Issues with Copilot Studio connector to Azure AI Search

Question

Issues with Copilot Studio connector to Azure AI Search

Shabash, Boris 40

I have a copilot agent in Copilot Studio which is using an Azure AI Search index as a data source. Up until Friday May 16th, it was working fine. Today, when connecting the same search index (vectorized) to my copilot studio and then querying it I get the following error:

Error Message: The output returned from the connector was too large to be handled by the agent. Try reducing its size by utilizing available connector filters or by limiting the number of configured action outputs. Error Code: AsyncResponsePayloadTooLarge

This is telling me Copilot Studio is receiving the response differently than it used to several days ago

Is there a new workaround to get around this, my solution requires Azure AI Search as the data source?
If there isn't, what is the timeline to get the connector to behave again?
Are there actual connector filters available for the AI Search service connector? or is the error implying I have to manually create a power automate/copilot set of actions to filter the response? If so, how can this be done given the response is too large to even retrieve?

Please provide me answers for my questions, not just links to the documentation, I've read the documentation and there is no solution there for this issue

Karan Shewale 170 Reputation points Microsoft External Staff

2025-05-20T10:20:13.31+00:00

Thanks for bringing this issue to our attention. We will check and update you soon.
Marco Riezzo 20 Reputation points

2025-05-20T12:21:11.1666667+00:00

same issue here!
Patrick Martin Castro 25 Reputation points

2025-05-26T11:08:25.18+00:00

Please help, im also facing this issue
Naeemeh YadollahPour 0 Reputation points

2025-05-28T13:11:44.2166667+00:00

I had the same problem, I have used custom connector and defined top returned parameters and now works fine. But I have another question, do we need to embed(vectorize) the user query as well?

3 answers

Your answer

Karan Shewale 170 Reputation points Microsoft External Staff

2025-05-20T10:20:13.31+00:00

Thanks for bringing this issue to our attention. We will check and update you soon.
Marco Riezzo 20 Reputation points

2025-05-20T12:21:11.1666667+00:00

same issue here!
Patrick Martin Castro 25 Reputation points

2025-05-26T11:08:25.18+00:00

Please help, im also facing this issue
Naeemeh YadollahPour 0 Reputation points

2025-05-28T13:11:44.2166667+00:00

I had the same problem, I have used custom connector and defined top returned parameters and now works fine. But I have another question, do we need to embed(vectorize) the user query as well?

Answer 1

Hi Shabash, Boris , Marco Riezzo

I’ve analyzed the issue where our Copilot agent, which connects to Azure AI Search, is now failing with the following error:

“The output returned from the connector was too large to be handled by the agent.” Error Code: AsyncResponsePayloadTooLarge

What’s Causing This?

This error is the result of recent changes in Copilot Studio (around May 2024) that enforce stricter limits on how much data it can process from connectors like Azure AI Search. When too many results or large content fields are returned, Copilot stops processing and throws this error.

This is expected behavior going forward and not a temporary bug.

What You Can Do to Fix It:

Option 1 – Limit Data Using Azure Search Query Parameters

You can reduce the response size by adjusting your Azure Search query. The two key options are:

$top: Limits how many records are returned.

$select: Lets you pick only the fields you actually need (e.g., title, summary).

Sample Query:

http

https://<your-search-service>.search.windows.net/indexes/<your-index>/docs
?api-version=2023-11-01-preview
&search=your-query
&$top=3
&$select=title,summary

This keeps the response compact and avoids exceeding Copilot’s payload limits.

References:

Query Parameters for Azure Cognitive Search

Azure Cognitive Search REST API

Option 2 – Use Power Automate to Control the Response

If you need more flexibility or need to further reduce the response before Copilot receives it, you can create a Power Automate flow. This method allows you to:

Call Azure Search within the flow

Use filters and limit the fields/records

Format the output and return only what's essential

This helps keep the response lightweight and under the size limit.

References:

Copilot Studio and Power Automate Integration

Power Automate: Using HTTP Actions

Additional Points:

There is no patch or fix planned from Microsoft — this behavior is permanent.

Azure AI Search doesn’t support built-in filters in Copilot, so we have to manage that at the query or flow level.

Avoid returning full documents or unnecessary metadata — these are often what push the response over the limit.

Thanks,

Karan Shewale.

