Plans
One subscription for control.
One meter for work.
The platform fee buys the control plane — tenants, governance, AI autonomy. The meter is ABUs: published per-GB rates for the work the gateway does. No savings-share math anywhere.
Starter
$0
Individual analyst / evaluation · pay-as-you-go
- 1 tenant
- ABUs metered, no minimum
- AI Rule Generation view only
Land
Growth
$499/mo + ABUs
Data team
- 3 tenants
- Governance & audit included
- AI Rule Generation suggest
Adopt
Pro
$2,999/mo + ABUs
Platform team
- 10 tenants
- Routing & policy included
- AI Rule Generation auto-apply
Expand
Enterprise
Custom
Data Platform / Enterprise / OEM / Custom
- Unlimited tenants · VPC / BYOC
- Committed-use ABU discounts
- AI Rule Generation autonomous
- Enterprise POC — 30 days against a target you set, pilot support included
Standardize
Pick your plan in the app. Starter needs no card.
The meter
ABUs — published rates, no savings audit
$0.10/GB
Cache hit — the warehouse never ran
$0.01/GB
Pass-through — 10× cheaper by design
$0.60/GB/mo
Cache storage in your bucket
$0.0001/eval
Rule evaluations — complex policy pays
- Annual and multi-year discounts
- The bill concentrates where the warehouse didn’t run
- AI traffic grows the meter
The common questions
What you’re billed on, and how to get back out
What counts as a cache hit?
One query answered by the gateway instead of by your warehouse. It meters at $0.10 per GB served, against $0.01 per GB for anything passed through — ten times cheaper by design, because a pass-through is work your warehouse still had to do. Both are recorded in your own logs alongside latency and bytes, so the number you are billed on is a number you can count yourself.
What happens if we stop?
You point the hostname back at the warehouse and you are native again in minutes. There is no driver to uninstall, no workbook to rewrite, and no copy of your data on our side to retrieve. Leaving is the same move as arriving.
A miss is forwarded untouched, which sets a floor while you are on it: no query can end up slower or costlier than it already is.
How do we know it worked before we commit?
Point one dashboard at the gateway and read hit rate, latency and bytes from cache on your own traffic — not estimated by us. Most teams have their answer the same afternoon. The Enterprise POC formalizes that into a target you set.
Don’t take the pricing on faith either.
The scan reads your own query history and tells you what repetition is costing you, before you sign up for anything.
Read-only · your token is never stored · nothing to install