AI Services

On This Page

Overview

Use the AI Services screen in the Admin Console to control which generative AI models are used in your Appian applications. As AI providers release new and more powerful models, this screen allows you to adopt them in a secure and controlled way to benefit from the latest improvements in performance and accuracy.

Note: Model Updates Required
AI models will be deprecated in certain AWS regions. See which models are impacted and what action is required.

The AI Services screen contains the following tabs:

Appian: This tab allows you to manage the generative AI models provided by Appian.
Google Cloud and Microsoft Azure: Appian features that previously relied on Google Cloud and Azure for generative AI have been deprecated and no longer require configuration in these tabs.

How it works

Selecting a model on the Appian tab makes it available to power various AI capabilities in your environment. For some newer models, you must also select an inference profile. For Appian sites in a US region, the US inference profile is selected by default.

An inference profile is a setting that controls which geographic regions can process your AI data. Unlike older settings that were limited to a single region, inference profiles allow processing in multiple pre-defined regions. This provides greater flexibility for data governance while ensuring access to the most powerful models.

Impact on AI capabilities

Changing the model affects Appian's AI features in different ways.

AI capabilities

The following AI capabilities will automatically begin using newly selected models:

AI Agents
CMS Data Type Recommendation
Composer
Data Fabric Chat
Process Insights
Sample Data Generation
Test Case Generation

Generative AI skills

Generative AI skills behave differently than other AI capabilities when you change models.

New Skills: By default, new AI Skills are created with the latest available model in the environment. However, developers can select any available model instead.
Existing Skills: Your existing AI skills will not automatically upgrade. To use the new model, developers will see a banner in the prompt builder with instructions to perform the upgrade. If a new model is not selected, existing skills will continue to use the previously available models.

Tip: When a model is deprecated, AI skills don't automatically switch to a new model. Instead, they default to a region-specific model until you update the AI skill. Developers must manually update each AI skill to use a supported model. If your site is outside the US, you may also need to choose an inference profile in the Admin Console to keep your AI skills running.

In the US, customers already have an inference profile enabled. If an AI skill still uses Sonnet 3.5 by March 1, 2026, Appian will automatically switch it to the region's default model to prevent application issues.

Inference profiles: Control where your AI data is processed

Inference profiles define the regions through which data may move when using the model. Unlike older cross-region settings that limited you to a single region, inference profiles offer more flexibility by allowing processing in multiple pre-defined AWS regions.

Inference profiles are required for some generative AI models, particularly newer models. The required profile is automatically enabled for sites in US regions. Administrators for sites in other regions, such as Europe or Asia-Pacific, must manually select an inference profile before legacy models are retired.

After you select an inference profile, the model card displays the supported regions for that profile. This lets you confirm where data could be processed before saving your selection. See the Security and Compliance page for details about regional availability.

Scroll or drag to pan · ⌘+/⌘− to zoom · Esc to close

Note: Selecting an inference profile outside your site's primary region may introduce data latency.

Considerations when enabling AI models

Selecting a model on the Appian tab changes which models are available for your AI capabilities. Before you enable a model, keep these considerations in mind:

System administrators can select models and inference profiles in the Admin Console. For sites in a US region, a US inference profile is enabled by default.
For FedRAMP sites, AI models are automatically configured based on your service agreements and cannot be selected in the Admin Console.
If you select an inference profile that doesn't match your site's region, Appian automatically uses one of the regions defined in that profile to connect to the generative AI service. This does not cause an error.

Model deprecation schedule

Models are periodically deprecated by their respective providers, such as Anthropic and AWS. When this happens, Appian updates the models that are available to ensure the best performance, security, and compliance. Older models may be deprecated as part of this lifecycle. We communicate these changes to help you plan ahead.

The following table lists upcoming model deprecations.

Deprecation date	Model	Impacted regions
March 1, 2026	Anthropic Claude 3.5 Sonnet	- US East (N. Virginia) us-east-1 - US West (Oregon) us-west-2 - Europe (Frankfurt) eu-central-1 - Europe (Zurich) eu-central-2
April 28, 2026	Anthropic Claude 3.7 Sonnet	All regions where this model is offered

Note: Make sure you update your models before the deprecation date to prevent disruptions.

Update your AI model before March 1, 2026 to avoid downtime

To ensure Appian AI capabilities use the most capable models, this section provides guidance on an upcoming change from Amazon Web Services (AWS).

AWS announced that the Anthropic Claude 3.5 Sonnet model will no longer be supported in certain AWS regions after March 1, 2026. This section helps you determine whether your Appian site is impacted and outlines any actions needed to keep your AI capabilities working.

Check this table to determine whether an update is needed for your region.

Update required	Update encouraged	No update needed
- US East (N. Virginia) us-east-1 - US West (Oregon) us-west-2 - Europe (Frankfurt) eu-central-1 - Europe (Zurich) eu-central-2	- Asia Pacific (Mumbai) ap-south-1 - Asia Pacific (Sydney) ap-southeast-2 - Canada (Central) ca-central-1 - Europe (Ireland) eu-west-1 - Europe (London) eu-west-2 - Europe (Paris) eu-west-3 - South America (Sao Paulo) sa-east-1	- Asia Pacific (Tokyo) ap-northeast-1 - Asia Pacific (Seoul) ap-northeast-2 - Asia Pacific (Singapore) ap-southeast-1

Update required

Impact: Starting March 1, 2026, the following tools will no longer work in these regions unless you update your configuration: AI Agents, CMS data type recommendation, Composer, Data Fabric chatbot, generative AI skills, sample data generation, Summary tab in process insights, Test case generation.
Action required: The US inference profile is automatically configured for sites in these regions, enabling access to supported models like Claude Sonnet 4.5. To complete the update, ensure your existing generative AI skills are upgraded to use a newer model before March 1, 2026.

Update encouraged

Impact: AI capabilities already use a fallback model (like Claude 3 Sonnet) or they're already unavailable.
Recommended action: Although no action is required to prevent disruption, we encourage updating your AI model configuration. Upgrading to a more advanced model improves results and enables additional tools, including Appian Composer, Test case generation, and AI Agents. When you choose an inference profile, your data is handled across the regions included in that profile while using AI features. Before choosing one, make sure the available profiles align with your organization's policies and regulatory needs.

No update needed

Impact: Claude 3.5 Sonnet will remain supported in these regions. AI capabilities will continue to work as they do today. You can still upgrade to newer models to benefit from the latest advancements in generative AI.

Feedback

Was this page helpful?