ClovieWelcome
What is Clovie?
Clovie is a governance platform for AI agents. It answers three questions in one place: which agents your organization runs, what they are allowed to do, and what they cost. Your agents keep using the model providers and SDKs they use today; they send their requests through the Clovie gateway, which checks each one against your policies and budgets and records what happened.
Who is this guide for?
This guide is written for people with the Org admin or Admin role. With another role you see fewer pages, and some pages open read-only. If you need more access, ask an Org admin in your organization.
Monitor first, then enforce
Nothing you set up blocks traffic until you decide it should. Every new policy starts by only watching: Clovie records what it would have done, so you can check its effect on real requests before you turn it on.
Connect your agents
How the gateway works
The gateway sits between your agents and the model providers (such as OpenAI, Anthropic or Google). An agent calls the gateway with its own Clovie key instead of calling the provider directly. The gateway identifies the agent, applies your policies and budgets, forwards allowed requests to the provider with your provider key, and records the request, the response and every decision.
Add a provider key and enable models
Go to AI Governance → Models → Provider credentials and select Add credential to store your provider API key. Then, in AI Governance → Models → Catalog, enable the models your agents may use.
Register an agent
Go to AI Governance → Agents → Registry and select Register agent. A name is all you need; you can add the environment, the owner team and a monthly budget. Clovie then shows the agent's key (it starts with cv_) once, so copy it right away. In your agent, change the base URL to the Clovie gateway and use this key as the API key. Nothing else in your code changes.
Check that it works
Until the agent sends its first request, its status is Waiting for first call. After that it changes to Active. Open the agent to see its usage and runs. The Kill switch on the agent's page stops all of its requests at once, without redeploying anything.
Find agents you didn't register
AI Governance → Agents → Shadow agents lists callers that tried to use the gateway without being registered, with how often and when. Register one in a click. Discovery looks for AI usage in the systems you connect, so you can bring it under governance too.
Policies and approvals
Monitoring and enforcing
A policy is a rule that decides whether a request goes through, is blocked, or waits for a person to approve it. A new policy is always Monitoring: it records what it would have decided and never changes a request. When you trust it, switch it to Enforcing. Only Policy admins, Admins and Org admins can do that, and Clovie first shows you how the policy would have treated your recent traffic.
Start from the Policy hub
AI Governance → Policies → Policy hub has ready-made packs, for example for data privacy, spending limits and common regulations. Install a pack and its policies appear in My policies, monitoring.
Try a policy before you rely on it
From My policies, open the Policy simulator. Pick an agent, a model and an environment and see whether the request would be allowed, denied or would need approval.
Approvals
When a policy asks for approval, the request waits in AI Governance → Policies → Approvals until someone approves or denies it, or the request expires. New requests appear on their own, with the time left to decide.
Routing
AI Governance → Policies → Routing can send simple requests to smaller, cheaper models and keep complex ones on the models you choose. Compare the answers first before you switch it on for real traffic.
Costs and usage
Where the money goes
AI Governance → Costs shows what your agents spend, by model, team and agent, over the period you choose. Cost explorer lets you slice spend any way you need, and Token usage shows how many tokens each source, model, team and person uses.
Budgets and alerts
In AI Governance → Costs → Budgets, set a budget for an agent, a team or your whole organization. You choose whether it only alerts you or also stops requests once it is used up. Anomalies flags unusual spikes in spend.
Chargeback and negotiated prices
Allocation and Statements split spend by team for internal billing, ready to export for your finance team. If you pay providers less than list price, add your rates in Negotiated prices so every cost is shown at what you actually pay.
Security and risk
Protect sensitive data
In Identity and access → AI settings → Data and privacy, choose what sensitive data (such as email addresses, keys or card numbers) is masked before a request reaches a model provider, and how long request data is kept.
Incidents
AI Governance → Activity → Incidents tracks AI incidents from detection to resolution, with severity, timeline, cause and follow-up actions. Clovie opens one for you when a kill switch fires or a policy catches a violation, and you can add your own.
Risk register
AI Governance → Compliance → Risk register keeps a structured list of your AI risks: what could happen, the controls in place, who owns it and by when, with links to related incidents and policies.
Send events to your security tools
Identity and access → AI settings → Export to SIEM streams policy decisions, kill switches, approvals and other events to your SIEM or any webhook.
Compliance and audit
Audit log
AI Governance → Activity → Audit log has one row for every decision the gateway made: the time, the agent, the policy, the decision and the reason. A request that a monitoring policy would have blocked shows as Would deny. Open a row to see the full run in Agent runs, with the request, the response and every policy that looked at it.
Evidence for auditors
AI Governance → Compliance → Compliance packs creates signed evidence packs for frameworks such as SOC 2, NIST AI RMF and the EU AI Act, covering the period you choose. Each pack includes a summary, the decisions and approvals, and the policies that were in force.
Identity and access
Overview
Identity and access → Overview shows your organization at a glance: people, invitations, groups, projects and connections.
Users and groups
In Identity and access → Users and groups, invite people, choose their role, and suspend or remove them. Groups put people into teams and give a whole team access to projects at once.
Connections
Identity and access → Connections is where you connect your cloud accounts and other tools, such as Slack or your ticketing system, so Clovie can discover AI usage and send you notifications. Cloud connections use short-lived, keyless access that you approve in your own cloud.
Single sign-on
Identity and access → Single sign-on connects your identity provider (SAML or OIDC, for example Okta or Microsoft Entra ID) so people sign in with their work account.
AI settings and organization
Identity and access → AI settings holds organization-wide AI settings: models and routing, data and privacy, limits and budget, and connectors. Identity and access → Organization holds your organization's name and security settings; only Org admins can change it.
Settings and roles
Your settings
Manage → Settings has your preferences (theme, time zone), your security settings, such as multi-factor authentication, and your notifications.
Projects
Manage → Projects is where you create projects and choose who works in them. Each project has its own members and connections. Use the project selector at the top to switch between them.
Roles
Org admins have full control of the organization, including billing, single sign-on and security settings. Admins manage people, projects, integrations and settings. Policy admins can do everything an editor can, and also switch policies to enforcing, manage approvals and export evidence. Editors register agents and create policies that monitor. Agent operators register and edit agents and view their runs. Members see data and take everyday actions but can't change configuration. Viewers have read-only access.
Audit history
Manage → Audit log records what happened in your organization, and its Security events tab lists sign-ins, role changes and other security events, ready to export.
Next steps
Connect your first agent
Add your provider key, register an agent in AI Governance → Agents → Registry and point it at the gateway. Your first requests show up within minutes.
Install a policy pack
Pick a pack in AI Governance → Policies → Policy hub. It starts monitoring, so you can see what it would do before you enforce it.
Set a budget
Add a budget in AI Governance → Costs → Budgets so you hear about spend before it becomes a surprise.
Invite your team
Invite the people who own your agents and policies in Identity and access → Users and groups.
Need help?
Ask the assistant on any page, read the FAQ, or get in touch with our team.