Overview of AssistPro

Overview of AssistPro

Note: This feature is in public preview. To gain access, please contact your account manager and/or customer success contacts.

Companion Video

Prerequisites

AssistPro leverages Azure OpenAI. As of the release of this feature, Azure customers are required to request access to Azure OpenAI via this form. In addition, Azure OpenAI has per-region and pre-subscription quotas.

Once access to OpenAI has been granted, you can proceed with the rest of the document.

Enable AssistPro

Note: Microsoft requires that the subscription where AssistPro is deployed is whitelisted for OpenAI. In addition, this feature is in private preview and requires the Nerdio’s team to enable it.

To enable AssistPro:

  1. In Nerdio Manager, at the MSP level, navigate to SettingsIntegrations.

  2. In the AssistPro tile, select Deploy.

  3. Enter the following information:

    • Type the names for the resources Nerdio AssistPro deploys as part of its infrastructure.

    • Customize resources tags: Optionally, expand this to enter custom tags.

  4. Once you have entered all the desired information, select OK.

    The deployment task starts and can take some time depending on the Azure region and other conditions. You can follow the task's progress in the Settings Tasks section.

Use AssistPro

Once all Azure resources for AssistPro are deployed, you may use AssistPro.

To use AssistPro:

  1. In the upper-right corner of Nerdio Manager, select the AssistPro icon.

    Nerdio Manager connects to the bot and downloads the chat history.

  2. Users can ask any question related to Nerdio Manager and send them to AssistPro. After a short period of time AssistPro, displays an answer.

  3. You may select any of the view options:

    • Sidebar:

    • Floating window:

    • Detached window:

  4. To toggle among the different views, select the View icons.

  5. To clear the AssistPro chat history, in the header, select the Trashcan icon. This creates a new chat with no previous message content.

Submit Feedback

AssistPro uses Azure OpenAI, powered by Large Language Model (LLM) that has been augmented with Nerdio-specific information. Due to this being an LLM, answers are not deterministic. See the following Microsoft article for more details.

When an user notices incorrect answer, they can submit feedback in the following ways:

  • Select the feedback icon.

  • Follow Nerdio’s standard support escalation path.

Cost of AssistPro

AssistPro has the following paid components:

  • App Service

  • Application Insights

  • Azure Cosmos DB account

  • Key vault

  • Log Analytics workspace

  • Azure OpenAI service

  • Azure AI Search

Customers can get accurate pricing of the components by using the Azure Pricing calculator .Here are the details on how customers can get the exact cost of AssistPro:

  • App Service: The price depends on the App Service plan that Nerdio Manager is using. See the following Microsoft article for more details.

  • Application Insights: The price depends on the usage of AssistPro. AssistPro uses this service at a lower frequency, and we don’t expect more than 0.5 GB of storage per month. The pricing calculator can help to model that.

  • Azure Cosmos DB Account: 25 USD for 1 GB at 500 RU/s.

  • Key Vault: This service charges per 10,000 transactions. See the following Microsoft article for more details.

  • Azure OpenAI Service: This service’s cost depends on usage of AssistPro and the number of input and output tokens that are being used in each interaction. See the following Microsoft article for more details.

  • Azure AI Search: AssistPro uses the basic tier for this service that is priced at $0.11 per hour , which is approximately $80 per month. See the following Microsoft article for more details.

Note: This is an estimate and not a guarantee of the cost. The Azure costs must be monitored.

Putting all this together, the estimated cost for AssistPro when using out of the box and with up to 7 users as of version 5.7 of Nerdio Manager is $110 -$120 per month.

Functional Considerations

Every chat via AssistPro creates two indexes in the Azure Cosmos database.

LLM implementation work with tokens. The number of tokens is a combination of system prompt, input from user, and output from LLM. The number of tokens defines how much “memory” about a previous exchange in the current conversation the bot has. If a conversation is long, larger than the max token count configured for the model, older data is dropped. However, we expose all the chat history in the UI until the chat history is deleted using the delete history button.

Deployment Considerations

By default, AssistPro deploys all resources in the same region. When possible and applicable, we deploy free or lowest paid tier resources, and as of v5.0 that is not configurable.

AssistPro deploys its own app service.

Feature Limitations

  • Only accounts that are assigned to the AVD admin role can interact with AssistPro.

  • Accounts with the Reviewer role can see the AssistPro option but it does not load.

  • Disabling the feature doesn’t delete Azure resource deployed when the feature was enabled.

  • Users cannot control the throttling limit per day and/or month.

  • Users cannot control what data we store.

  • There is no support for notifications.

  • There is no mechanism for identifying and filtering out false positives.

  • In v5.0, a maximum of 7 users can interact with AssistPro at the same time.

  • In v5.0, implementation uses Azure AI Search.

Was this article helpful?

0 out of 0 found this helpful
Have more questions? Submit a request

Comments (0 comments)

Please sign in to leave a comment.