Minor edits.

Author: Sherry Yang

learn-pr/wwl-data-ai/get-started-with-generative-ai-and-agents/includes/2-generative-ai-models.md

@@ -67,7 +67,7 @@ Choosing the right model in Foundry starts with understanding **your workload, t
 |**Multimodal**| Phi‑4‑multimodal‑instruct, GPT‑5.x chat multimodal, Mistral‑Large‑3 | Support for images/audio in chat completions|
 
 > [!NOTE]
-> When the use case is well‑defined, instead of choosing a model from the model catalog, you may choose a [**Foundry tool**](https://azure.microsoft.com/en-us/products/ai-foundry/tools/?msockid=2bbfe2e7589c63f40fd5f7ea5c9c654c#Tools). Foundry tools are powered by prebuilt models that provide predictable performance, built‑in compliance, and fast time‑to‑value without custom modeling.  
+> When the use case is well‑defined, instead of choosing a model from the model catalog, you may choose a [**Foundry tool**](https://azure.microsoft.com/products/ai-foundry/tools/?msockid=2bbfe2e7589c63f40fd5f7ea5c9c654c#Tools). Foundry tools are powered by prebuilt models that provide predictable performance, built‑in compliance, and fast time‑to‑value without custom modeling.  
 
 #### Score and compare models in Foundry 
 

learn-pr/wwl-data-ai/get-started-with-generative-ai-and-agents/includes/3-using-generative-ai-models.md

@@ -14,7 +14,7 @@ Unlike the user prompt, which is the end-user request or question (example: Wher
 
 The playground is a useful bridge between Foundry and code. After you test representative prompts, you can use the same system and user prompts and parameter values in your code. The playground provides code that can call your Foundry deployment via the OpenAI‑compatible *Responses* API. The code is essentially what is running when you use the chat interface to configure settings and send user prompts. 
 
-![Screenshot of code example in Foundry portal that is based in the playground.](../media/chat-code.example.png)
+![Screenshot of code example in Foundry portal that is based in the playground.](../media/chat-code-example.png)
 
 You can take the code as a starting point for creating your own chat client. 
 

learn-pr/wwl-data-ai/get-started-with-generative-ai-and-agents/includes/5-using-an-agent.md

@@ -3,7 +3,7 @@
 ![Screenshot of publishing agent in the Foundry portal.](../media/publish-agent.png)
 ![Screenshot of agent after it is published in the Foundry portal with the endpoint information.](../media/agent-published.png)
 
-Once you've published an agent, you have a stable endpoint to integrate the agent anywhere. You can use an agent from a client application by using the **Foundry Projects SDK** to connect to the project and call it from a client using the **Project API**.
+Publishing an agent gives you a stable endpoint to integrate the agent in code. You can use an agent from a client application by using the **Foundry Projects SDK** to connect to the project and call it from a client using the **Project API**.
 
 The Project API enables you to:
 - Integrate agents into web apps, bots, or backend workflows
@@ -52,11 +52,11 @@ print(f"Response output: {response.output_text}")
 ```
 
 You can evaluate an agent the same way as a model—testing reasoning steps, tool use, and orchestration quality.
-This is helpful when building multi‑step or tool‑enhanced agents. In production environments, model behavior can drift. Foundry supports continuous evaluation to monitor relevance, safety, accuracy, cost and token efficiency. 
+In production environments, model behavior can drift. Foundry supports continuous evaluation to monitor relevance, safety, accuracy, cost, and token efficiency. 
 
 ## Managing costs  
 
-A Foundry agent itself is free — the real cost comes from what it uses:
+A Foundry agent itself is free; the real cost comes from what it uses, mainly:
 - The models it calls
 - The tools it invokes
 - The data services it connects to
@@ -72,7 +72,7 @@ A Foundry agent itself is free — the real cost comes from what it uses:
 |Deployment settings|For example: pay-as-you-go|
 |External knowledge connections|Fabric, SharePoint, etc.|
 
-The more complex and data-rich your agent, the higher the cost. The cost of using a model is influenced by three variables:
+The more complex and data-rich your agent, the higher the cost. Three of the main cost drivers of a model are:
 
 - Billing model (pay‑as‑you‑go vs. provisioned throughput)
 - Deployment option (Standard, Serverless API, Managed Compute)