Forum Discussion
Helper IIIChatgpt like Model in Power BI
Hello,
We are trying to work on AI use case with Power BI and facing difficulties doing that. our problem statement is -
we have a Power BI Report which is having metadata of 10 oracle tables. Everytime when oracle releases its new feature notes we wanted co-pilot/ any LLM/Any AI agent to read the release notes and existing semantic model and let us know if there are any tables getting impacted by new oracle releases.
I tried importing oracle release notes pdf to power bi but since it is unstructured data, its not working. I am also trying to see if we can integrate any chatgpt like ai into Power BI Report so that it can access external resource such as oracle release notes and compare with meta data data present in Semantic model.
I am currently stuck here with no options found. Please let me know if anyone has come through the scenario or aware of any solutions
TIA
Hey vamshikrishna20 ,
First, the Reality
What you’re trying to do cannot be done directly inside Power BI alone.
Power BI is great at:
- Showing data
- Modeling data
- Creating dashboards
But it is not designed to read unstructured documents like PDFs and “understand” them or compare them intelligently with your semantic model.
That’s why importing the Oracle release notes PDF into Power BI didn’t work. It can extract text, but it cannot interpret meaning.
What You Actually Need
Your requirement is:
When Oracle releases new feature notes, automatically check if any of our 10 Oracle tables in Power BI are impacted.
To do this, you need something that can:
- Read the Oracle release notes (PDF)
- Understand which tables or columns are mentioned
- Compare that with your Power BI semantic model
- Tell you what is impacted
This requires an AI/LLM layer outside Power BI.
Simple Working Solution (Practical Approach)
Here’s how you can realistically implement it:
Step 1 – Use an AI Service to Read Release Notes
Use something like:
- Azure OpenAI
- OpenAI API
- Any enterprise LLM tool
This AI will:
- Read the Oracle release notes PDF
- Extract table names, column changes, deprecations, etc.
- Convert them into structured output
Step 2 – Extract Your Power BI Model Metadata
From Power BI, you can export:
- Table names
- Column names
- Relationships
This can be done using:
- XMLA endpoint
- Tabular Editor
- Power BI REST API
Store this metadata somewhere (SQL table, Excel, SharePoint, etc.).
Step 3 – Compare Using AI
Now send both things to the AI:
- Oracle release notes summary
- Your Power BI model metadata
Ask it:
“Tell me which of my tables are impacted and what kind of impact it is.”
The AI will return a structured response.
Step 4 – Show Results in Power BI
Store the AI output in:
- SQL
- Excel
- SharePoint
- Fabric Lakehouse
Then simply build a Power BI dashboard that shows:
| Table | Impact Type | Severity | Action Required |
Now Power BI is just displaying the results — not doing the AI work.
If You Are Using Microsoft Stack
If your company uses Microsoft ecosystem, the best enterprise approach is:
- Microsoft Fabric
- Azure OpenAI Service
This makes integration much smoother and secure.
If it solved your issue, feel free to mark it as the solution so others can benefit too.
Thanks for being part of the community.
8 Replies
- tharunkumarRTK
Super User
Interesting use case
To the best of my knowledge, As of today, native Power BI Copilot primarily works with the metadata available within the semantic model (tables, columns, measures, relationships, synonyms, descriptions, etc.) and report-level metadata (visuals, filters, slicers, pages etc). It does not support bringing in external documents like Oracle release notes as a custom RAG source.
I believe you can achieve this in many ways like using a fabric data agent OR by using any model in Azure AI services. For better control I would suggest you
-
Set up Azure AI Services and provision a model
-
Use Microsoft Fabric Fabric notebook to:
-
Extract semantic model metadata (via semantic link).
-
Extract and clean text from Oracle release notes (PDF).
-
-
Implement a Retrieval-Augmented Generation (RAG) pattern:
-
- You can access the Azure AI model from fabric notebook
https://learn.microsoft.com/en-us/fabric/data-science/ai-services/ai-services-overview -
Feed the release notes as RAG
-
Compare release note content against your semantic model metadata.
- You can access the Azure AI model from fabric notebook
-
Ask the model to identify potential impacts on tables or fields based on release changes.
This way, the AI model analyzes both the semantic metadata and the Oracle documentation together and provides the impact analysis.
Connect on LinkedIn
read my blogs here: techietips.co.in
Did I answer your question? Mark my post as a solution! If I helped you, click on the Thumbs Up to give Kudos.
