-
Notifications
You must be signed in to change notification settings - Fork 3.3k
feat(docs): Add docs for injecting custom instructions to Ask DataHub, Docs Gen, + Classification #15197
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merged
jjoyce0510
merged 4 commits into
datahub-project:master
from
jjoyce0510:jj--customize-prompts-instructions
Nov 5, 2025
Merged
feat(docs): Add docs for injecting custom instructions to Ask DataHub, Docs Gen, + Classification #15197
Changes from all commits
Commits
Show all changes
4 commits
Select commit
Hold shift + click to select a range
File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Some comments aren't visible on the classic Files Changed page.
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,127 @@ | ||
| import FeatureAvailability from '@site/src/components/FeatureAvailability'; | ||
|
|
||
| # Ask DataHub | ||
|
|
||
| <FeatureAvailability saasOnly /> | ||
|
|
||
| **Ask DataHub** is DataHub's conversational AI assistant that brings intelligent, context-aware help directly to where you work. Using Ask DataHub, you can ask questions about your organization's metadata and get instant, accurate answers about your data landscape. | ||
|
|
||
| ## What Can Ask DataHub Do? | ||
|
|
||
| Ask DataHub empowers your organization to navigate & understand your entire data ecosystem with ease. | ||
|
|
||
| Key capabilities include: | ||
|
|
||
| ### Natural Language Search | ||
|
|
||
| Search for data assets using plain English instead of memorizing exact table names or technical identifiers. Ask DataHub understands context and synonyms to help you find what you need. | ||
|
|
||
| ### Find Trustworthy Data | ||
|
|
||
| Discover high-quality, reliable data assets based on usage patterns, documentation quality, ownership information, and data quality metrics. Ask DataHub helps you identify the most authoritative sources for your analysis. | ||
|
|
||
| ### Impact Analysis | ||
|
|
||
| Quickly assess how changes to a data asset will ripple through your organization. Ask DataHub can trace lineage and identify all downstream dependencies, helping you make informed decisions before making changes. | ||
|
|
||
| ### Understand Social and Organizational Context | ||
|
|
||
| Get insights into the human side of your data: | ||
|
|
||
| - **Ownership**: Find out who owns and maintains specific assets | ||
| - **Expertise**: Identify domain experts and frequent users | ||
| - **Collaboration**: Understand which teams work with particular datasets | ||
| - **History**: Learn about past usage patterns and organizational knowledge | ||
|
|
||
| ### Assess Data Quality | ||
|
|
||
| Quickly understand the health and reliability of your data assets: | ||
|
|
||
| - View assertion results and data quality scores | ||
| - Understand freshness and validation status | ||
| - Identify potential issues before using data | ||
| - Get context on historical quality trends | ||
|
|
||
| ### Write SQL Queries | ||
|
|
||
| Generate first-draft SQL queries to answer specific analytical questions, accelerating your data exploration and analysis workflows. | ||
|
|
||
| ## Where You Can Use Ask DataHub | ||
|
|
||
| ## Getting Started | ||
|
|
||
| To start using Ask DataHub: | ||
|
|
||
| 1. **In Slack or Teams**: Simply mention @DataHub followed by your question in any channel or chat | ||
| 2. **Enable the feature**: As of DataHub Cloud v0.3.13, Ask DataHub is in public beta and can be enabled in Settings → AI | ||
| 3. **Start asking questions**: Try queries like: | ||
| - "What tables contain customer email data?" | ||
| - "Who owns the sales_facts table?" | ||
| - "What would break if I change the orders table schema?" | ||
| - "Find me the most reliable revenue datasets" | ||
| - "Show me tables with data quality issues" | ||
|
|
||
| ### Slack | ||
|
|
||
| Ask DataHub is available in Slack by mentioning @DataHub in any channel. This brings DataHub's intelligence directly into your team conversations. | ||
|
|
||
| Learn more: [Ask DataHub in Slack](../../managed-datahub/slack/saas-slack-app.md#ask-datahub-in-slack) | ||
|
|
||
| ### Microsoft Teams | ||
|
|
||
| Similarly, you can use Ask DataHub in Microsoft Teams by mentioning @DataHub in channels and chats. | ||
|
|
||
| Learn more: [Ask DataHub in Teams](../../managed-datahub/teams/saas-teams-app.md#ask-datahub-in-slack) | ||
|
|
||
| ### DataHub UI | ||
|
|
||
| Ask DataHub is coming soon to the DataHub UI, providing seamless access to AI assistance while browsing your data catalog. | ||
|
|
||
| Ask DataHub is currently available in the DataHub UI as part of a private beta. If you're interested in enabling this feature, please reach out to your DataHub representative. | ||
|
|
||
| ## Customize Ask DataHub | ||
|
|
||
| As of v0.3.15, you can customize how Ask DataHub responds to queries by configuring custom instructions. These are injected into AI context to tailor the AI assistant's behavior to match your organization's specific needs, terminology, and guidelines. | ||
|
|
||
| ### Configuring Custom Instructions | ||
|
|
||
| To configure custom instructions: | ||
|
|
||
| 1. Navigate to **Settings > AI** in your DataHub instance | ||
| 2. Locate the **Ask DataHub** section | ||
| 3. Enter your custom instructions in the **Instructions** field | ||
|
|
||
| That's it! | ||
|
|
||
| <p align="center"> | ||
| <img width="70%" src="https://raw.githubusercontent.com/datahub-project/static-assets/main/imgs/saas/ai/add_custom_prompts_chat.png"/> | ||
| </p> | ||
|
|
||
| Custom base prompts can be used to: | ||
|
|
||
| - Define response style and tone | ||
| - Define organization specific terminology or terms | ||
| - Guide the model on how to navigate Glossary Terms, Tags, Domains, and properties. | ||
| - Guide the assistant toward specific of recommendations (e.g. helping differentiate production over staging assets) | ||
|
|
||
| ### Default Organization View | ||
|
|
||
| By default, Ask DataHub applies your organization's default **Search View** (if one is set). This means the assistant will prioritize finding assets from a narrower set of assets, enabling it to separate the signal from the noise by default. | ||
|
|
||
| This ensures that Ask DataHub's responses are always relevant to your team's specific data landscape, without requiring additional filtering in your queries. | ||
|
|
||
| :::info Note | ||
| After updating custom prompts or organization view settings, it may take up to 5 minutes for changes to take effect. | ||
| ::: | ||
|
|
||
| ## How It Works | ||
|
|
||
| Ask DataHub leverages your complete metadata graph to provide intelligent, context-aware responses. The AI assistant considers: | ||
|
|
||
| - Asset names, descriptions, and documentation | ||
| - Lineage relationships (upstream and downstream) | ||
| - Ownership and domain information | ||
| - Usage patterns and popularity metrics | ||
| - Data quality and assertion results | ||
| - Tags, glossary terms, and classifications | ||
| - Schema information and sample values |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Uh oh!
There was an error while loading. Please reload this page.