Agents (Chat-based Apps)

An Agent enables a conversational user experience, similar to ChatGPT but with your own data, unique capabilities and defined objectives. It's an incredibly powerful new interaction model that we are just starting to explore.

Have a use case for Agent? We want to hear from you, click Need Help? in the App Builder.

Example Walkthrough

Building an Agent

To build an Agent, you need to establish the Agent goal and connect some tools or data to it. In this example, we are instructing the Agent to research the latest AI & LLM trends and create an informative, casual tweet thread. Then, we connect two tools: one app that uses Google Search to retrieve results from the web, and another app that executes a web browser to extract information from a webpage.

Configuring the Agent

  • System prompt: The system prompt is used to define the behavior and objective of the assistant, giving high-level instructions about how the assistant should behave. For example, the system prompt could be "You are an assistant that speaks like Shakespeare", "You are a JSON API", "You are an expert math teacher".

    You are using the System prompt to set the overall context for the conversation. You can include important context in here and use it provide important information to the model that will aid it in fulfilling it's objective.

  • Example Interactions: These prompts provide the model with example exchanges and help in directing the conversation. User prompts typically contain a user's question or command, while assistant prompts carry the assistant's response.
    If you correctly define the objective in the System prompt, you may not need to provide examples in the User / Assistant format, but this can be helpful, particularly if you are looking for very specific output formats like JSON.

Configuring an Agent Tool

  • Select app: You must have previously created an Airops Tool App. Use the dropdown menu to select the app you wish to connect to the Agent.
  • Select Version: Pin down which published version of the tool you want the assistant to use.
  • Description: Describe to the Agent how and when to use the app tool.

Configuring a memory store tool

  • Select a Memory Store: You must have previously created an AirOps Memory Store. Use the dropdown menu to select the memory store you wish to connect to the Agent.
  • Description: Describe to the Agent how and when to use the memory store tool. An example would be "Use this when answering questions about math".

Interacting with a Chat Agent

You interact with a Chat Agent using a chat box, just like ChatGPT. Looking at our earlier example, let's see how a chat with this assistant might go. Let's say we ask it to make a thread of tweets about news on Large Language Models (LLMs). The Assistant uses the tools we gave it to do this job: it googles LLM news, opens the page in a web browser, and then writes a thread of tweets based on the page's content.

Use cases

Chat Agents can be tailored to a variety of use cases beyond the scenario outlined above. Below are a few examples:

  1. Customer Support: Chat Agents can be equipped to handle common queries and issues customers face. They can be linked to internal databases to fetch real-time information or status updates, and connected to tools like email or CRM software to enable more direct support when needed.
  2. Sales and Marketing: Chat Agents can assist in lead generation and nurturing, reaching out to potential customers, providing information on products or services, and even helping to schedule meetings or calls with sales representatives.
  3. Data Analysis: By linking a Chat Agents with data analytics tools and databases, users can ask for specific data reports, insights, or visualizations. This could be a great way to enable non-technical team members to leverage data more effectively.
  4. Human Resources: Chat Agents can help streamline various HR functions. They could be programmed to answer common employee queries about policies, assist with onboarding by providing relevant documents and resources, schedule meetings with HR personnel, etc.
  5. Healthcare: Chat Agents can be used in telehealth applications, providing initial symptom checks, appointment scheduling, medication reminders, and general health advice. They can be connected to electronic health records (EHRs) to provide personalized care.
  6. Education and Learning: Assistants can be a useful tool for students, providing answers to queries, help with homework, or explanations of complex concepts. They can also assist teachers in managing class schedules, assignments, and providing resources.
  7. Travel and Hospitality: A Chat Agent can be linked to travel apps and databases to provide information on flights, hotel bookings, weather conditions, local attractions, and more. They could even automate the booking process for users.

Remember, the key to building a useful Chat Agent is to clearly define its purpose and equip it with the right tools and access to the necessary data.

Whatโ€™s Next