Proud to be a Super User!
-
- MohdZaid_
Hey vamshikrishna20 ,
First, the Reality
What you’re trying to do cannot be done directly inside Power BI alone.
Power BI is great at:
- Showing data
- Modeling data
- Creating dashboards
But it is not designed to read unstructured documents like PDFs and “understand” them or compare them intelligently with your semantic model.
That’s why importing the Oracle release notes PDF into Power BI didn’t work. It can extract text, but it cannot interpret meaning.
What You Actually Need
Your requirement is:
When Oracle releases new feature notes, automatically check if any of our 10 Oracle tables in Power BI are impacted.
To do this, you need something that can:
- Read the Oracle release notes (PDF)
- Understand which tables or columns are mentioned
- Compare that with your Power BI semantic model
- Tell you what is impacted
This requires an AI/LLM layer outside Power BI.
Simple Working Solution (Practical Approach)
Here’s how you can realistically implement it:
Step 1 – Use an AI Service to Read Release Notes
Use something like:
- Azure OpenAI
- OpenAI API
- Any enterprise LLM tool
This AI will:
- Read the Oracle release notes PDF
- Extract table names, column changes, deprecations, etc.
- Convert them into structured output
Step 2 – Extract Your Power BI Model Metadata
From Power BI, you can export:
- Table names
- Column names
- Relationships
This can be done using:
- XMLA endpoint
- Tabular Editor
- Power BI REST API
Store this metadata somewhere (SQL table, Excel, SharePoint, etc.).
Step 3 – Compare Using AI
Now send both things to the AI:
- Oracle release notes summary
- Your Power BI model metadata
Ask it:
“Tell me which of my tables are impacted and what kind of impact it is.”
The AI will return a structured response.
Step 4 – Show Results in Power BI
Store the AI output in:
- SQL
- Excel
- SharePoint
- Fabric Lakehouse
Then simply build a Power BI dashboard that shows:
| Table | Impact Type | Severity | Action Required |
Now Power BI is just displaying the results — not doing the AI work.
If You Are Using Microsoft Stack
If your company uses Microsoft ecosystem, the best enterprise approach is:
- Microsoft Fabric
- Azure OpenAI Service
This makes integration much smoother and secure.
If it solved your issue, feel free to mark it as the solution so others can benefit too.
Thanks for being part of the community.
v-karpurapudCommunity Support
Thank you for posting your query in the Microsoft Fabric Community Forum, and thanks to tharunkumarRTK , MohdZaid_ and RicardoTraNa for sharing valuable insights.
Could you please confirm if your query has been resolved by the provided solution?If you have any more questions, please let us know and we’ll be happy to help.
Regards,
Microsoft Fabric Community Support Team.
RicardoTraNaResponsive Resident
Hi! Have you tried Power BI Modeling MCP Server?.. in these cases can be really useful! I recommend that you use it directly from GitHub Copilot.
Regards!
vamshikrishna20Hi RicardoTraNa - I dont think Modelling MCp server from Github co-pilot cannot read oracle release notes or any other external documentation. let me know if iam wrong
- RicardoTraNa
Thats correct!
So you need something like this:
Oracle Release Notes
↓
Document Parser
↓
Chunk + Tag
↓
Embeddings
↓
Vector DB (Azure AI Search)
↓
↘
LLM Agent ←→ MCP Power BI Modeling Server
↓
Impact Analysis
↓
Impact Results Table
↓
Power BI Dashboard
- v-karpurapud
We have not received a response from you regarding the query and were following up to check if you had the opportunity to review the information provided. Please feel free to contact us if you have any further questions.
Thank You.
- querixa
Helper IV
Integrating unstructured PDFs like Oracle release notes directly into Power BI can be challenging, as you've experienced. One approach is to use a separate AI service to process the PDFs, extract structured insights, and then feed those into your semantic model for comparison. This often involves Azure OpenAI or similar APIs to analyze the text and identify potential table impacts.
In my own work, I've found that embedding an AI assistant directly into the dashboard can help bridge this gap, allowing natural language queries against both your data model and external documents. For example, a tool like PowerMind enables you to ask questions about your metadata and cross-reference it with processed release notes without needing a separate Copilot license.
You might want to check out this LinkedIn demo showing how an AI assistant can be integrated into Power BI for similar use cases: https://www.linkedin.com/feed/update/urn:li:activity:7438407183161065472. It could give you some practical ideas on handling unstructured data alongside your semantic model.
