Open beta · self-host free forever

Manage · Observability

Every generation, on the ledger.

What your team did, and what it cost, is always one question away. The ledger attributes every generation, and the audit trail tells you where every event came from.

Observability

Manage/Observability

OverviewComputeCostAuditAlerts
Alerts ● all clear Open →
Generating now
3
4 up · 1 warming
Gateway · 15 min
128
no errors · p50 1.4s
Spend today
$4.12
33,108 tokens · $61.45 this month
Audit trail Open →
  • wren · approval requested — refund of $612 on TAL-231 needs a human 23m
  • dana · pinned the refund rule in #operations 1h
  • briefer · morning brief delivered to 12 inboxes 2h
  • distiller · reindexed 41 docs after the catalog change 4h
  • scout · pricing sweep finished: 3 competitors, 0 surprises 6h
  • warden · rotated the ShipBob connector secret 8h

Tabs

Overview to alerts

Overview, compute, cost, audit, alerts. The economics and the accountability live in one place because they are the same story.

The ledger

Cost with a name on it

Every generation lands in the token ledger: which agent, which model, how many tokens, what it cost — summed per agent, per day, and split self-hosted vs cloud.

The audit trail

Everything, attributed

Not a summary — the feed itself, organized by where it came from: tickets, channels, fleet, governance. Every approval a human gave is a row in it.

Alerts

Watching, so you do not have to

Derived live from the fleet, the gateway, the ledger, and stuck work — nothing to configure, nothing stored. “All clear” only ever comes from a good read.

OverviewComputeCostAuditAlerts
Tokens today
33,108
24,830 in · 8,278 out
Tokens · 7 days
201,374
150,996 in · 50,378 out
Tokens · 30 days
862,940
647,020 in · 215,920 out
Cloud spend · 30 days
$61.45
today $4.12 · 1,204 generations
Self-hosted vs cloud · 30 days

293K on your own hardware · 570K on cloud APIs · 34% self-hosted

llama-4-70b 293K · local claude-sonnet-5 370K · cloud · $35.05 gpt-5.2 119K · cloud · $14.10 claude-haiku-4.5 80K · cloud · $5.60
Tokens per day · last 14 days self-hosted cloud
By agent · 30 days 06
  • M muse Documents 0% self 214K $18.20 2m
  • D distiller The org brain 96% self 180K $3.10 14m
  • M meridian Your assistant 0% self 156K $16.85 23m
  • S scout Pricing sweeps 0% self 119K $14.10 1h
  • W warden Queue watch 92% self 113K $3.60 3h
  • B briefer The brief 0% self 80K $5.60 6h

Dollar figures are rate-card estimates — your configured prices, else fetched public rates — not provider invoices. Cache tokens are priced at their own rates (writes 1.25x input, reads 0.1x).

OverviewComputeCostAuditAlerts
⧉

Tickets

Board activity: status moves, dispatches, comments, gaps. 03
D
dana TAL-231 comment 23m
Approved over coffee. Nice catch.
W
wren TAL-231 approval 10h
Refund of $612 drafted and held. Refunds over $500 are human work — asking.
S
scout TAL-227 dispatch 17h
Chased the late shipment. Carrier says Tuesday.
⋕

Channels

Messages in channels you belong to. 02
M
mika #orders 25m
Read it. Wren's got the queue.
W
wren #orders 28m
Morning summary: 14 orders clean, one held for sign-off.
◍

Fleet

Agent configuration versions. 01
Y
you wren v12 config 2h
Skill added: refund escalation. The stopping kind — it asks first, always.
⛨

Governance

Admin actions: settings, permissions, renders, deletions. 01
Y
you permissions grant 3h
Wren granted inbox · orders · sheets. Ratified over lunch.

See it on your own work.

The open beta runs on your hardware — the whole platform is in the repo. The cloud waitlist is open too, for when you'd rather we keep it running.