Genesys Cloud - Main

Hi everyone!

I'd like to share a scenario we are currently investigating involving Genesys Cloud Agent Copilot, Knowledge Fabric, and a SharePoint Knowledge Base. We would like to understand whether other organizations have experienced similar behavior.

One of our customers maintains its Knowledge Base in SharePoint, using web articles originally published in the .ASPX format.

To make this content available through Knowledge Fabric, the following approach was implemented:

• The SharePoint articles were converted from .ASPX to .HTML using Power Automate.

• The converted articles were made available to Genesys Cloud through Knowledge Fabric.

• The content was successfully processed, but during operational testing, Agent Copilot did not consistently suggest the expected articles.

• To avoid delaying the pilot, the same articles were also imported manually into Genesys Cloud. In this scenario, the articles could be used by Copilot as expected.

During the investigation, a possible request-volume limitation was initially considered. The configured limit was increased from 20 requests per minute to 200 requests per minute, but the behavior persisted.

An interesting observation occurred during an off-hours test: when only one agent was logged in, Copilot successfully suggested articles from the integrated Knowledge Base. During tests with multiple agents operating simultaneously, the suggestions were not consistently displayed.

We understand that this observation alone does not confirm a concurrency or scalability limitation, and the case is still being investigated with Genesys Care. However, it raises some important questions:

• Has anyone experienced similar behavior when using Knowledge Fabric with SharePoint content?

• Are there any known concurrency, throughput, indexing, or request-volume considerations for Agent Copilot article suggestions?

• Could the number of simultaneously logged-in agents affect the retrieval of Knowledge Fabric content?

• Are there recommended monitoring tools or logs that can help identify where requests are being delayed or rejected?

• Is converting SharePoint articles from .ASPX to .HTML considered a supported and recommended approach for this type of integration?

• What architecture or configuration has worked successfully in larger environments with multiple concurrent agents?

This topic is especially relevant for organizations that already maintain their Knowledge Base in SharePoint and want to use the same content in Agent Copilot without having to manually recreate and maintain articles in Genesys Cloud.

I'd really appreciate hearing from anyone who has implemented a similar architecture or encountered comparable behavior. Any recommendations, lessons learned, or troubleshooting guidance would be very helpful. 👇

#AICopilot(Agent,Supervisor,Admin)

------------------------------
João Luiz
CX Analist
------------------------------

Thanks for sharing such a detailed case. The observation about Copilot working successfully with only one agent logged in is particularly interesting.

I agree that this alone wouldn't be enough to conclude that there is a concurrency or scalability limitation, but I think it is a useful clue to investigate alongside the request volume and indexing behavior.

Regarding the SharePoint content format, I found a recent Genesys Community discussion where a Genesys moderator confirmed that .aspx pages need to be converted/saved as HTML before they can be properly ingested by the SharePoint connector. So the .ASPX → .HTML approach seems consistent with the current supported ingestion model.

One thing I would be interested in understanding is whether the issue happens at the Knowledge Fabric ingestion/indexing stage or later during Copilot retrieval. For example, if the content is successfully indexed and searchable when tested directly, but Copilot suggestions become inconsistent only under higher concurrent usage, that could help narrow down where the problem is occurring.

I also think it would be useful to compare the behavior with the same content imported directly into Genesys Cloud, since that seems to work consistently in your pilot. That could help isolate whether the issue is related to the SharePoint connector, Knowledge Fabric, or the Copilot retrieval layer.

I'd be very interested to hear what Genesys Care finds, especially if they can confirm whether there are any documented throughput, concurrency, or monitoring considerations for this architecture.

------------------------------
Joao
Analyst CX
------------------------------

Thanks for the thoughtful insights! Your suggestions will be very helpful in narrowing down where the issue may be occurring. We'll share any relevant findings from Genesys Care as the investigation progresses.

------------------------------
João Luiz
CX Analist
------------------------------

Original Message:
Sent: 09-01-2026 13:51
From: Joao Saraiva
Subject: Knowledge Fabric and Copilot scalability with SharePoint content

Thanks for sharing such a detailed case. The observation about Copilot working successfully with only one agent logged in is particularly interesting.

I agree that this alone wouldn't be enough to conclude that there is a concurrency or scalability limitation, but I think it is a useful clue to investigate alongside the request volume and indexing behavior.

Regarding the SharePoint content format, I found a recent Genesys Community discussion where a Genesys moderator confirmed that .aspx pages need to be converted/saved as HTML before they can be properly ingested by the SharePoint connector. So the .ASPX → .HTML approach seems consistent with the current supported ingestion model.

One thing I would be interested in understanding is whether the issue happens at the Knowledge Fabric ingestion/indexing stage or later during Copilot retrieval. For example, if the content is successfully indexed and searchable when tested directly, but Copilot suggestions become inconsistent only under higher concurrent usage, that could help narrow down where the problem is occurring.

I also think it would be useful to compare the behavior with the same content imported directly into Genesys Cloud, since that seems to work consistently in your pilot. That could help isolate whether the issue is related to the SharePoint connector, Knowledge Fabric, or the Copilot retrieval layer.

