Flow Basics

ALPHA RELEASE

Switchboard is currently in preview (v0.1.0-preview.17). APIs may change between releases.

Contact flows are the heart of Amazon Connect - they define how calls are handled, routed, and processed. This guide covers the fundamentals of building contact flows with Switchboard.

What is a Contact Flow?

A contact flow is a sequence of actions that Amazon Connect executes when handling a customer interaction. Think of it as a flowchart that defines:

• What customers hear (messages, prompts)
• What input is collected (DTMF, voice)
• Where calls are routed (queues, agents)
• What logic is executed (Lambda functions, conditions)
• How calls end (disconnect, transfer, voicemail)

Flow Types

Amazon Connect supports several flow types:

Type	Purpose	Entry Point
Contact Flow	Main inbound/outbound flows	Phone number, transfer
Customer Queue	While customer waits in queue	Set by routing
Customer Hold	While customer is on hold	Agent places on hold
Customer Whisper	Played to customer before connecting	Before agent connection
Agent Whisper	Played to agent before connecting	Before customer connection
Transfer to Agent	When transferring to specific agent	Quick connect
Transfer to Queue	When transferring to queue	Queue transfer

Most Common: Contact Flow (90% of use cases)

Your First Flow

Switchboard offers two equally valid approaches for building flows. Choose the one that fits your style and requirements.

Using Attributes

The declarative approach with minimal code:

using NickSoftware.Switchboard;

[ContactFlow("WelcomeFlow")]
public partial class WelcomeFlow : FlowDefinitionBase
{
    [Message("Welcome to our contact center")]
    public partial void Welcome();

    [TransferToQueue("GeneralSupport")]
    public partial void Transfer();
}

Using Fluent Builders

The programmatic approach with full control:

using NickSoftware.Switchboard;

public class FlowDefinitions
{
    public static IContactFlow CreateWelcomeFlow()
    {
        return new FlowBuilder()
            .SetName("WelcomeFlow")
            .PlayPrompt("Welcome to our contact center")
            .TransferToQueue("GeneralSupport")
            .Build();
    }
}

CDK Integration

using Amazon.CDK;
using NickSoftware.Switchboard;

var app = new App();

var builder = Host.CreateApplicationBuilder();

builder.Services.AddSwitchboard(options =>
{
    options.InstanceName = "MyContactCenter";
    options.Region = "us-east-1";
})
.AddFlowDefinitions(typeof(Program).Assembly);

var host = builder.Build();
var switchboard = host.Services.GetRequiredService<ISwitchboardApp>();

app.Synth();

Flow Structure

Every flow has three core components:

1. Entry Point

Where the flow begins execution:

[ContactFlow("MainInbound")]
public partial class MainInboundFlow
{
    // This is the entry point (first method)
    [Message("Thank you for calling")]
    public partial void EntryPoint();

    // Subsequent actions...
}

2. Actions

Individual steps in the flow:

[ContactFlow("CustomerService")]
public partial class CustomerServiceFlow
{
    // Action 1: Play message
    [Message("Welcome")]
    public partial void WelcomeMessage();

    // Action 2: Get input
    [GetUserInput("Press 1 for sales, 2 for support", MaxDigits = 1)]
    public partial void GetMenuChoice();

    // Action 3: Transfer
    [TransferToQueue("Sales")]
    public partial void TransferToSales();
}

3. Transitions

How actions connect:

[ContactFlow("MenuFlow")]
public partial class MenuFlow
{
    [Message("Main menu")]
    public partial void MainMenu();
    // Automatically transitions to next action

    [GetUserInput("Press 1 or 2", MaxDigits = 1)]
    [Branch(OnDigit = "1", Target = nameof(Sales))]
    [Branch(OnDigit = "2", Target = nameof(Support))]
    public partial void GetChoice();
    // Transitions based on user input

    [TransferToQueue("Sales")]
    public partial void Sales();

    [TransferToQueue("Support")]
    public partial void Support();
}

Common Flow Patterns

Simple Welcome and Transfer

The most basic flow:

[ContactFlow("SimpleFlow")]
public partial class SimpleFlow
{
    [Message("Thank you for calling. Connecting you to an agent.")]
    public partial void Welcome();

    [TransferToQueue("GeneralSupport")]
    public partial void Transfer();
}

IVR Menu

Collect input and route accordingly:

[ContactFlow("IvrMenu")]
public partial class IvrMenuFlow
{
    [Message("Welcome to Acme Corp")]
    public partial void Welcome();

    [GetUserInput(
        "For sales, press 1. For support, press 2. For billing, press 3.",
        MaxDigits = 1,
        Timeout = 5
    )]
    [Branch(OnDigit = "1", Target = nameof(Sales))]
    [Branch(OnDigit = "2", Target = nameof(Support))]
    [Branch(OnDigit = "3", Target = nameof(Billing))]
    [Branch(OnTimeout = true, Target = nameof(Timeout))]
    public partial void GetMenuSelection();

    [TransferToQueue("Sales")]
    public partial void Sales();

