Billing, seats & onboarding

What this page covers

Every organization in Elliptic has an edition (its plan), a seat count (who is billable and who is free), a pool of AI credits, and, when it is new, a get-started checklist. This page is the admin-facing reference for all four. It explains how a seat is counted, how a plan maps to a seat limit and a list of features, how the per-seat AI credit pool is sized, and how the onboarding checklist is computed from your live workspace. Everything here matches what the product actually does, so you can read your numbers with confidence.

You will find the controls under Settings → Members. The Plan & licensing card and the Seats card sit at the top of that page, above the invite form and the member list. The get-started checklist appears on your organization home until you finish or dismiss it.

Seats

A seat is a membership in your organization. Elliptic splits seats into billable and free based on the member's role, counts your agents on their own line, and shows the full breakdown by role. Nothing here requires a payment method to read. It is accounting, so you always know where you stand against your plan's limit.

Billable seats vs free seats

Your organization has four roles: owner, admin, member, and guest. The first three use a paid seat. Guests are free. So the seat math is simple:

Billable seats is the sum of your owners, admins, and members. Free (guests) is the count of guests. Total members is billable plus free. Agents (non-billable) is the number of your active AI users. Only billable seats are measured against your plan's seat_limit, so adding guests or agents never pushes you over the line.

This is also why guest is the right role for an outside collaborator. A guest joins specific projects with a limited project role (viewer or commenter), takes a free seat, and is not left holding task assignments. When you demote a member to guest, Elliptic caps their project access to viewer or commenter and clears any task assignments they held. Promoting someone from guest to member is the moment they start consuming a billable seat.

The breakdown by role

The Seats card shows three headline numbers, billable seats, free (guests), and agents (non-billable), followed by a row of badges with the exact count per role. The badges appear in a fixed order, owner, admin, member, then guest, and a role only shows when it has at least one member. So a small team might read owner · 1, member · 4, while a larger org adds admin · 2 and guest · 3.

Editions and plans

An edition is the plan your organization is on. Every org starts on Free. The plan is a single setting that records three things: a billable seat limit, a per-seat AI credit figure shown on the card, and a list of features for that tier. Elliptic ships four plans.

Each step up raises the seat limit and the per-seat AI credit figure on the card, and the feature list grows. The Free plan lists the core surfaces, projects, tasks, notes, and search. Pro adds dashboards, automations, and cycles. Business adds the governance features, and Enterprise carries *, which the card renders as a single all features badge.

The feature badges

The Plan & licensing card lists your edition's features as small badges, with underscores rendered as spaces, so audit_log reads as audit log. The badges to know about:

• `sso` (single sign-on), `audit_log` (a record of sensitive RBAC actions), and `custom_roles` (org-defined roles with a granular permission set) appear on Business and above.
• `dashboards`, `automations`, and `cycles` appear on Pro and above.
• Enterprise carries *, which the card renders as a single all features badge.

The over-limit flag

Your edition compares your billable seats against your plan's seat limit. When billable seats exceed the limit, the card shows a warning: you are over your seat limit, upgrade to add more billable members. When you are within the limit, it instead shows how many seats remain on your current plan, for example 2 seats remaining on Free.

Changing the plan

Changing the plan is an admin action. The Plan & licensing card has a plan dropdown listing all four editions, Free, Pro, Business, Enterprise. Pick a plan and the change applies immediately: the seat limit, the per-seat AI credit figure on the card, and the feature badges all update to the new edition. The API restricts the change to admins and owners, so when a member or guest tries to switch plans the request is rejected.

1. 1
   Open Settings → Members

   The Plan & licensing card is the first card on the page, above the Seats card and the invite form.
2. 2
   Read your current edition

   The card shows your plan, your billable seats against your seat limit, any non-billable agents, and the AI credits per seat for the tier. Below that is either seats remaining or the over-limit warning.
3. 3
   Pick a new plan

   Open the plan dropdown and choose a higher or lower edition. Owners, admins, and members carry over as the same billable seats under the new limit.
4. 4
   See the card refresh

   The feature badges, seat limit, and the AI credits figure all update to the chosen plan right away.

AI credits

Elliptic meters AI work as AI credits. One AI action, any single call Elliptic makes to your model, summarizing a meeting, answering a brain question, an agent acting, uses one credit. The credit pool is sized by your billable seats, so as your team grows, your monthly allowance grows with it. You read it on the AI credits card under Settings → AI.

The pool is 500 credits per billable seat per month, and your monthly limit is 500 multiplied by your billable seats (with a floor of one seat, so a brand-new solo org still gets a full pool). Usage is counted per calendar month and resets on the first of each month.

Usage is tracked against the same AI run log that records everything your models do. The AI credits card reports used, limit, and remaining for the current month, with a progress bar that turns red past 90 percent, so an admin can see how much of the pool is left before month-end. Because the count is just AI runs, the credit pool and your run history always agree.

The onboarding checklist

A new workspace shows a Get started checklist on the organization home. It is not a static tutorial. Every item is computed from your live workspace data, so a step ticks itself the moment the real thing exists. This makes it a reliable map of how far an org has come, which is exactly what you want as the admin standing up a new team.

The steps

There are six steps, each backed by a live count in your organization:

Because each step reads a real count, there is no separate checkbox to tick. Create a project and the first step is done. Invite a second person and the teammate step is done. The checklist reports how many of the six are complete, the total, and whether the whole list is finished.

How admins use it

The checklist is the standard path for getting a workspace operational, and each step links straight to where you do it: the project, task, and cycle steps open Projects, the teammate step opens Settings → Members, the note step opens your workspace browser, and the AI step opens Settings → AI. Work down the list and you end up with a project, a first task, a second teammate, a planned cycle, a written note, and a connected model key, which is everything an org needs to actually run.

• It shows progress at a glance. A progress bar and an _N of 6 done_ label sit at the top, so you can see how complete the setup is without reading every line.
• Done steps are crossed off. Each finished step gets a check and a strikethrough, while open steps keep a link to go act on them.
• Connecting AI is on the list. Because every AI feature runs on your own key, the last step nudges you to add your provider key so agents, summaries, and brain answers can run.
• It is dismissable, and it gets out of the way. You can dismiss the checklist with the close button, and it also disappears on its own once all six steps are complete.