Generative AI is an artificial intelligence technology that uses language models to generate original content, providing the ability to understand and respond in natural language.
In Copilot Studio, you can use the following Generative AI features to retrieve and create content, individually or in combination.
- Create an agent . No need to manually draft the topic; an empty agent can generate answers based on the knowledge sources you specify, such as websites and files.
- Leveraging AI's general knowledge . When the "Allow ungrounded responses" option is enabled, the agent can answer general questions unrelated to your specific knowledge sources or topics.
- Draft your theme using natural language . Describe what you want your theme to do, and Copilot Studio will generate it for you. Your agent includes conversational responses and various node types. Use the suggested default theme as a starting point for further development.
- Write prompts using natural language . Describe the prompt you want to create, and Copilot Studio will generate it for you. You can use the suggested default prompt as a starting point for further development.
- Create agent flows using natural language . Describe the overall flow you want your agent to follow, and Copilot Studio will generate it for you. You can use the suggested default flow as a starting point for further development.
- Use orchestration generation . This allows the agent to select the most appropriate topics, tools, agents, and knowledge sources at runtime.
Copilot Studio's Generative AI features change the way you build agents, significantly reducing manual work and configuration.
Prerequisites
- I have a Copilot Studio account.
- Understand the limitations of generating responses.
- Learn about Azure OpenAI.
Generating answer
With generated answers, your agent can find and present information from multiple sources, both internal and external. You don't need to manually create numerous topics that may not answer all of your customers' questions. Use generated answers as your primary source of information or as a backup when user-generated topics fail to answer user questions. By using generated answers, you can quickly create and deploy an effective agent.
What has changed?
When a traditional chatbot cannot determine the user's intent, it will ask the user to rephrase their question. If, after two prompts, the chatbot still cannot determine the user's intent, it will forward the request to a live agent.
Before forwarding to the live agent, the Copilot Studio agent, capable of generating responses, will use natural language processing (NLP) to:
- Analyze what the user types to determine what they are asking.
- Search, gather, and analyze relevant information from designated sources, such as your company website or various other sources including SharePoint.
- Summarize the search results in easily understandable language and send it to the user agent.
Your workflow might look like this:
- You create an agent and test it thoroughly. By default, your agent has the generative orchestration feature enabled in the Generative AI settings .
- After testing, you publish your agent. It can immediately provide answers, assistance, and guidance to your customers.
- You create individual topics for frequently asked questions. These topics can be developed from analysis of previous agents or existing support issues.
General knowledge of AI
In addition to generated answers, you can allow your agents to use AI's general knowledge to find and present information to answer customer questions. General knowledge eliminates the need to manually create numerous topics, which may not address all customer questions. It can also be useful when existing topics fail to meet customer intent.
What is the general knowledge base of AI?
General knowledge of AI applies AI capabilities to access and deliver information, insights, and support across a wide range of topics.
Why should you use it?
- Accessibility : Agents have immediate access to a vast repository of information and knowledge across multiple fields.
- Flexibility : Capable of addressing a wide range of topics and tasks, becoming a versatile resource for diverse needs.
Note: While general knowledge of AI can provide valuable information and support, you should evaluate the information objectively and consider consulting other sources for further verification or clarification when necessary.
What features are supported?
Creating AI-based content may be subject to limitations in usage or file size regulation.
Note : Long prompts may cause errors in generated response nodes due to usage limits or capacity throttling. Use short, focused prompts or break down queries.
Limit
Limits are default constraints applied to agents, limiting the frequency with which you can send messages to an agent. The purpose of limits is to regulate the client's service load, helping to protect the service from overload and preventing clients from using resources unnecessarily.
Agents capable of generating responses have a limit on the number of queries they can make to retrieve answers from the URL you specify. Regular conversations using the agent's topic are subject to standard limits and restrictions.
Language
To see a list of supported languages, see the Language Support section .
You've just finished reading the article "Overview of AI-based agent creation" edited by the TipsMake team. We hope this article has provided you with many useful tech tips and tricks. You can search for similar articles on tips and guides. Thank you for reading and for following us regularly.