    [TransferToQueue("Support")]
    public partial void Support();

    [TransferToQueue("Billing")]
    public partial void Billing();

    [Message("We didn't receive your selection")]
    [Loop(Target = nameof(GetMenuSelection), MaxIterations = 3)]
    public partial void Timeout();
}

Smart IVR with Speech Recognition

Use Amazon Lex for natural language input with DTMF fallback:

Attribute-Based:

using Switchboard.Attributes;
using Switchboard.Actions;

[ContactFlow("SmartIVR")]
public partial class SmartIVRFlow : FlowDefinitionBase
{
    [Action(Order = 1)]
    [Message("Welcome to Acme Corporation")]
    public partial void Welcome();

    [Action(Order = 2)]
    [GetCustomerInput(
        InputType = CustomerInputType.Sequential,  // Try ASR, fallback to DTMF
        Text = "I can help you with sales, support, or billing. What do you need?",
        LexBot = "AcmeIVR",
        LexBotAlias = "Production",
        ConfidenceThreshold = 0.6,
        FallbackText = "Or press 1 for sales, 2 for support, 3 for billing")]
    [MaxDigits(1)]
    public partial void GetDepartment();

    [Action(Order = 3)]
    [Branch(Input = "$.CustomerInput")]
    public partial void RouteByDepartment()
    {
        [Case("Sales")]  // Lex intent or DTMF digit "1"
        void Sales()
        {
            [Message("Connecting to sales...")]
            void NotifySales();

            [TransferToQueue("SalesQueue")]
            void TransferSales();
        }

        [Case("Support")]  // Lex intent or DTMF digit "2"
        void Support()
        {
            [Message("Transferring to support...")]
            void NotifySupport();

            [TransferToQueue("SupportQueue")]
            void TransferSupport();
        }

        [Case("Billing")]  // Lex intent or DTMF digit "3"
        void Billing()
        {
            [Message("Routing to billing...")]
            void NotifyBilling();

            [TransferToQueue("BillingQueue")]
            void TransferBilling();
        }

        [Default]
        void NoMatch()
        {
            [Message("I didn't understand. Let me connect you to an agent.")]
            void NotifyNoMatch();

            [TransferToQueue("GeneralQueue")]
            void TransferGeneral();
        }
    }
}

Fluent API:

using Switchboard;
using Switchboard.Configuration;
using Switchboard.Enums;

var flow = new FlowBuilder()
    .SetName("SmartIVR")

    .PlayPrompt("Welcome to Acme Corporation")

    .ThenGetInput(input =>
    {
        // Primary: ASR with Amazon Lex
        input.Primary.Mode = InputMode.ASR;
        input.Primary.LexBot = "AcmeIVR";
        input.Primary.LexBotAlias = "Production";
        input.Primary.Locale = "en_US";
        input.Primary.Prompt = "I can help you with sales, support, or billing. What do you need?";
        input.Primary.ConfidenceThreshold = 0.6;

        // Fallback: DTMF
        input.Fallback.Mode = InputMode.DTMF;
        input.Fallback.PromptText = "Or press 1 for sales, 2 for support, 3 for billing";
        input.Fallback.MaxDigits = 1;

        // Enable Sequential Mode (ASR → DTMF)
        input.EnableFallback = true;
        input.FallbackTriggers = FallbackTrigger.AnyError;
    })
    .RouteByInput(router =>
    {
        router.OnIntent("Sales", flow =>  // Lex intent
        {
            flow.PlayPrompt("Connecting to sales...")
                .ThenTransferToQueue("SalesQueue");
        })
        .OnDigit("1", flow =>  // DTMF fallback
        {
            flow.PlayPrompt("Connecting to sales...")
                .ThenTransferToQueue("SalesQueue");
        });

        router.OnIntent("Support", flow =>
        {
            flow.PlayPrompt("Transferring to support...")
                .ThenTransferToQueue("SupportQueue");
        })
        .OnDigit("2", flow =>
        {
            flow.PlayPrompt("Transferring to support...")
                .ThenTransferToQueue("SupportQueue");
        });

        router.OnIntent("Billing", flow =>
        {
            flow.PlayPrompt("Routing to billing...")
                .ThenTransferToQueue("BillingQueue");
        })
        .OnDigit("3", flow =>
        {
            flow.PlayPrompt("Routing to billing...")
                .ThenTransferToQueue("BillingQueue");
        });

        router.OnNoMatch(flow =>
        {
            flow.PlayPrompt("I didn't understand. Let me connect you to an agent.")
                .ThenTransferToQueue("GeneralQueue");
        });
    })

    .Build();

Learn More

See Speech Recognition (ASR) for comprehensive ASR documentation including:

• Amazon Lex bot setup
• Confidence threshold tuning
• Multi-language support
• Advanced routing by intent

Business Hours Check

Route differently based on time:

Attribute-Based:

[ContactFlow("BusinessHoursFlow")]
public partial class BusinessHoursFlow
{
    [PlayPrompt("Welcome to our support center")]
    public partial void Welcome();

