Enterprise
Dragons for
enterprise.
For the platform lead whose auditors, security review, and procurement counterparties need evidence — not screenshots — that the agents running inside the business acted within their authority. Dragons gives you the receipts. The license, deployment topology, and support envelope are negotiated to fit your perimeter.
See also Security model & receipt chain → /security
The audit problem
Your auditors ask for evidence that your agents acted within their authority. Today, the answer is a Slack thread, a screenshot, and a verbal walkthrough.
Autonomous systems inside an enterprise produce side effects: they file tickets, post to ledgers, touch customer records, move money. When the auditor — internal or external — asks which agent did what, under whose authority, when, and against which scope, the burden is on the platform team to reconstruct that history from logs that were never designed to answer the question. The reconstruction is slow, expensive, and unverifiable by a third party.
Dragons inverts that. Every action an agent takes resolves to a signed identity, a scope-bounded lease, and a hash-chained receipt. The audit is a replay, not a meeting.
What enterprise gets
Three states, marked honestly. What ships today in the open-source control plane, what is available under a signed enterprise agreement, and what is on the roadmap.
Ships today
Receipt-pack export for auditors
Every agent action is signed, lease-attributed, and hash-chained in the WorkLedger. Export a date-range pack and an auditor replays goal → action → evidence → outcome against the chain — without live tenant access and without trusting Dragons as a narrator. This is the unit of audit, and it ships in the open-source control plane.
shippedShips today
Tenant-local runtime, no payload exfiltration
Execution and the WorkLedger live inside your tenant boundary. The control plane reasons over hashes, lease scopes, and heartbeat records — not over the contents of the work. Procurement asks "where does the data live"; the answer is "inside your perimeter."
shippedUnder enterprise agreement
On-prem deployment
A dedicated control-plane deployment inside your network, with an enterprise licensing rider on top of the open-source core. Available under a signed agreement — scope, deployment topology, and operational handoff are negotiated per customer. Not a self-serve toggle today.
agreementUnder enterprise agreement
SSO / SAML, custom data residency
Single sign-on (SAML or OIDC) and region-pinned data residency are provisioned per enterprise contract. Both are configured against your identity provider and your region constraints — not a generic checkbox. Talk to us before assuming a specific IdP or region is supported.
agreementRoadmap
SOC2 audit posture
Dragons is not SOC2-certified today. The mechanisms an auditor expects to see — signed identity, scoped authorization, tamper-evident evidence, attributable action history — are in the product because they are the product. A formal Type II audit is on the roadmap; if your procurement requires a current certificate, talk to us and we will tell you exactly where we are.
roadmapUnder enterprise agreement
SLA + dedicated success engineer
A named success engineer, response-time SLA, and an escalation path that does not route through a generic support queue. Coverage hours, severity tiers, and incident-handling expectations are written into the contract — not implied by a pricing page.
agreementThe receipt-pack for auditors
The wedge. Every agent action signed, chain-linked, replayable. Export a date-range pack; the auditor verifies it themselves, against the chain, without live access to your tenant.
A receipt-pack is the unit of evidence Dragons hands to an auditor. It is a self-contained
artifact: identities, leases, the contiguous receipt chain for the window, the liveness
record, and a verifier script. The auditor runs the verifier against the pack, walks
prev_hash by prev_hash, and either
the pack is intact or the tool reports the first index that fails. There is no narrator.
date_range The bounded window the pack covers. Inclusive of the receipt index at each endpoint. manifest_set The set of agent identities active in the window — each by manifest_hash. lease_set Every authorization grant referenced by a receipt in the window, with scope and validity. receipt_chain The contiguous hash-chained receipt sequence. Each entry commits to the previous prev_hash. heartbeat_record The liveness trace for the window — running, degraded, autohealing, quarantined, stopped. verifier_script A standalone replay tool. Walks the chain, recomputes hashes, reports the first index that fails — or confirms the pack is intact. Receipts are append-only. A correction is a new receipt that references the original; retroactive edits are forbidden by the chain itself. The auditor sees the correction in the same way they see the original action — as a signed, attributable event.
License
Licensing details for enterprise deployments are under operator ratification.
The open-source core is governed by its repository license. The licensing terms that apply to a dedicated enterprise deployment — including any rider on top of the open license, redistribution boundaries, and procurement-ready legal summary — are being finalized and will be the canonical statement for procurement and legal review. Contact us for the current canonical language; we will not publish provisional terms on this page that could later contradict a signed agreement.
Talk to us
Open an enterprise conversation.
Tell us the scope: number of agents, regions, identity provider, audit framework, target deployment topology. We will respond with a deployment proposal, the current licensing summary, and an evidence-pack sample against a realistic workload — not a brochure.
enterprise@dragons.dev