Signed receipt
Every command writes one. Actor, scope, payload hash, decision, version applied. The audit answer to 'who changed this and when?' lives here.
Stays queryable for the life of the tenant
Day Zero
Guided receipt proof in 15 minutes, not 15 days.
KarmanFlow preview setup targets a seeded tenant with policies, a webhook, a first command, and a full audit trail before your team has to trust it. The first 15 minutes are not a setup form; they are a guided operating proof with the evidence path visible.
What actually happens
The first 15 minutes of a KarmanFlow tenant.
Every step below is tied to a smoke path we expect preview cells to prove on a fresh tenant. The minute markers are targets; the artifacts are the evidence we inspect before calling a tenant ready.
Workspace appears
Your tenant, default environment, owner, locations, sales channels, and a small catalog are seeded for you. No checklist, no wizard, no empty grids.
Preview tenantPolicies wired
Approval boundaries, oversell rules, and notification routing are set with sensible defaults. You can change them later; you do not have to today.
Policy in effectWebhook ready
An inventory event webhook is registered against a test sink. Every change you make from this point follows the same audit shape your team will use when production evidence is attached.
Subscription registeredFirst command
You run one inventory adjustment. Real command, real validation, real version check. The same command shape your operators will use on day 180.
Command acceptedProof artifacts land
The proof path targets four replayable artifacts: a signed receipt, an immutable event, an outbound webhook delivery, and a registry entry that AI agents can call when enabled.
Receipt + event + delivery + tool entryConsole reads them
Your operator console shows the change, the audit trail, the webhook delivery state, and the agent-available actions. The workspace is no longer a static demo; it is a preview operating model.
Preview view
What the path proves
Four proof artifacts. Not a slogan.
"Audit trail" is two words on most platforms. On KarmanFlow it is four concrete artifacts the preview path is designed to produce before operators have to trust the system, and that downstream systems (analytics, partners, AI agents) read from without touching the write path.
Immutable event
One per state change. Past-tense, append-only. Downstream systems, analytics, and AI agents replay from this stream without touching the write path.
Streams to your warehouse, BI, and agent context
Webhook delivery
Signed payload pushed to your registered sink with retry, dead-letter, and replay. Same path connectors and partner systems use.
Visible in connector health + delivery history
Agent-callable action
The same command is registered as a typed tool in the agent catalog. A model can propose it; the same approval boundary still applies.
Discovered by Claude, ChatGPT, and any MCP client
What we don't do
Three things you'll notice are missing.
Most evaluation friction comes from setup choices that don't survive contact with production. KarmanFlow's Day-Zero stance is built around what is absent, not what is added.
No empty grids waiting to be filled in.
Most platforms drop you into a dashboard of empty tables and a 14-step checklist. KarmanFlow seeds a preview tenant the moment access opens. Day 0 begins with data, not chores.
No fake data that won't survive contact with production.
The seed shape mirrors the tenant shape intended for production. Same command surface, same approval boundary, same audit trail. The remaining work is evidence and rollout, not a separate marketing build.
No proof you have to build yourself.
The guided proof path targets a receipt, an event, a delivery, and an agent-callable action during the first preview session. Your evidence chain is visible before operators have to trust a black box.
For your engineering team
The same seed for every environment you run.
The bootstrap template is a typed file (`karmanflow.bootstrap.v1`). One command stands up the same tenant shape for your staging cell, CI test runs, partner sandboxes, and contributor preview environments. The Day-Zero you see is the Day-Zero your team gets, not a marketing build that vanishes on signup.
bun run smoke:day0Runs against any tenant you have credentials for. Idempotent; re-running on an already-seeded tenant is a no-op. Same script powers the CI smoke gate on every release.
Get to receipt one
See the 15-minute proof on your operating model.
We'll map the seed to your shape (DTC + marketplaces, single 3PL, multi-warehouse, brand wholesale) so the first receipt is something your team would actually act on.