> ## Documentation Index
> Fetch the complete documentation index at: https://hebno.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> Hebno is a B2B AI control platform for growing teams (30-250 people). Use British English (organisation, not organization).
> Describe only what Hebno supports today. If unsure, say so.
> Spend limits use the organisation display currency in the app.
> Start with /index and /getting-started/quickstart for Hebno orientation.

# Spend and limits

> Hard and soft limits, scopes, billing periods, hierarchy rules, and what happens when a cap is reached.

Hebno tracks AI spend and enforces **limits** you set at organisation, team, member, or project level.

## Limits vs subscription

| Concept                  | What it is                                           |
| ------------------------ | ---------------------------------------------------- |
| **Subscription (seats)** | What you pay Hebno monthly per active member         |
| **Limits (AI spend)**    | Internal caps on model usage within a billing period |

Seat billing and AI spend limits are separate. Limits control how much model usage your organisation can incur before requests are blocked or alerts fire.

## Hard vs soft

| Mode     | Behaviour                                                                              |
| -------- | -------------------------------------------------------------------------------------- |
| **Hard** | Blocks new chat requests when spend reaches the limit for the current period           |
| **Soft** | Sends alerts at a threshold and when the limit is reached; does **not** block requests |

Only **hard** limits stop new chat requests. Soft limits are for visibility and operator notification.

When a hard limit blocks a request, the member sees an error and no model call is made.

## Scopes

Set a limit on one of four scopes:

| Scope            | Applies to                             |
| ---------------- | -------------------------------------- |
| **Organisation** | Total spend across the whole workspace |
| **Team**         | Spend attributed to one team           |
| **Member**       | Spend attributed to one person         |
| **Project**      | Spend attributed to one client project |

A single request is checked against every **hard** limit that matches the conversation's organisation, member, team, and project tags.

## Billing period

Limits reset on **UTC calendar months**. The period start is the first day of the month at 00:00 UTC.

Reports and limit checks use the same period boundary.

## Hierarchy rules

When admins create limits, Hebno validates that child budgets do not exceed parents:

* Sum of **team** limits cannot exceed the **organisation** limit
* Sum of **member** limits within a team cannot exceed that **team** limit
* Members not on a team: personal limits cannot exceed the organisation limit
* **Project** and **organisation** scopes skip hierarchy checks against siblings

This prevents configuring caps that contradict each other.

## Alerts

For each limit you can set an **alert threshold** (percentage of the cap). When spend crosses the threshold, admins receive notifications.

At 100% of a soft limit, alerts fire again. At 100% of a hard limit, subsequent requests are blocked.

## Reserve

When viewing limits in admin, **Reserve** is the unallocated budget remaining at organisation or team level after assigned child limits. Use it to see headroom before assigning more caps.

## Example

An agency sets:

* Organisation hard limit: €2,000/month
* Team hard limit (Client A): €800/month
* Member soft limit (freelancer): €100/month with alert at 80%

Client A work tagged to that team counts toward both team and organisation totals. The freelancer gets warnings at €80 but is not blocked unless a hard limit applies to them.

## Next steps

* [Set limits](/docs/guides/set-limits) - admin walkthrough
* [Attribution](/docs/concepts/attribution) - how team and project tags affect spend roll-ups