Billing
Wavetable’s pricing is a modest base platform fee plus metered usage. Seats are deliberately a minor dimension: agents and automations do much of the work here, and charging mainly by chair undercounts what the system does for a small team.
See pricing for current tiers and rates.
What is metered
| Meter | Counted when |
|---|---|
| Events ingested | A connector or the ingest API delivers an event into your log |
| Intent runs | A compiled workflow executes |
| AI tokens | The design plane, memory generation, or edge agent uses a model |
| External sends | An approved message actually leaves the system |
| MCP calls | An external agent calls a tool on your endpoint |
What is not metered
- Reading your own data: Canvas views, record pages, and Inbox review are not usage.
- Failed or blocked actions: a send that was never approved is not a send.
- Rebuilds and adjudication: the bookkeeping that keeps projections consistent is Wavetable’s cost, not yours.
Predictability
- Every workflow run records its cost, and unit economics are queryable inside the product itself: “what did the follow-up automation cost last month” is an Ask question.
- Budgets are enforced structurally. Automations carry budget grants; exceeding one stops the run and raises an Inbox escalation instead of quietly growing your bill.
- Overage on paid tiers is billed at published metered rates, monthly, from the same meters you can inspect.