The System page brings together the platform-level information and controls an administrator needs to keep an Ender Turing installation healthy. From here you can see which version is running, check your license and remaining capacity, track how much of that capacity each month consumes, upload a new license file, install or activate ASR languages and platform components, and — for organizations that sign in through a federated identity provider — configure how OIDC users are provisioned.

This page is intended for platform owners and administrators rather than day-to-day users.

The horizontal tabs across the page separate the System overview from related configuration. Select Usage, the final tab, to open monthly and LLM usage reporting.

Only users with the system management permission can open this page. If your role does not include that permission, the System entry does not appear in the Settings menu.

In multi-tenant deployments, several cards on the page are hidden because version, container, and license-file management are handled centrally rather than per organization.

The first card on the page shows the system identity and update status:

If online updates have been disabled at the system level, a banner explains that updates are offline-only. When archives containing updates have been uploaded, an Unpack uploaded archives button appears so you can prepare them for installation.

The License section summarizes what your license grants:

The first card on the Usage tab is Monthly usage. It tracks consumption month by month, starting with the current month. Each row is labeled with the month and year, and the current month is marked Month to date so you can tell that its numbers are still growing.

Two views are available through the toggle in the card title:

By default the most recent two months are shown. Use the chevron toggle on the right side of the card title to expand the view to the most recent six months, or collapse it back to two.

Only conversations that are attributed to an agent count toward the active-agent figures. Conversations without an agent still count toward the conversation count and total duration.

Some organizations have this card configured to split the active-agent figure into two categories instead of showing a single number. When the split is switched on, the Active agents view replaces the single bar for each month with two bars:

An agent's category is based on how many separate days they were active in that month:

Several conversations on the same day count as one active day. Every active agent falls into exactly one category, and the two counts always add up to the single Active agents figure you would see with the split switched off.

Active days — like the monthly grouping itself — are counted from the day each conversation was added to Ender Turing, not from the day the conversation originally took place. If your conversations arrive in bulk imports, several days of real activity can land on a single day here, which lowers an agent's active-day count.

Both bars are measured against the same active-agents limit from your license — the split does not give you two separate allowances. A short caption above the first month restates the 1–10 and 11+ day definitions.

The split applies to every month shown, not only to months after it was switched on. For the current month, an agent moves from part-time to full-time as soon as they become active on their eleventh distinct day, so a month-to-date row can change category mix as the month progresses.

The Show full-time and part-time active agents checkbox in the Monthly usage card title controls this behavior. It is only visible to Ender Turing staff accounts that also have the system management permission, and the same restriction is enforced on the server, so the setting cannot be changed from outside the product either.

The checkbox is off by default for every organization. Selecting or clearing it saves immediately — there is no separate Save button. While the change is being saved the checkbox is disabled, and once it succeeds the card reloads the full six-month history so every visible month reflects the new choice. If saving fails, an error notification appears and the checkbox returns to its previous state.

Administrators in your own organization who do not have the Ender Turing staff account type will not see the checkbox at all, but they do see the full-time and part-time bars once the split has been switched on for the organization.

The second card on the Usage tab reports large language model (LLM) cost and provider tokens over time. It is always available to users whose role has both the system management and automation management permissions; no reporting switch needs to be enabled.

Users with system management permission but without automation management permission continue to see Monthly usage, but the LLM Usage card is not loaded. Users with automation management permission alone cannot open the System Usage tab.

Filters and controls:

The card displays:

When you split By Ender or By Model, some usage may not be tied to a specific Ender or model — for example, usage that did not run through an automation. It is grouped together under an __HTML_PLACEHOLDER_0__ entry in the chart and table so the full picture still adds up.

The metric selector offers:

Token counts are shown as whole numbers, formatted for your interface language. Switching the metric does not change the other filters — the date range, prompt selection, and Split by grouping stay as you set them.

Token counts started being recorded only when this capability was added to your platform. Usage that was recorded earlier has no token information, and no estimate is produced for it. Where token data is missing you will see:

Missing values are never displayed as zero, so you can tell "we have no data" apart from "no tokens were used". Because a single missing record makes the whole aggregate unreliable, the rule applies upward as well: if any day contributing to a row is unavailable, that row's Total for the selected period also shows Unavailable. Input, cached input, and output tokens are tracked independently, so it is normal to see a real number for one of them and Unavailable for another.

Answers served from cache count as zero tokens. When Ender Turing reuses a previously generated response instead of calling the provider again, the run is recorded with zero input, cached input, and output tokens — because no provider request took place. This is a real zero, not missing data.

The Authentication card is shown only on deployments that use a federated identity provider (OIDC). On installations without OIDC configured, this card does not appear.

The card contains one control:

This control is the same option that can otherwise only be flipped by directly editing the underlying setting. If you turn it off, federated users without an existing Ender Turing account will be rejected at sign-in until they are invited by an administrator.

To verify which accounts came in through this flow, open a user from Users & Agents — accounts that were linked to a federated identity show an OIDC linked badge below the last login row.

When an update is detected and confirmed, an additional System update info section appears showing the new version, the changelog, and the current state of the in-flight updater container. From there you can:

These controls are only meaningful in single-tenant deployments where the platform is managed by your own administrators.

The ASR languages card lists the speech-recognition language packs that are part of your license. Each row shows the language name and the versions available. Version chips use color to indicate state:

By default the list hides obsolete entries: it always shows the latest version, the currently active version, and the two most recently downloaded versions, plus any beta or development builds when Show beta versions is on.

The Connectors and Services cards work in the same way as ASR languages. Each row lists an installable component and shows version chips that you can download or activate using the same color-coded indicators. These cards are visible in single-tenant deployments only.

The Upload License File expansion panel lets you apply a new or renewed license:

If the file is rejected, an error message appears under the file selector. After a successful upload the system checks for updates automatically so any new entitlements (such as new languages or higher limits) become visible without further action.

This panel is available in single-tenant deployments only.