*************************************************************************

If the response is helpful, please click "Accept Answer" and upvote it. You can share your feedback via Microsoft Teams Developer Feedback link. Click here to escalate.

Marco Riezzo 20 Reputation points

2025-05-21T06:41:31.0433333+00:00

So basically now Copilot Studio is offering the integration with Azure AI Search as knowledge Base but this integration is UN-usable since the context returned is too long and there is no built in way to reduce it.

Please, at least,

share with the community a detailed step-by-step guide to filter this output in Copilot

add possibility to pass filters like top/select in Copilot Studio
Shabash, Boris 40 Reputation points

2025-05-21T19:49:51.82+00:00

@Karan Shewale So does the Azure AI Search connector offer a way to add these parameters (e.g. $top) in its call to Azure AI Search?
Karan Shewale MSFT 0 Reputation points

2025-06-02T05:35:20.94+00:00

Shabash, Boris

Yes, you can use parameters like $top in your Azure AI Search queries to limit the number of results. If you’re using Power Automate or a custom action, you can add these parameters directly to your search call. In Copilot Studio, look for a place to add search options—if you don’t see it, using Power Automate might give you more control.
Karan Shewale MSFT 0 Reputation points

2025-06-02T05:36:28.9466667+00:00

Hi Shabash, Boris,

I wanted to check in and see if you were able to add parameters like $top to your Azure AI Search connector calls. Did that help with your issue?
Trann, Jason 5 Reputation points

2025-06-02T17:06:06.7966667+00:00

I'm not seeing any option to add search parameters. I entered the Endpoint URL and Admin Key, then selected an index. That's it. To me, it seems that this built in connector just isn't functional.
Shabash, Boris 40 Reputation points

2025-06-03T17:46:33.99+00:00

Hi @Karan Shewale MSFT
I wanted to let you know that no, I wasn't able to add such parameters because that option doesn't exist in copilot

Answer 2

Raja Kumar | MAQ Software 0

Steps which i followed to solve the problem:

Created custom connecter to fetch top 5 result.

Add a tool -> Connector -> Search (Hybrid semantic search) -> apply all the filter you want

Created a tool called Prompt(Answer based on given question and data)

User's image

And then printed the response

Marco Riezzo 20 Reputation points

2025-05-26T08:06:01.3966667+00:00

I can not find the tool "Prompt (Answer based on given question and data)".

Where is it located?
Patrick Martin Castro 25 Reputation points

2025-05-26T11:04:03.2633333+00:00

Hi Rajah. where can you find this tool
Raja Kumar | MAQ Software 0 Reputation points

2025-05-26T11:15:19.6133333+00:00

1.You can find the tool like this -

You can search - Provide answer based on given question
Patrick Martin Castro 25 Reputation points

2025-05-26T11:40:10.7933333+00:00

Hi Rajah. Can you provide a more detailed step by step or can you provide an image of your whole topic flow? This will be very helpful thank you. What are you using as trigger etc
Raja Kumar | MAQ Software 0 Reputation points

2025-05-26T11:46:33.7966667+00:00

Step1 - Add a Topic (Blank)

Step 2 Add a tool Semantic Hybrid Search

After connection u will see like this -

Step 3 Parsed the result coming from azure search -

Step 4. Added prompt tool -

Step 5.

Answer 3

Raja Kumar | MAQ Software 0

1.You can find the tool like this -

User's image

You can search - Provide answer based on given question

Patrick Martin Castro 25 Reputation points

2025-05-26T13:52:06.3433333+00:00

Im getting this error The connector 'Azure AI Search' returned an HTTP error with code 400. Error code: ConnectorRequestFailure. Conversation ID: 45f11e73-8886-470d-abe1-fe09ad454275. Time (UTC): 5/26/2025 1:51:29 PM. Even though I had setup the connector.
Brandon T Warner 0 Reputation points

2025-05-28T15:06:37.5566667+00:00

I'm also experiencing the Http error with code 400 when I submit a test prompt to my topic which is made up of the node workflow: Trigger (by agent) -> connector action (Semantic Hybrid Search Azure AI Search) -> Message (just trying to do a simple output. )

I tried two types of connections to my azure ai search instance (Key based and my entra ID via oAuth) with the same Http error with code 400. Both connections indicate the successful green checkmark.

Share via

Issues with Copilot Studio connector to Azure AI Search

3 answers

Your answer