Plans and limits

Two ways to buy

Builder plans (Starter, Growth and Scale) are for the agent you already run. Nothing is ever locked away: any tool family, any data source, any connection, the workspace database, the dashboard, memory and messaging work on every plan. What a plan sets is quantity (agents, schedules, pipelines, knowledge, database size), throughput, and the data credits it includes each month. AI is not included; you top up AI credits in the dashboard.

Managed plans (Solo, Team and Business) are for people running a hosted superagnt as a team. You buy seats and capability together: one price, a monthly credit allotment for each wallet, and limits sized to the tier.

The no-card trial

A builder account starts on a trial that never asks for a card. Its clock does not start at signup: it starts when you actually use the data or the database, so wandering off for a week costs you nothing.

1. 01

   Verify your email

   The first verified sign-in grants $3 of data credits and $1 of AI credit, once, and Growth limits. Nothing is granted before the inbox is proven.

2. 02

   Use it before the clock starts

   Tool discovery, connecting clients, schedules and every other data-free call cost nothing and do not move the clock. Billable data calls draw on the $3.

3. 03

   The 7-day clock starts

   At the 25th billable data call, the first time your agent, the API or the dashboard uses your workspace database, or when the credits run out, whichever comes first. Checking the database's status does not count. The trial ends 7 days later.

4. 04

   If the credits run out first

   Only billable data calls stop: each returns an offer instead of data. The database, schedules, agents, knowledge, the dashboard and messaging keep working until the trial ends.

5. 05

   When the trial ends

   Everything pauses and nothing is deleted. Your agent serves the upgrade link on every call, and your data is kept for 30 days. Choose a plan in that window and everything resumes where it stopped.

One trial per person, counted by email address: letter case, a +tag and, on Gmail, dots and googlemail.com all count as the same address. A disposable address or a second account for the same person starts without one: choose a plan, or ask a teammate to invite you to theirs. Accounts that join through a team invite or claim a workspace set up for them use that organization's plan instead.

What your agent gets at a limit

Every builder limit is a hard cap. When a call would cross one, the call does not run. Over MCP the tool result is an upgrade offer, never a protocol error, because clients bury errors and relay results. Over REST and in the dashboard it is HTTP 402 with the same offer under error.upgrade.

{
  "status": "requires_upgrade",
  "reason": "limit_reached",
  "message": "This organization has reached its schedules cap on the Starter plan (10 of 10); Growth ($149/mo) raises it to 50.",
  "tool": "agnt_schedules_create",
  "limit": {
    "resource": "schedules",
    "used": 10,
    "cap": 10,
    "tier": "builder_starter"
  },
  "current_tier": "builder_starter",
  "upgrade": {
    "tier": "builder_growth",
    "name": "Growth",
    "price_usd_month": 149,
    "included_data_credits_usd": 90,
    "unlocks": {
      "workspaces": 3,
      "agents": 15,
      "schedules": 50,
      "queues": 15,
      "webhookEvents": 150000,
      "dataJobPipelines": 15,
      "dataJobItems": 75000,
      "knowledgeCollections": 15,
      "knowledgeDocuments": 5000,
      "knowledgeStorageBytes": 21474836480,
      "dbTables": 50,
      "dbRows": 250000,
      "dbStorageBytes": 1073741824,
      "inboxes": 3,
      "messagingLines": 3
    }
  },
  "confirm_url": "https://app.superagnt.com/confirm/<token>",
  "human_step": "The call did not run: schedules is at the plan's cap. Everything that already exists keeps working. Nothing was charged. Share the confirm_url with a person on this account: they review Growth there and confirm it. You never pay or change the plan yourself. Retry once they have confirmed."
}

confirm_url is a page where a person reviews the plan and its price and confirms. No agent can move money on any surface. At Scale the offer names Enterprise with a contact link instead.

Pause, never delete

• Moving to a smaller plan is always allowed. Whatever sits over the new caps pauses: the oldest schedules, queues, pipelines, agents, inboxes and messaging lines up to the cap keep running, the rest skip with a plan_limit reason until you upgrade or remove some.
• The workspace database warns at 80% of a cap. Over the table cap, new tables are refused. After 24 hours over the table or row cap it stops taking new rows while reads, updates and deletes keep working, and SQL that only shrinks it (DELETE, TRUNCATE, DROP) still runs, so you can clean up or upgrade. Storage only warns until its headroom is used, then refuses writes that need more space.
• A trial that ends or a plan that lapses pauses everything at once. Nothing is deleted inside the retention window, and choosing a plan lifts the pause straight away.

Upgrading from inside an agent session

Three control tools cover the whole path from "what are we paying for?" to a page a person taps. None of them moves money, on any surface.

Valid targets for a preview or change are builder_starter, builder_growth, builder_scale, solo, team, business. Grandfathered, Toolkit and sales-led plans can be the source of a change but never the destination. A limit offer's confirm_url skips the preview step: it already names the plan.

What is enforced, by plan

On a builder plan, the trial or the Toolkit, every limit in the builder table is counted and enforced wherever the resource is created: MCP, REST, the dashboard and the Agnt Wizard alike.

On a managed plan, four resources are counted and enforced: workspaces, agents, databases and webhook events. Crossing one raises PLAN_LIMIT_REACHED with HTTP 402 and a message naming the resource, the usage and the cap. Seats are checked when someone is invited, with SEAT_LIMIT_REACHED. Everything else is uncapped.

An organization with no plan and no trial keeps the pay-per-call limits it already had, and whatever it already runs keeps running. Creating schedules, queues, data job pipelines or items, or knowledge returns a plan_required offer. A few dashboard paths stay uncapped for it: connecting an inbox, binding an iMessage number or a Slack DM, and switching a paused schedule or pipeline back on. Knowledge captured from email is not capped either. The same messaging lines added through MCP or the Agnt Wizard, and any other messaging line, return plan_required.

Rate limiting is separate from plan limits. On builder plans the rpm figure is one budget per organization, shared by REST and MCP; on managed plans it is per workspace, on REST. See Errors and rate limits for the full error surface, and Usage for where spend is recorded.