Agentic AppEngine overview (DPS)
- Latest Dynatrace
- Explanation
- 6-min read
Agentic AppEngine is the Dynatrace Platform Subscription capability group for running functions and agentic AI actions. A single agentic action (for example, an AI agent investigating a problem) can generate billed usage across three related but distinct billing types.
Dynatrace can generate three kinds of billed usage, as described in the following table. For current rates and cost examples, see Dynatrace pricing.
| Billed usage | What it meters | When it's metered | Examples |
|---|---|---|---|
AI Units | AI/LLM work: Inference performed by the driver LLM (the operator) for agentic reasoning, and any LLM-based tool that the driver LLM invokes. Available tools are listed in the MCP server documentation. |
|
|
Standard Function Calls | Non-LLM-based tools invoked directly or by the driver LLM. Available AI-invoked tools are listed in the MCP server documentation. |
|
|
Query consumption | The data (GiB) scanned when a tool executes a DQL query, billed under the corresponding query capability (such as Log - Query, Events - Query, or Traces - Query). Query consumption incurred by agentic AI carries the |
|
|
To see all three billing types for AI-generated usage in one place, use the ready-made Usage - Agentic AI dashboard. For more information, see Usage - Agentic AI.
Standard Function Calls
Standard Function Calls are the backend for your app. They are written in TypeScript and run within the Dynatrace JavaScript runtime. Consumption is measured in invocations.
There are three types of Standard Function Calls:
- App functions: These functions represent the backend of an app and are built, bundled, and deployed together with your custom app.
- Ad-hoc functions: Custom code that accomplishes a specific use case is referred to as an ad-hoc function. These functions can be invoked from within
Notebooks, 
Dashboards, 
Workflows, or directly via API. - Custom actions: A specific type of app function that, together with a UI component, can be declared as a custom workflow action to extend Workflows. Custom actions can then be selected as a task within Workflows and can be executed within a workflow.
All Standard Function Calls are deployed in an environment with 256 MiB RAM.
Standard Function Calls work out-of-the-box: No external hosting is required, and there is no need to worry about maintaining a runtime environment to execute logic or code.
Non-LLM-based tools invoked by Dynatrace's agentic AI are also billed as Standard Function Calls.
For more information, see Understand and manage Standard Function Calls consumption (DPS).
AI Units
Dynatrace generative and agentic AI capabilities are metered with AI Units. A single AI Unit is a normalized quantity of AI work performed when a prompt or request runs large-language-model (LLM) inference on your behalf. As the complexity and value of a task grows, so does its AI Unit consumption.
AI Units are consumed, for example, when a generative or agentic AI capability runs as a task in a Workflow, when an external agent calls generative or agentic AI capabilities through the Dynatrace MCP server, or when an API call or custom application invokes generative or agentic AI capabilities.
For more information, see Understand and manage AI Units consumption (DPS).
Query
The data scanned when a query is executed is billed under the corresponding query capability: Log - Query, Events - Query, or Traces - Query.
When a query is triggered by agentic AI (whether directly by the driver LLM or by a Standard Function Call it invokes), the resulting billing usage event carries the ai_generated flag. Filtering on ai_generated == true isolates all AI-triggered query usage and consumption across every query capability.
Example query: AI-generated query consumption (billable GiB)
fetch dt.system.events, scanLimitGBytes: -1| filter event.kind == "BILLING_USAGE_EVENT"| filter in(event.type, array("Log Management & Analytics - Query", "Events - Query", "Traces - Query"))| filter ai_generated == true| summarize billed_bytes = sum(billed_bytes), by:{event.type}| fieldsAdd billed_gib = round(toDouble(billed_bytes) / 1073741824, decimals: 3)
For details on how consumption is calculated for each query capability, see the capability's own page.