Activity and audit logs¶
See what has happened in your tenant: a security audit trail, how the team is adopting Adverity Atlas, and what it is costing. Grouped under Settings > Activity as three views, Audit Logs, Adoption, and Cost Monitoring.
Prerequisites¶
Admin role at the tenant level. Cost Monitoring can require a specific permission, if you cannot see it, ask a tenant admin.
Audit logs¶
Settings > Audit Logs is the security and compliance record: who did what, when, and from where. Every consequential action lands here, sign-ins, member and role changes, data source and setting changes, approval decisions.
Browse and filter: By event type, actor type (User, Service Account, or System), status (Success or Failed), and a time range with Last hour, Last 24 hours, Last 7 days, and Last 30 days presets
Open an entry: For its full detail: the actor, the target, and the before/after where it applies
Export: Download the filtered list as CSV, for compliance reporting or an external SIEM
Entries are immutable, and an actor removed later still appears on their past actions, so the trail stays complete.
Adoption¶
Settings > Adoption shows how your team actually uses Adverity Atlas, so you can tell whether a rollout is landing, and where to help. It is a set of charts over a time range you choose, grouped into four sections:
People & Onboarding: Daily active members, members by workspace role, and the invite funnel
Data Connectivity: Datasources by source type (cumulative), datasource health, and sync job outcomes
Agent & Chat Engagement: Messages per day, and top users by conversations
Knowledge & Curation: Glossary status, classifications approved, and the classification method mix
Cost monitoring¶
Settings > Cost Monitoring tracks what Adverity Atlas spends, so cost is visible before an invoice rather than after. Two things drive spend: LLM usage (tokens) and warehouse usage (the queries Adverity Atlas runs).
Break it down by:
Kind: LLM versus warehouse, the top-level split
Workspace: Which teams are driving cost
Model: Spend per LLM model, useful when comparing providers
Warehouse: Query cost per connected source
Note
Snowflake publishes a query’s compute credits several hours after the query finishes, so a recent
Snowflake query shows without a cost at first. Adverity Atlas revisits those queries once the
credits are available and prices them then, with nothing to configure. This needs the data
source’s role to read SNOWFLAKE.ACCOUNT_USAGE. A query that consumed no billable compute, for
example one served from Snowflake’s result cache, stays unpriced.
Tip
Group by workspace to find where cost concentrates, then by model or warehouse within that to see what is driving it. A single expensive recurring automation usually stands out immediately.
Permissions¶
View |
Who can see it |
|---|---|
Audit Logs |
Tenant Admin |
Adoption |
Tenant Admin |
Cost Monitoring |
Tenant Admin, or holders of the cost-view permission |
What’s next?¶
Members and roles, the access changes the audit log records
AI providers, the provider and models whose spend appears here
Automations, often the largest recurring line in cost monitoring