Ask Your Table - AI Assistant FAQ – MUI Support

Overview

The AI Assistant feature provides built-in AI-powered querying capabilities for the MUI X Premium Data Grid. It is not currently available for the Pro Data Grid.

We highly recommend reviewing https://mui.com/x/react-data-grid/ai-assistant/ as our docs always provide our most up-to-date technical information. The below covers all billing and general queries about the feature.

Pricing

We offer different tiers according to your query needs, all of which can be accessed and explored in our MUI Console: https://console.mui.com/products/ai-datagrid/overview.

Free Plan

You can generate a free API key from our console without any payment commitment. API keys can be generated here: https://console.mui.com/products/ai-datagrid/api-keys.

The free plan includes 200 queries a month, where billing is calculated on a calendar monthly basis.

The free plan does not include the option to BYOK (Bring Your Own Key), meaning you cannot use your own LLM. Our free tier works exclusively with our MUI Hosted service.

Paid Plans (Starter / Growth / Enterprise or BYOK)

If the free plan doesn't meet your needs, you can subscribe or upgrade to a paid plan directly through the console: https://console.mui.com/products/ai-datagrid/billing. Simply select your plan and add your payment information.

Query limits range from 1,500 for the starter plan to unlimited on the enterprise plan for the MUI hosted service.

For the Bring Your Own Key service, we bill on a Monthly Active Users (MAU) basis. The options are either Basic (100 monthly MAU) or Enterprise (unlimited users). Full information on the plans and distinctions can be found in the console.

Renewal is automatic, and billing limits are only available on the Bring Your Key plans.

Setup & Onboarding

You can access setup instructions, onboarding resources, and a product overview here: https://mui.com/x/react-data-grid/ai-assistant/

Frequently Asked Questions (FAQ)

Where are the AI Assistant queries processed? Are they handled by MUI servers or another third-party provider?

For the MUI Hosted service, the queries are processed by our third-party provider (currently Anthropic). Only the query itself is sent. We use anonymised samples of data so that the model can have a reference for the columns, but you can manually set the samples/examples of the data yourself.

Are any of the data sent through the assistant stored, logged, or reused in any way?

Yes, queries are logged for billing verification, but they are stored in an anonymised format that cannot be linked back to individual users and are retained for 12 months. We don't use user data for training.

Is there a way to anonymise or filter the data before it is sent for processing, or do you provide any recommendations to help us manage that?

The data is automatically anonymized, and the Data Grid samples data from independent rows randomly, but as mentioned above, you can provide your own examples for each one of the columns without sharing any real data.

Does MUI provide a proprietary generative AI API for this feature?

No, there is no API that we provide; it's a built-in Data Grid feature.

Is it possible to integrate external APIs such as OpenAI, Google Gemini, etc.?

Yes, using the Bring Your Own Key plan you're free to integrate the Data Grid with your choice of LLM.

Are there any specific technical requirements to enable and use the AI Assistant?

Nothing major, we'll just provide you with the API keys for the plan you're interested in, and the rest of this integration guide will help you set it up. If you would like to use your own LLM, you can generate your own API key.

What is the process to update the plan?

This can be done directly in the console.

Can we limit the number of queries per month so no additional costs are incurred?

Billing limits can only be set for the BYOK options, but not for the paid plans. We recommend monitoring usage in the console and upgrading if you're frequently hitting your limit. MUI does not charge for additional queries over your plan - the LLM will simply stop processing queries.

What counts as a query? Do development calls count towards this limit?

Yes - any call that that consumes inference resources when submitting a request to the AI Assistant service and and yields a response will count as a query, whether it's made during development or in production. A failed prompt or lack of response will not count as a query.