Genesys Cloud - Main

Good morning everyone,

I'm setting up the Knowledge section in Architect (Ask for Intent > Answer tab) and noticed you can choose between Standard, Highlight, and Generative as the Content Type.

I understand Generative uses an LLM to synthesize an answer from the article that best matches the customer's query, but I had a few questions:

- Is there any case where Standard/Highlight actually performs better than Generative?
- Does Generative add noticeable latency due to the extra retrieval step?
- Has anyone tested this in production and noticed a real difference in customer satisfaction?

Thanks!

#ArchitectandDesign

------------------------------
Matheus Bortolotto
------------------------------

Hello, Matheus

Yes, Standard and Highlight can perform better in some scenarios. Standard is a good choice when the article is already short and direct, while Highlight works well with structured content that contains clearly defined answers. They are also more predictable because the response stays closer to the approved content, which is important for regulated or sensitive use cases.

Generative is usually more effective for longer articles or questions that require a more contextual and conversational response. It may add some latency because the platform first retrieves the most relevant article and then uses the LLM to generate the answer. In most cases, though, the improvement in readability and customer experience can justify that additional processing time.

For production, I recommend running a controlled comparison. Response time is important, but accuracy, containment rate, fallback rate, and customer feedback provide a better view of the real impact. The quality and structure of the knowledge base will also strongly influence the results, regardless of the selected content type.

------------------------------
---------------------
Leticia Roque Oliveira
Senior Full Stack Developer
---------------------
------------------------------