We’re building Ever Works in public. Some things may be incomplete, missing, or broken while we continue improving the platform.We’re building Ever Works in public — expect a few rough edges.
Feature · Usage & cost
An AI workforce consumes model calls, credits and compute, and those are the resources Ever Works tracks. Every run is metered and written to a ledger you can read, budgets are checked before an AI call runs rather than after the bill arrives, and the Fleet view shows the machines your Agents are working on. Organizations and Teams record who owns which Works and Agents, so every number has an owner.
Usage is recorded per run, from the run that caused it — not smeared across a monthly total.
Agent runs, pipelines and AI calls debit credits; your plan’s monthly allowance and a daily free allowance add them. Your balance is simply the sum of a ledger you can read line by line under Settings → Billing.
Settings → Usage & Credits takes one period — a calendar month or a rolling 7 or 30 days — and breaks it down per day, by model, by Agent and by Work, with Export CSV for every usage event in it.
Each run in Sessions shows its duration, tokens, changed files and spend, and usage appears while the run is still going — so an expensive run is obvious before it finishes.
Point the platform at your own model keys and those runs cost no credits at all.
Caps sit at several levels and add up: whichever applicable cap is strictest wins.
Set a monthly spend cap under Settings → Account → Usage & Budget, with an alert threshold. Once spend reaches it, the next AI call is refused before it runs — or, if you allow overage, you are alerted and the call goes ahead.
A Mission can cap the spend of a single tick, how many Works a tick may spawn and how many items each produces, and can require approval above a forecast. Missions, Ideas and Works each show a Budget card with current spend, the cap that applies and the share used.
Every Agent can carry its own spend cap — hourly, daily, weekly or monthly — checked before each AI call. An Agent that keeps failing pauses itself rather than burning spend.
Crossing a budget threshold files a notice in your Inbox with the spend and the cap, and an account past its cap appears under Needs attention on the dashboard.
Compute is a resource too, and so is knowing which team a cost belongs to.
Settings → Fleet lists every enrolled machine with what its worker is doing — idle, working, paused, throttled or quarantined — and its recent job history. Give a node CPU and memory ceilings and it stops taking new work while it is above them.
Nodes of a Kubernetes cluster you configured appear in the same Fleet view, so self-hosted capacity is not a separate dashboard.
A team groups Agents and people and lists the Works, Agents, Missions, Ideas and Tasks it owns, so “which team runs this site?” has an answer you can look up.
Missions, Ideas, Works, Agents and Teams belong to an Organization you switch between from the sidebar, and per-Work roles decide who may edit what.
Every figure above is counted from real usage records; nothing is estimated. A few boundaries are worth knowing, and plan allowances and prices live on the pricing page rather than here.