Agents are best described as NLU (Natural Language Understanding) modules. They can be used by your app, product, or service to transform natural user requests into actionable data. This transformation occurs when a user input matches one of the intents inside your agent. Intents are the predefined or developer-defined components of agents that process user requests.

Agents can be designed to manage a conversation flow with the help of contexts, intent priorities, slot filling, responses, and fulfillment via webhook. The following diagram shows the handling of a user request. An agent encompasses the Dialogflow components.

Create an agent


You can access the settings for an agent by following these steps:

  1. Go to the Dialogflow console
  2. Select your agent
  3. Click the gear icon settings next to the agent name


  • Description: Description of your agent. Displayed in the Web Demo for your agent.
  • Default Time Zone: Default time zone for agent.
  • Google Project:
    • Project ID: GCP project linked to agent.
    • Service Account: Service account for authentication.
  • API Version: API version for agent. Select V2 API for all new agents. Dialogflow Enterprise Edition only supports the V2 API, so the API version cannot be selected.
  • Beta Features: (V2 API only) Toogle to enable beta features for your agent.
  • API Keys V1:
    • Client Access Token: (V1 API only) For end-user interaction with /query, /context and /userEntites endpoints only.
    • Developer Access Token: (V1 API only) Used for agent modification. Do not share.
  • Log Settings:
  • Delete Agent: Completely deletes agent and cannot be undone. If the agent is shared with other users, those users must be removed from the agent before you can delete it.


Add multiple languages and their respective locales to make your agent multilingual.

Choose a language from the list and click the Save button. To add a locale, if available, hover over the listed language and click + Add locale.

For more information, see Multi-language Agents.

ML settings (machine learning)

  • Match Mode: This setting defines the algorithms that should be used for machine learning for all intents in which machine learning is enabled.
    • There are two types of algorithms used to match intents:
      • Rule-based grammar matching:
        • Pros:
          • Accurate with a small or large number of training phrase examples
          • Models are updated quickly
        • Cons:
          • Matching is slow if there are many training phrase examples
          • May produce incorrect results if the agent uses @sys.any or "Allow automated extension" frequently
      • ML matching:
        • Pros:
          • Accurate with a large number of training phrase examples
          • Matching is fast
        • Cons:
          • Inaccurate with a small number of training phrase examples
          • Models are updated slowly
          • Less accurate than grammar matching for agents with many/complex composite entities or training phrase templates
    • You can select one of two modes that use these algorithms:
      • Hybrid: This mode first attempts a rule-based grammar match. If a match is not made, it switches to ML matching. This mode is best for most cases.
      • ML Only: This mode only performs ML matching.
  • ML Classification Threshold: To filter out false positive results and still get variety in matched natural language inputs for your agent, you can tune the machine learning classification threshold. Intent matches have an intent detection confidence value in a range from 0.0 (completely uncertain) to 1.0 (completely certain). If the confidence value is less than the ML classification threshold, a fallback intent is triggered. If there are no fallback intents defined, no intent will be triggered.

  • Automatic Spell Correction:

    If this is enabled and user input has a spelling or grammar mistake, an intent will be matched as though it was written correctly. The detect intent response will contain the corrected user input. For example, if a user enters "I want an aple", it will be processed as though the user entered "I want an apple". This also applies to matches involving both system and developer entities.

    Spell correction is available for all languages supported by Dialogflow.

    Warnings and best practices:

    • Spell correction can't correct ASR (automatic speech recognition) errors, so we don't recommend enabling it for agents using ASR inputs.
    • It is possible for corrected input to match the wrong intent. You can fix this by adding commonly mismatched phrases to negative examples.
    • Spell correction increases the agent's response time slightly.
    • Spell correction should not be used with AoG.
    • Spell correction is trained on general user queries. If an agent is defined using domain-specific jargon, the corrections may be undesired.

Export and import

  • Export as ZIP: Exports the agent as a zip file. For more information, see Exported JSON Fields.
  • Restore from ZIP: Overwrites the current agent with the supplied zip file.
  • Import from ZIP: Adds intents and entities to the current agent from the supplied zip file. If any existing intents or enties have the same name as those in the zip file, they will be replaced.

Environments BETA

This feature allows you to publish and make past or current versions of your agent available to a specific environment. Multiple environments can be set up depending on your needs, like "Development", "Testing", and "Production".

Clicking the PUBLISH A VERSION button takes a snapshot of your agent's intents and entities, and lets you optionally deploy that version to one of your environments.

To learn more about this feature, see Versions and Environments.


These settings are used to share agent access with other developers. See Sharing for more information.

Speech BETA

These settings are specific to the speech response and telephony features. For more information, see Speech Response and Telephony.

Advanced BETA

Currently, there is only one feature controlled from these settings. For more information, see Sentiment Analysis

Was this page helpful? Let us know how we did:

Send feedback about...

Dialogflow Enterprise Edition Documentation