Agent apps provide a new way for creating virtual agents using large language models (LLM). Rather than defining Conversational Agents (Dialogflow CX) flows, pages, intents, and transitions; you provide natural language instructions and structured data. This can significantly reduce the virtual agent creation and maintenance time, and enable brand new types of conversational experiences for your business.
If you still require the explicit control provided by Conversational Agents (Dialogflow CX) flows in certain conversational scenarios, you can combine the power of Vertex AI and flows in a single hybrid virtual agent.
Limitations
The following limitations apply:
- Agents don't support sending a call companion SMS from the Default Welcome Intent route in the Default Start Flow, but you can enable the call companion SMS option in standard flows.
- The Agent Builder console does not support Conversational Agents (Dialogflow CX) flows.
- Agents created in the Vertex AI console don't open in the Conversational Agents (Dialogflow CX) console, unless the project is granted access to the restricted access feature in the Conversational Agents (Dialogflow CX) console. To request access, contact your Google account team.
Language support
Agent apps support the same languages as Gemini models.
Region support
The following regions are supported:
global
eu
(multi-region)europe-west1
europe-west2
europe-west3
northamerica-northeast1
us
(multi-region)us-central1
us-west1
Settings
This section describes the settings available for agent apps.
General
The following general settings are available for agent apps:
-
A human-readable name for your agent app.
-
The agent app region.
-
If enabled, changes to the agent app are not permitted.
Logging
The following logging settings are available for agent apps:
-
If enabled, logs will be sent to Cloud Logging.
-
If enabled, conversation history will be available. Indicates whether you would like Google to collect and store redacted end-user queries for quality improvement. This setting does not affect whether conversation history is used to generate agent responses.
-
If enabled, conversation history is exported to BigQuery. The Enable Conversation History setting must also be enabled.
GenAI
The following generative AI settings are available for agent apps:
-
Select the generative model that agents should use by default.
-
Select the input token limit for the generative model. This is the maximum token size for input sent to the model. Depending on the model, a token can be somewhere between one character and one word. Smaller token limits have lower latency, but the model input size is limited. Larger token limits have higher latency, but the model input size can be larger.
-
Select the output token limit for the generative model. This is the maximum token size for output received from the model. Depending on the model, a token can be somewhere between one character and one word. Smaller token limits have lower latency, but the model output size is limited. Larger token limits have higher latency, but the model output size can be larger.
-
The temperature for a LLM lets you control how creative the responses are. A low value provides more predictable responses. A high value provides more creative or random responses.
-
List of phrases that are banned for generative AI. If a banned phrase appears in the prompt or the generated response, the agent will return a fallback response instead.
-
Configure sensitivity levels of safety filters with respect to different Responsible AI (RAI) categories. Content will be assessed against the following four categories:
Category Description Hate speech Negative or harmful comments targeting identity and/or protected attributes. Dangerous content Promotes or enables access to harmful goods, services, and activities Sexually explicit content Contains references to sexual acts or other lewd content Harassment Malicious, intimidating, bullying, or abusive comments targeting another individual Content is blocked based on the probability that it's harmful. The sensitivity level can be customized by choosing one of Block few (blocking only high-probability instances of harmful content), Block some (medium and high probability instances), and Block most (low, medium, and high probability) for each category. You can also get access to the Block none restricted option that disables RAI checks for the category after submitting a risk acknowledgment request for your project and receiving approval.
For more information, see configure safety attributes.
-
You can check the enable prompt security check setting to enable prompt security checks. When enabled, the agent will attempt to prevent prompt injection attacks. These attacks may be used to reveal parts of the agent prompt or to provide responses the agent is not supposed to supply. This is accomplished by sending an additional LLM prompt that checks whether the user query is possibly malicious.
This setting is enabled by default for new agents. If your agent was created prior to the release of this setting, the setting is not enabled.
Git
These settings provide a Git integration. Follow the instructions to configure the integration.
Versions and Environments
You can create multiple versions of your agents (known as playbooks in Dialogflow CX) and deploy them to environments. See the versions and environments documentation to learn about best practices when creating versions and environments.
Agent apps versus Dialogflow agents
Over the coming months, the generative features of Conversational Agents (Dialogflow CX) will be migrating to Vertex AI. Once the migration is complete, the two consoles will have feature parity. This documentation focuses on the Vertex AI features, but the concepts are the same.
Agent features are restricted access from the Conversational Agents (Dialogflow CX) console. The following table describes the currently supported agent generative features for each:
Capability | Conversational Agents (Dialogflow CX) | Vertex AI |
---|---|---|
Release stage | Private GA | Public Preview |
Flows | ✅ | |
Input/output parameters for agents | ✅ | |
Function tools | ✅ (not supported in simulator) | ✅ |
Save simulator and conversation history as example | ✅ | |
Custom prompts for data store query rewrite and summarization | ✅ | |
Test cases | ✅ | |
Import and export | ✅ | |
VPC SC | Webhooks ✅ | Open API Tools ✅ |
Security and compliance | CMEK, AxT, DRZ | CMEK, AxT, DRZ |