Overview

Flows are a powerful tool for guiding deterministic agent behavior. You can use them to gain greater control by breaking down conversations into discrete steps to guide your agent in complex tasks like booking, altering, and canceling reservations.

A flow is a series of steps that guide the agent through a structured interaction. Each step in a flow has its own unique prompt, similar to the knowledge base content field. This modular design makes flows particularly useful for scenarios like collecting user information, processing requests for reservations or bookings, or structuring dynamic conversations with complex logic.

Flows are triggered by specific functions or action fields, and once a flow is active, the agent will progress through its steps sequentially, following defined transition rules.

​

Key features

​

1. Step-by-step prompts

Each step in a flow has a dedicated prompt to keep processing focused. This helps ensure the agent processes one task at a time, and does not include any superfluous information in its answers.

​

2. Hidden prompts

Only the current step’s prompt is visible to the model, keeping the conversation on track. This prevents the agent from “seeing” future steps.

​

3. Transition functions

Transition functions move the agent from one step to the next. These functions are defined within the flow and can only route to other steps within the same flow.

​

4. Global functions

Functions not tied to specific steps can be used throughout the agent’s workflows. These functions allow for reusable logic, such as API calls or user input validation.

​

5. ASR biasing

Flows support ASR (Automatic Speech Recognition) biasing to improve transcription accuracy. This is especially useful for collecting alphanumeric data like postcodes or phone numbers.

​

Best practices for flows

​

Keep steps concise

• Ensure each step handles a single cohesive task, like collecting a name or date.
• When handling related pieces of information like the date, time, and number of people for a restaurant booking, combine them into one step for cohesion.

​

Use meaningful function names

• Use descriptive names for functions (for example, SaveLastName or CollectPhoneNumber).
• Avoid naming functions based on future tasks (e.g., GoToNextStep), as this can confuse the model.

​

Keep your design modular

• Use flows to encapsulate logic for specific tasks, like booking a reservation or handling a customer query.
• Incorporate fallback mechanisms to handle unexpected inputs.

​

Always exit flows

• Always include an explicit exit function (ExitFlow) to prevent the agent from looping through unnecessary steps.
• Exiting the flow resets the agent’s focus, allowing it to handle new user queries effectively.

​

Advanced features

​

Conditional logic

Flows can include conditions to dynamically adapt responses based on user input. For example:

• If a user provides a specific date, the agent can check availability and guide the user accordingly.

​

Variables in prompts

Flows support the use of variables to dynamically fill in information during interactions. This allows for personalized and context-aware responses.

​

ASR biasing

• Configure ASR biasing for specific steps to improve transcription accuracy for numbers, postcodes, or other alphanumeric inputs.
• Boost keywords to prioritize relevant user inputs during transcription.

​

Example flow

Here’s an example of a flow for making a restaurant reservation:

1. Start step: Collect the date, time, and number of people.

   • Prompt: “What date and time would you like to book a table for? How many people will be dining?”
   • Function: CollectReservationDetails

2. Check availability:

   • Call an API to check availability based on user input.
   • If available: Transition to the confirmation step.
   • If unavailable: Prompt the user to choose a different time.

3. Confirmation step:

   • Prompt: “Great! Your table is booked for [date] at [time] for [number of people]. Is there anything else I can help you with?”
   • Function: ExitFlow

​

Connecting steps

You can connect flow steps from the prompt field:

1. Type ”/” in the prompt field.
2. Scroll down until you find Steps.
3. You will see a searchable and scrollable list of all the current flow steps. Select one to create a flow function.
4. Name your transition flow function, or retain the autogenerated name.in

​

Managing transition functions

Find the Flow functions button in the bottom-right corner of the flow editor. Click it to open a modal displaying a searchable list of all the functions associated with the flow.

1. If no transitions exist, you can manually create them by connecting steps.

2. When selecting a step to connect, you can name the transition rather than relying on an auto-generated name.

3. If a name is already in use, an error message will prompt you to resolve it before proceeding.

4. Once a transition is created, the UI displays its references, showing how many times it appears in prompts.

5. Transitions can be managed in two ways:

   • Directly from a step, by selecting an option from the context menu.
   • In the flow editor, by selecting an existing transition and connecting steps accordingly.

6. Users can also duplicate transitions, in which case the duplicate will automatically be uniquely named.

​

Troubleshooting

• Missing exit step: Failing to exit a flow can lead to the agent repeating unnecessary steps or ignoring new user queries.
• Overloading steps: Including too much information in a single step can overwhelm both the agent and the user.
• Poorly-named functions: Functions with vague or misleading names can cause unexpected behavior. Remember that when using the PolyAI Agent Studio, all title and description fields matter to the LLM.