Final Acrolinx updates and manual review

Author: ChrisHowd

learn-pr/wwl-azure/develop-ai-enabled-apps-using-github-copilot-sdk/includes/2-what-ai-agent.md

@@ -1,14 +1,14 @@
-An AI agent is a goal-driven AI entity that can reason, take actions, and adapt autonomously in pursuit of an objective. Unlike a simple chatbot that only responds to user queries with pre-scripted answers, an AI agent can make decisions, use tools or APIs, and maintain memory of context to achieve a goal without step-by-step human guidance. In essence, chatbots respond to individual queries; AI agents pursue goals through multi-step reasoning and action.
+An AI agent is a goal-driven AI entity that can reason, take actions, and adapt autonomously in pursuit of an objective. Unlike a simple chatbot that only responds to user queries with prescripted answers, an AI agent can make decisions, use tools or APIs, and maintain memory of context to achieve a goal without step-by-step human guidance. In essence, chatbots respond to individual queries; AI agents pursue goals through multi-step reasoning and action.
 
 ## How AI agents differ from chatbots
 
-A traditional chatbot might answer "What's my account balance?" with information from a database. An AI agent, on the other hand, could not only fetch your balance but also proactively alert you if it detects unusual spending, or even execute tasks like transferring funds or scheduling a meeting with your financial advisor. The agent reasons about your goal (managing your finances) rather than just reacting to a single question.
+A traditional chatbot might answer "What's my account balance?" with information from a database. An AI agent, on the other hand, could fetch your balance and proactively alert you if it detects unusual spending. The agent could even execute tasks like transferring funds or scheduling a meeting with your financial advisor. Agents are able to comprehend your goal (managing your finances) rather than just reacting to a single question.
 
 The key distinction is that chatbots follow predetermined conversational paths, while agents dynamically plan and execute multi-step workflows to achieve an outcome.
 
 ## Key capabilities of AI agents
 
-Modern AI agents are powered by advanced AI models (often large language models, or LLMs) and are defined by several core abilities that distinguish them from basic automation:
+Modern AI agents are powered by advanced AI models, often large language models (LLMs), and provide several core abilities that distinguish them from basic automation:
 
 - **Reasoning and planning**: Agents use AI models to interpret requests, break them into steps, and decide what actions to take. Rather than following a fixed script, they formulate plans based on the current situation.
 
@@ -18,7 +18,7 @@ Modern AI agents are powered by advanced AI models (often large language models,
 
 - **Iterative execution**: Agents operate in a loop of reasoning and acting. They reason about what to do next, take an action (like calling a tool), observe the result, and then decide whether to take another action or provide a final response. This loop continues until the agent's goal is accomplished or it reaches a stopping condition.
 
-These capabilities let an AI agent operate with a degree of independence. The agent uses its AI reasoning to decide *what* needs to be done, it acts by calling tools or issuing commands to do it, and it iterates by checking outcomes and refining its plan as needed.
+These capabilities let an AI agent operate with a degree of independence. Agents use AI reasoning to decide *what* needs to be done. With a plan in place, agents call tools or issue commands to complete the defined tasks. Once the tasks are complete, agents check their outcomes, refine their plan as needed, and iterate until they achieve the desired result. This cycle of reasoning, acting, and observing is what enables agents to handle complex, multi-step goals that go beyond simple question-answering.
 
 ## How AI agents compare to automation scripts
 
@@ -36,14 +36,14 @@ AI agents are most effective in scenarios where tasks:
 - Involve multiple steps or decisions.
 - Benefit from automation with minimal human oversight.
 
-If a task is straightforward and single-step (like "show latest sales figures"), a simple query or bot may suffice. But if it's complex (like "find all customers who overpaid and initiate refunds" or "monitor my servers and fix issues"), an agent is more suitable because it can handle the decision process and execution.
+If a task is straightforward and single-step (like "show latest sales figures"), a simple query or bot might suffice. But if it's complex (like "find all customers who overpaid and initiate refunds" or "monitor my servers and fix issues"), an agent is more suitable because it can handle the decision process and execution.
 
 The following table compares the capabilities of chatbots, automation scripts, and AI agents:
 
 | Capability | Chatbot | Automation script | AI agent |
 |---|---|---|---|
 | **Input handling** | Responds to specific queries | Follows predefined triggers | Interprets open-ended requests |
-| **Decision making** | Pre-scripted paths | Hard-coded rules | AI-driven reasoning |
+| **Decision making** | Prescripted paths | Hard-coded rules | AI-driven reasoning |
 | **Tool use** | Limited or none | Fixed integrations | Dynamic tool selection |
 | **Multi-step tasks** | Limited | Sequential only | Adaptive, iterative |
 | **Error handling** | Falls back to default | Stops or retries | Reasons about alternatives |

learn-pr/wwl-azure/develop-ai-enabled-apps-using-github-copilot-sdk/includes/3-what-github-copilot-sdk.md

@@ -4,7 +4,7 @@ The GitHub Copilot SDK is a multi-platform toolkit (currently in Technical Previ
 
 Building a full-featured AI agent from the ground up is complex. You need to handle conversation state management across turns, decide when to call which external API, ensure the AI's responses stay within boundaries, manage multiple AI models for different tasks, and implement safety measures. Building agentic workflows from scratch means you've essentially built a mini platform before you even get to your business logic.
 
