Skip to main content

Usage & Limits

The Usage & Limits page shows how much of everything your company has consumed — AI tasks, executions, credits, storage, phone minutes — and how much headroom each limit leaves. It is the first place to look when something is blocked or you want to understand what drives your bill.

What it's for​

Almost everything in Autessa is metered: how many resources you can create, how often agents and tools run, how many LLM credits copilots consume, how much you store. This page collects all of those meters in one place, grouped into collapsible category cards with progress bars per metric. It refreshes on demand and shows fixed windows (this minute, today, this month) rather than custom date ranges.

From the builder sidebar, open the Company section and click Usage & Limits. The page header shows when the data was last fetched and a Refresh button. For a per-task operational log (which agent ran, when, with what flags), see AI Tasks under Agentic Workflows in the sidebar.

Key concepts​

What counts as a task​

AI tasks are the primary billing unit:

  • A conversation with an agent, or a standalone LLM or tool run, counts as one task.
  • A multi-agent (manager) conversation counts as its configured number of sub-agents — the sub-agents do not bill separately on top.
  • A task with evaluation enabled adds one extra task for the evaluation pass.
  • Voice minutes and semantic search (RAG) calls are metered surcharges attached to a task, not separate task counts. Voice time is rounded up to whole minutes.

Each task is also tagged with a coverage state: covered by your free monthly sandbox allowance, covered by prepaid credit, or billable at month end. Sandbox and credit-covered tasks cost nothing extra.

The category cards​

  • AI Tasks — total tasks this month, billable (overage) tasks, voice tasks and minutes, semantic search calls, and evaluated tasks, alongside cost-so-far and projection tiles. Shown when billing is enabled for your company.
  • Resource Limits — how many Tools, Agents, Multi-Agents, Custom Apps, and Database Tables you have created against your company's caps.
  • Execution Limits — agent, tool, and multi-agent executions, each with a per-minute rate limit and a monthly quota.
  • LLM Credits — daily, weekly, and monthly credit meters for AI features, computed from input and output tokens with the weights shown on the card.
  • Phone Calls — phone numbers in use and call minutes this month.
  • Storage — storage units consumed, with the per-column-type cost grid (regular, vectorizable, file, and autocomputed columns are weighted differently).
  • Semantic Search — search requests this minute, today, and this month.
  • Browser & Scraping — web search, basic scraping, and JavaScript scraping usage, plus the per-minute rate limit.
  • Vibecode (AI Coding) — daily and weekly credit meters for the App Builder AutoBot, when enabled.
Rate-limit windows reset on their own schedule — per-minute meters clear within a minute, monthly meters on the first of the month. If a run is rejected for a limit, check the matching category here before assuming an error.

How to check why something is blocked​

  1. Open Company → Usage & Limits and click Refresh.
  2. Find the category matching the failing action — creating a resource maps to Resource Limits, a rejected run maps to Execution Limits or LLM Credits, a copilot refusal maps to LLM Credits or Vibecode.
  3. Look for meters at or near 100%. Per-minute limits clear on their own; monthly caps and resource caps need a higher limit or cleanup.
  4. If the AI Tasks card shows a budget or hard-limit badge, your own spending caps are the blocker — adjust them on the Billing page.

How to audit individual AI tasks​

  1. In the sidebar, open Agentic Workflows → AI Tasks.
  2. Pick a month from the period selector (the last six months are available).
  3. Scan the table — each row shows when the task ran, its type, model, who ran it, status, and flags for RAG, evaluation, and voice.
  4. Click a row for the full task detail.