I'd be very interested to hear what Genesys Care finds, especially if they can confirm whether there are any documented throughput, concurrency, or monitoring considerations for this architecture.

------------------------------
Joao
Analyst CX
------------------------------

Hi everyone,

Very interesting scenario. We have seen similar discussions around Knowledge Fabric and Agent Copilot where the main challenge is not necessarily the content itself, but the retrieval path between the external knowledge source, indexing, and Copilot suggestion engine.

Since the same articles work when imported directly into Genesys Cloud, I would investigate the differences in the Knowledge Fabric flow:

• Confirm the indexing status and last synchronization time for the SharePoint-based content.
• Validate the HTML generated from ASPX conversion (structure, metadata, hidden elements, duplicated content, etc.).
• Check whether Copilot is failing to retrieve the article or retrieving it but not presenting it due to relevance/confidence scoring.
• Compare behavior under different concurrency levels, especially considering the single-agent vs. multiple-agent observation.

The concurrency point is particularly interesting. I agree it does not necessarily indicate a platform limitation, but capturing request/response behavior during both scenarios could help identify whether there is throttling, latency, or availability impact somewhere in the chain.

For organizations using SharePoint as the source of truth, it would also be valuable to understand the recommended architecture and supported patterns for large-scale Knowledge Fabric deployments.

Looking forward to hearing the outcome from Genesys Care. This is a very relevant topic as more customers are trying to leverage existing enterprise knowledge repositories with Agent Copilot.

------------------------------
Lucas
------------------------------

Thanks for the valuable insights! These points, especially around indexing, relevance scoring, and concurrency behavior, will be very helpful in guiding our investigation. We'll share any relevant findings from Genesys Care.

------------------------------
João Luiz
CX Analist
------------------------------

Original Message:
Sent: 09-01-2026 13:56
From: Lucas Oliveira
Subject: Knowledge Fabric and Copilot scalability with SharePoint content

Hi everyone,

Very interesting scenario. We have seen similar discussions around Knowledge Fabric and Agent Copilot where the main challenge is not necessarily the content itself, but the retrieval path between the external knowledge source, indexing, and Copilot suggestion engine.

Since the same articles work when imported directly into Genesys Cloud, I would investigate the differences in the Knowledge Fabric flow:

• Confirm the indexing status and last synchronization time for the SharePoint-based content.
• Validate the HTML generated from ASPX conversion (structure, metadata, hidden elements, duplicated content, etc.).
• Check whether Copilot is failing to retrieve the article or retrieving it but not presenting it due to relevance/confidence scoring.
• Compare behavior under different concurrency levels, especially considering the single-agent vs. multiple-agent observation.

The concurrency point is particularly interesting. I agree it does not necessarily indicate a platform limitation, but capturing request/response behavior during both scenarios could help identify whether there is throttling, latency, or availability impact somewhere in the chain.

For organizations using SharePoint as the source of truth, it would also be valuable to understand the recommended architecture and supported patterns for large-scale Knowledge Fabric deployments.

Looking forward to hearing the outcome from Genesys Care. This is a very relevant topic as more customers are trying to leverage existing enterprise knowledge repositories with Agent Copilot.

------------------------------
Lucas
------------------------------

congratulations!!🎉🎉🎉🥳🥳

------------------------------
Alexandre
------------------------------

Hi Joao,

Thanks for sharing your questions.

Product are actively working at natively supporting ingesting Sharepoint sites. If we can deliver it, it will be Q1 2027.

Re limits, you can request this to be increased above 200 to meet your needs via a care ticket.

I would also suggest enabling the new auto-suggestion settings (Detect knowledge questions and Refine search queries( in Agent Copilot to ensure knowledge is searched for at the most appropriate times.

You can read more about them here: https://help.genesys.cloud/articles/configure-the-genesys-agent-copilot-settings/

Amanda

------------------------------
Amanda Halpin
Principal Product Manager, Knowledge @ Genesys
------------------------------

Very interesting scenario, especially the difference observed between testing with a single agent and multiple agents simultaneously.

The SharePoint + Knowledge Fabric approach is also something that can be very relevant for larger environments, so it would be great to understand whether there are any concurrency or throughput considerations involved.

I'm curious to see what Genesys Care finds in this case and whether others in the Community have seen similar behavior.

------------------------------
Gustavo Nogueira
CX - Data Analyst
------------------------------

Hi João,

The one-agent test is an interesting clue. I'd try repeating the same customer question with the same agents and queue, then compare a SharePoint article with its manually imported copy. If the imported article keeps appearing while the Knowledge Fabric version does not, that gives Genesys Care a much narrower place to investigate.

I'd also ask them to check whether Copilot retrieved the article but chose not to suggest it, or whether the retrieval request failed or took too long. Those would look similar to an agent, but point to different issues.

Please share what you find. We're interested in this approach too, especially for teams that already maintain their content in SharePoint.

------------------------------
Fabio Vieira
------------------------------

Hi João Luiz O Alves!

A very interesting use case. Thanks for sharing.

------------------------------
Juliano Fernandes Mendes Pimentel De Paiva
Genesys Analyst
------------------------------