Pricing
Free to see, read-only. Optimization is 20% of verified savings — $0 if we save you nothing.
20% of verified savings
Nothing is billed unless savings are confirmed against your own Databricks billing data. You keep 80% of every verified dollar.
Custom
For teams that need an agreed rate, an annual agreement and help through a security review.
| What you get | Standard20% of verified savings | EnterpriseCustom |
|---|---|---|
| Fee | 20% of verified savings | Custom rate, in your signed order form |
| Billed only on savings confirmed from your own Databricks billing data | Included | Included |
| Read-only onboarding | Included | Included |
| Every action gated by policy and a single-use signed token, and recorded in a hash-chained ledger | Included | Included |
| Annual agreement | Not included | Included |
| Security review support: scopes documentation and an architecture walkthrough | Not included | Included |
| Change-control-ready policy recommendations | Not included | Included |
The verified-savings guarantee: if we don’t save you money, you don’t pay us. We bill only savings verified from your own billing data, and if you disagree with a bill, we go through it with you.
FAQ
We compare what the resource cost in your own billing data in the 24 hours before an action with the 24 hours after it, at list price, and bill only savings verified that way. It’s a before/after comparison, not a controlled experiment, and every result says so. If you disagree with a bill, email us within 15 days and we’ll go through it with you. See the methodology.
Before any hibernate, the safety gate looks for work on the cluster: a running job, queued job, notebook or session refuses it, and if Digital Tap cannot see every job it does not guess — the hibernate needs a person’s approval. Clusters on your protected list are never touched. The idle-detection agent leaves alone a cluster launched in the last 10 minutes, and proposes a hibernate only once a cluster has been idle past its threshold, 30 minutes by default. When Digital Tap hibernates a cluster on its own and it is resumed through Digital Tap’s clusters API within 15 minutes, that counts as a mistake: the cluster’s idle threshold rises, up to four times the base, and falls back as hibernates go cleanly. And by default you run in copilot mode: nothing runs until an owner or admin approves it on the Approvals page.
Yes. It’s read-only by architecture — it cannot make changes — and it produces your actual waste number. We do it free because once you see the number, turning on optimization is an easy call. That’s the whole business model.
A scoped credential for your Databricks workspace — Digital Tap connects to Databricks today; other platforms will follow as each meets our integration standard. For the assessment, read-only permissions are enough. Credentials are encrypted at rest, and optimization permissions are only requested when you activate it.
At $50K/mo cloud data spend, the 29% industry estimate of cloud waste (Flexera 2026) would mean ~$14.5K/mo of savings — if that much is verified, you keep ~$11.6K and our fee is ~$2.9K. Your real number comes from your assessment, not an estimate. If verified savings are zero, the fee is zero.
Connect read-only in minutes. Get a verified waste map in days. Turn on governed action only when the evidence is worth it.
Get the read-only assessment