Tiers & limits
What the free, pro and enterprise tiers offer — and the practical limits you should know about.
Status: planning. Tier-based gating is not enforced in the platform today. Every project is created as
freeand no feature gate currently checks the tier. The tiers below are the intended plan structure. Before publishing this page externally, confirm with engineering and product which limits are actually enforced.
The three tiers
| Tier | For | Approximate scope |
|---|---|---|
| Free | Trials, individual use, internal experiments | One workspace, small data, basic reasoning |
| Pro | Single-team production use | Multiple projects, higher data volumes, all specialist analysts, scheduled trackers |
| Enterprise | Org-wide deployment | Unlimited projects, SSO, audit log retention controls, per-tenant configurations, priority support |
The platform's database tracks the tier on each project. The UI displays a tier badge. Quota and feature gates are roadmap work — you can use any feature regardless of tier today.
Practical limits (what the code actually enforces)
These limits exist independent of tier:
| Limit | Value |
|---|---|
| Uploaded skill archive size | 5 MB |
| Audit log export per request | 50,000 rows |
| DuckDB query result rows | 10,000 (truncated server-side) |
| API key expiry | 1 to 3,650 days, or no expiry |
| Session lifetime | 30 days |
| Audit log pagination | 50 events per page |
| File upload practical limit | ~500 MB per file (above this, use BigQuery) |
Limits that are NOT enforced today (despite tier appearing in the UI)
- Number of projects per workspace
- Number of sources per project
- Number of trackers running concurrently
- Reasoning runs per day
- API calls per minute
- Storage volume
These will be added as part of the tier rollout. Today they're effectively unlimited (subject to the operator's GCP quotas).
How to know which tier you're on
Open Settings → Project. The tier badge appears at the top of the project's settings card.
Upgrading
Today: contact your admin. There's no in-app upgrade flow. Pro and Enterprise are arranged with the operator (Kapiva) directly.
Operator-level limits
Your operator has set quotas on the underlying infrastructure: BigQuery slot reservations, Cloud SQL connection pools, GCS request rate. If you hit one of these, you'll see a clear error in the UI; surface it to your admin who can adjust the operator-level quota.
Tips
- Don't over-engineer for a tier you're not on. Build the project. If you hit a limit, ask.
- Watch the dev-side limits. The 10,000-row DuckDB cap is the one users notice — if your data exceeds that, route the query through BigQuery instead.
- Talk to your admin about retention if you're regulated. Audit log retention, source-version retention, and reasoning-run retention are all operator-configurable.