plan selection.

Platforms

Noveum supports API, Linux, self-hosted, and web use. The combination of managed cloud, VPC, BYO ClickHouse, on-premise, Kubernetes, and Helm deployment options gives infrastructure teams meaningful choices, while the Python and TypeScript tracing SDKs cover both languages.

Who it's for

Noveum is best for teams running production chat, voice, or multi-agent workflows that need agent-level tracing, evaluation, and a reviewable path from failures to fixes. Its voice and audio evaluation suite and integrations with LiveKit and Pipecat make it relevant to voice-agent teams in particular. Smaller teams can start on Free, but those with higher ingestion, longer retention needs, or more collaborators should compare the subscription limits before committing.

It is a weaker fit for buyers who need generally available runtime guardrails now, or who need SSO/SAML below Scale. Teams with compliance requirements should also account for GDPR and SOC 2 Type II being in progress.

Pros and cons

Strength: connected trace-to-fix workflow. Tracing, calibrated evaluation, backtesting, and pull-request delivery sit in one product, with human review retained for proposed fixes.

Strength: voice and multi-agent coverage. Evaluation includes voice and audio, and native integrations span agent frameworks and voice stacks.

Strength: deployment choice. Managed cloud, VPC, BYO ClickHouse, on-premise, Kubernetes, and Helm options accommodate varied infrastructure preferences.

Trade-off: capacity is plan-bound. Free caps storage at 2 GB and retention at 30 days; paid plans scale those allowances at rising monthly prices.

Trade-off: credits expire on subscriptions. Monthly credits reset without rollover, so variable usage may be less convenient than one-time add-on credits.

Trade-off: some enterprise needs have thresholds or pending status. SSO/SAML is specified for Scale and above, while GDPR and SOC 2 Type II remain in progress and NovaGuard is beta or coming soon.

Alternatives

AI Agent Evaluation Tools is a useful place to compare the category. Choose W&B Weave if its free plan’s 1 GB/mo Weave data ingestion, 5 GB/mo storage, and evaluation, tracing, and scorers better match your starting needs. Future AGI AI Evaluation SDK is a stronger candidate if an Apache 2.0-licensed platform with Docker, Python SDK, and Node SDK self-hosting options is the priority.

Consider Opik if running open-source observability and evaluation locally is important. DeepEval suits readers looking for a free, Apache 2.0-licensed LLM evaluation framework with a local and CI/CD test runner. Promptfoo is worth comparing if its free Community plan’s 10k red-team probes per month and local or self-hosted operation fit the job.

Google Cloud Agent Evaluation may fit teams that prefer pay-as-you-go computation-based metrics, billed at $0.00003 per 1k input characters and $0.00009 per 1k output characters. Maxim AI offers a free Developer plan with up to three seats, one workspace, 10k logs per month, and three-day data retention. Strands Evals is another alternative.

Verdict

Choose Noveum if your production agent team wants tracing, evaluation, and validated fixes delivered for human review, especially across voice and multi-agent systems. Its strongest case is the joined workflow and flexible deployment; look elsewhere if you need generally available runtime guardrails, SSO/SAML on a lower tier, or more predictable capacity without monthly credit resets.