Skip to main content

AI Management

The AI Management section centralizes operational visibility and configuration for AI usage in Pyplan. From this area, we can review execution traces, maintain providers and models, and define token-based pricing tiers used for cost calculation.

To open this section, we go to the left sidebar and expand AI Management.

info

In this section, access depends on permissions. We need visibility permissions to review traces, and management permissions to create, edit, or delete providers, models, and pricing tiers.

1. AI Traces​

The AI Traces page helps us inspect AI execution history and investigate behavior for agents, tools, and model calls.

1.1 What we can review in the trace list​

In the main table, we can review key execution data such as:

  • Workflow name
  • Agent
  • Provider
  • Model
  • Tools count
  • Token usage
  • Duration
  • Execution date
  • Status

We can also:

  • Search by text from the page header search box.
  • Sort columns to focus on recent or relevant executions.
  • Use pagination to navigate large result sets.

1.2 Filters available in AI Traces​

We can refine results using:

  • Provider filter
  • Status filter
  • Date from and Date to filters

These filters help us narrow down investigations when we need to analyze a specific period, provider, or execution outcome.

1.3 Viewing trace detail​

To inspect a specific execution, we select a row and click View trace.

In the detail view, we can analyze:

  • The span tree of the execution
  • Parent-child span relationships
  • Status and duration per span
  • Trace payload and error details when available
tip

When troubleshooting, we can first filter by status and date range, then open the trace detail to inspect spans in execution order.

2. Providers​

The Providers page is where we manage AI vendors available in the platform.

2.1 Provider list​

The list displays providers with:

  • Name
  • Code

From this page, we can search, sort, and select a provider.

2.2 Creating a provider​

To create a provider:

  1. We click New provider.
  2. We complete the required fields:
  • Name
  • Code
  1. We save the record.

2.3 Editing and deleting a provider​

To update a provider:

  1. We select a provider from the table.
  2. We click Edit provider.
  3. We update fields and save.

To remove a provider:

  1. We select a provider.
  2. We choose Delete provider.
  3. We confirm in the dialog.

3. Models​

The Models page defines which AI models are available and how they are associated with providers.

3.1 Model list and filter​

The table includes:

  • Model code
  • Provider
  • Description
  • Active status

We can filter by provider directly in the provider column filter and combine it with search and sorting.

3.2 Creating or editing a model​

To create a model:

  1. We click New model.
  2. We select Provider.
  3. We enter Model code.
  4. We optionally add a Description.
  5. We define whether the model is Active.
  6. We save.

To edit a model, we select it from the list and click Edit model.

3.3 Activating or deactivating models​

From the list, we can change the Active switch for each model. This allows us to keep model definitions while controlling operational availability.

4. Pricing​

The Pricing page manages token-based pricing tiers for each model.

4.1 Pricing tier list​

Each tier contains:

  • Model
  • Tokens min
  • Tokens max (optional, unlimited when empty)
  • Input price
  • Cache price
  • Output price

We can filter the table by model, and we can also search, sort, and paginate.

4.2 Creating or editing a pricing tier​

To create a tier:

  1. We click New tier.
  2. We select the Model.
  3. We define Tokens min and, optionally, Tokens max.
  4. We enter Input price, Cache price, and Output price.
  5. We save.

To edit a tier, we select a row and click Edit tier.

4.3 Validation rules for ranges​

When we define ranges, the platform validates pricing tiers to keep consistency.

  • Tokens max must be greater than Tokens min.
  • Tiers for the same model must not overlap.
warning

If ranges overlap for the same model, the platform rejects the update until the token intervals are adjusted.

Summary​

With AI Management, we can manage AI operations end-to-end:

  • We monitor executions in AI Traces.
  • We maintain vendors in Providers.
  • We configure available engines in Models.
  • We control token-cost definitions in Pricing.

This structure helps us keep AI behavior observable, configurable, and aligned with governance and cost control needs.