- 10,000 requests/month
- 25 Clyvel AI credits/month
- 7 days of retention
- Traces, tokens, latency and costs
- No automatic overages
Reduce your AI costs.Get more from every agent.
Identify costly requests, compare models and discover savings opportunities with Clyvel AI while keeping performance in view.

Everything you need to run your AI agents.
Every request, explained.
Inspect requests, trace waterfalls, tokens and latency. Explore quality evaluations, regression datasets, prompt versions and reports.
Learn more
Turn AI spend into clarity.
Compare estimated costs by agent, provider and model. Track budgets and validate potential savings with real usage and quality evidence.
Learn more
Find your next optimization.
Compare optimization opportunities, inspect their evidence and validate quality before applying changes.
Learn more
Meet Clyvel AI. Your operation, explained.
Ask which agent costs most, investigate failures and run evidence-backed audits. Review and approve supported actions where enabled.
Learn more
Stay ahead of every signal.
Investigate regressions with Watchtower, compare traffic against a baseline and review incidents, reliability objectives and alerts.
Learn more
Operate with confidence.
Manage roles, policies, API keys, data retention, audit history and approvals from one workspace.
Learn more
Bring every agent into view.
Connect existing agents through SDK, frameworks, OpenTelemetry or Gateway—on your terms.
Learn moreOperate with confidence.
Manage roles, policies, API keys, data retention, audit history and approvals from one workspace.
Learn moreGET STARTED
Connect your agents. See the bigger picture.
Four ways to connect your existing tools and turn every received request or trace into useful insights.
Choose how to connect
Connect through SDK, a framework, OpenTelemetry or AI Gateway, depending on your application.
Set up your chosen connection
Instrument your agent, connect its exporter or use Gateway to centralize model requests.
See your agents in action
Explore usage, costs, performance and traces to find actionable optimization insights.
Common questions.
Read the documentation →How do I connect my AI agents?
Choose the path that fits your stack: Python/JavaScript instrumentation, frameworks and automation, OpenTelemetry OTLP/HTTP or AI Gateway. Your agent stays in your infrastructure. Gateway is the option for routing model calls through Clyvel.
What can I see in Clyvel?
Bring requests, traces, tokens, P95 latency, incidents, estimated model costs and budgets into one operational view. Explore Watchtower, quality evaluations, regression datasets, prompt versions, reports and exports.
How does Clyvel AI help my team?
Ask about costs, incidents and agents, or run a quick, deep or audit analysis using workspace evidence. Supported actions can be reviewed and approved by your team when enabled; execution is scoped to permitted operations.
How can I find savings opportunities?
Analyze costs by agent, model and provider, identify workloads worth optimizing and review potential model or routing savings. Clyvel helps you evaluate recommendations with operational evidence before applying them.
What's included in a Clyvel plan?
Plans provide monitoring and management capacity, monthly request allowances and Clyvel AI credits according to the selected tier. General setup guidance is free in Clyvel AI credits. Successful quick, deep and audit data lookups cost 1, 2 and 3 credits per turn. Model-provider usage is billed separately. Free has no overages; paid overages require activation and explicit authorization. Approved actions, where enabled, cost 2 to 4 credits.
Can I manage my agents with my team?
Use shared operational views, budgets, roles, policies, approvals and audit history to coordinate decisions. Clyvel helps your team understand activity and keep sensitive actions under human control.