-The Copilot SDK removes that complexity by providing a pre-built, production-tested agent engine. The SDK handles authentication, model management, chat sessions, and streaming. You focus on what gets built on top of those building blocks—your domain-specific tools and business logic.
+The Copilot SDK removes that complexity by providing a prebuilt, production-tested agent engine. The SDK handles authentication, model management, chat sessions, and streaming. You focus on what gets built on top of those building blocks—your domain-specific tools and business logic.
 
 ## Architecture overview
 
@@ -35,7 +35,7 @@ The SDK provides several built-in capabilities that simplify building AI agents.
 
 ### Production-grade agent loop
 
-The SDK provides the same proven execution loop that powers GitHub Copilot CLI. This loop manages multi-turn conversations, calls tools, and iterates until the task is complete. You don't need to code the planning and orchestration logic.
+The SDK provides the same proven execution loop that powers GitHub Copilot CLI. This execution loop manages multi-turn conversations, calls tools, and iterates until the task is complete. You don't need to code the planning and orchestration logic.
 
 ### Tool orchestration
 
@@ -111,7 +111,7 @@ The SDK uses an event-driven communication model. When you send a message to a s
 - **SessionErrorEvent**: An error occurred during processing.
 - **ToolExecutionStartEvent** and **ToolExecutionCompleteEvent**: A tool was invoked.
 
-You subscribe to these events to capture the agent's responses and handle errors. For non-streaming scenarios, use `AssistantMessageEvent` to collect the full response. For streaming scenarios (like showing a typing indicator in a chat UI), use `AssistantMessageDeltaEvent` to render tokens as they arrive.
+You subscribe to these events to capture the agent's responses and handle errors. For nonstreaming scenarios, use `AssistantMessageEvent` to collect the full response. For streaming scenarios (like showing a typing indicator in a chat UI), use `AssistantMessageDeltaEvent` to render tokens as they arrive.
 
 ### Session hooks
 

learn-pr/wwl-azure/develop-ai-enabled-apps-using-github-copilot-sdk/includes/4-examine-ai-agent-use-cases.md

@@ -6,9 +6,9 @@ An online retail company deploys an AI agent to handle customer support and orde
 
 ### How it works
 
-The agent connects to the company's backend APIs through registered tools. It has tools for looking up orders, creating return requests, issuing refunds, and sending customer communications. When a customer says, "My order arrived damaged, what can I do?", the agent verifies the order details through an API call, then responds with both empathy and action: "I've initiated a return for you and scheduled a courier pickup for tomorrow. You'll be refunded to your original payment method."
+The agent connects to the company's backend APIs through registered tools. It has tools for looking up orders, creating return requests, issuing refunds, and sending customer communications. When a customer says, "My order arrived damaged, what can I do?", the agent verifies the order details through an API call, then responds with both empathy and action: "I've initiated a return for you and scheduled a courier pickup for tomorrow. You'll receive a refund that's applied to your original payment method."
 
-If the agent encounters a request outside its capability (like a complex policy dispute), it escalates to a human support representative and provides all the context it gathered during the conversation.
+If the agent encounters a request outside its capability (like a complex policy dispute), it escalates to a human support representative and provides the context it gathered during the conversation.
 
 ### Why it matters
 
@@ -27,7 +27,7 @@ This approach automates the time-consuming parts of sales—research, data entry
 
 ## Finance and accounting automation agent
 
-An enterprise finance department uses an AI agent to handle routine financial operations: invoice processing, expense reconciliation, and transaction matching. Each day, the agent processes incoming invoices by extracting relevant data (vendor name, amounts, due dates), then cross-references them against purchase orders in the ERP system.
+An enterprise finance department uses an AI agent to handle routine financial operations: invoice processing, expense reconciliation, and transaction matching. Each day, the agent processes incoming invoices by extracting relevant data (vendor name, amounts, due dates), then cross-references them against purchase orders in the enterprise resource planning (ERP) system.
 
 If the invoice matches the purchase order within acceptable variance thresholds, the agent approves it for payment. When the agent detects discrepancies—like an invoice amount that exceeds the purchase order by more than the allowed tolerance—it flags the issue and routes it to a human accountant with a detailed explanation.
 
@@ -45,7 +45,7 @@ This type of agent helps reduce both stockouts (lost sales) and excess inventory
 
 A technology company uses an AI agent as a first responder to system alerts and incidents. When a monitoring alert fires (such as high CPU usage on a database server), the agent investigates by running diagnostic scripts, analyzing logs, and attempting known remediation steps.
 
-For example, if the agent detects a long-running query causing high CPU, it can terminate the query and verify that performance returns to normal. If it resolves the issue, it logs the actions taken and closes the incident. If the issue persists after exhausting its troubleshooting steps, the agent escalates to a human engineer with all the diagnostic context it gathered.
+For example, if the agent detects a long-running query causing high CPU, it can terminate the query and verify that performance returns to normal. If it resolves the issue, it logs the actions taken and closes the incident. If the issue persists after exhausting its troubleshooting steps, the agent escalates to a human engineer with the diagnostic context it gathered.
 
 This agent is particularly valuable for its speed—it can begin investigating within seconds of an alert, rather than the minutes it takes for a human to notice and respond. Initial deployments typically restrict the agent to safe, reversible actions (like restarting services or clearing temp files) and require human approval for riskier operations.