    [CheckHoursOfOperation("BusinessHours")]
    public partial void CheckIfOpen();

    // Branch methods would be generated for "InHours" and "OutOfHours"
    // (Source generator support coming soon)
}

Fluent API:

var flow = new FlowBuilder()
    .SetName("BusinessHoursFlow")

    .PlayPrompt("Welcome to our support center")

    .CheckHoursOfOperation("BusinessHours")
    .WhenInHours(inHours => inHours
        .PlayPrompt("Our agents are available now. Please hold.")
        .TransferToQueue("Support")
        .Disconnect())
    .WhenOutOfHours(afterHours => afterHours
        .PlayPrompt("We're currently closed. Our business hours are Monday-Friday, 9am-5pm.")
        .PlayPrompt("s3://my-bucket/voicemail-instructions.wav")
        .Disconnect())

    .Build();

Customer Authentication

Verify customer before routing:

[ContactFlow("AuthenticatedFlow")]
public partial class AuthenticatedFlow
{
    [Message("Welcome. For your security, we need to verify your identity.")]
    public partial void Welcome();

    [GetUserInput("Please enter your 4-digit PIN", MaxDigits = 4, Timeout = 10)]
    public partial void GetPin();

    [InvokeLambda("CustomerAuthFunction", Timeout = 5)]
    [OnSuccess(Target = nameof(Authenticated))]
    [OnError(Target = nameof(AuthFailed))]
    public partial void ValidatePin();

    [Message("Thank you. You are verified.")]
    [SetAttribute("Authenticated", "true")]
    [TransferToQueue("VIPSupport")]
    public partial void Authenticated();

    [Message("Invalid PIN. Please try again.")]
    [Loop(Target = nameof(GetPin), MaxIterations = 3)]
    public partial void AuthFailed();
}

Callback Flow

Offer callback instead of waiting:

[ContactFlow("CallbackFlow")]
public partial class CallbackFlow
{
    [Message("Welcome")]
    public partial void Welcome();

    [GetQueueMetrics("Support", Metric = "CallsInQueue")]
    [CheckAttribute("CallsInQueue", CompareType.GreaterThan, "10")]
    [OnTrue(Target = nameof(OfferCallback))]
    [OnFalse(Target = nameof(TransferDirectly))]
    public partial void CheckQueueDepth();

    [Message("The current wait time is over 10 minutes")]
    [GetUserInput("Press 1 to hold, or 2 to receive a callback", MaxDigits = 1)]
    [Branch(OnDigit = "1", Target = nameof(TransferDirectly))]
    [Branch(OnDigit = "2", Target = nameof(CreateCallback))]
    public partial void OfferCallback();

    [TransferToQueue("Support")]
    public partial void TransferDirectly();

    [Message("We'll call you back within the hour")]
    [SetCallback(Queue = "Support", DelayMinutes = 5)]
    [Disconnect]
    public partial void CreateCallback();
}

Flow Actions Reference

Message Actions

Play text-to-speech:

// Basic message
[Message("Hello")]
public partial void BasicMessage();

// With voice and language
[Message("Bonjour", Voice = "Celine", Language = "fr-FR")]
public partial void FrenchMessage();

// Dynamic from config
[Message(ConfigKey = "welcomeMessage")]
public partial void DynamicMessage();

Input Collection

Collect DTMF or voice input:

// DTMF input
[GetUserInput("Enter your account number", MaxDigits = 8, Timeout = 10)]
public partial void GetAccountNumber();

// Voice input
[GetUserInput("Please state your request", InputType = InputType.Voice, Timeout = 5)]
public partial void GetVoiceInput();

// With validation
[GetUserInput("Enter 4-digit PIN", MaxDigits = 4)]
[ValidateInput(Pattern = @"^\d{4}$", OnInvalid = nameof(InvalidPin))]
public partial void GetValidatedPin();

[Message("Invalid PIN format")]
public partial void InvalidPin();

Queue Operations

Transfer to queues:

// Basic transfer
[TransferToQueue("Support")]
public partial void Transfer();

// With timeout and priority
[TransferToQueue("VIPSupport", Timeout = 300, Priority = QueuePriority.High)]
public partial void TransferVip();

// Set working queue (for queue flow type)
[SetWorkingQueue("Support")]
public partial void SetQueue();

Lambda Integration

Invoke AWS Lambda:

// By ARN
[InvokeLambda(
    FunctionArn = "arn:aws:lambda:us-east-1:123456789:function:MyFunction",
    Timeout = 8
)]
public partial void CallLambda();

// By name (resolved automatically)
[InvokeLambda(FunctionName = "CustomerLookup")]
public partial void LookupCustomer();

// With error handling
[InvokeLambda("ExternalAPI", Timeout = 5)]
[OnSuccess(Target = nameof(ApiSuccess))]
[OnError(Target = nameof(ApiError))]
public partial void CallApi();

[Message("API call successful")]
public partial void ApiSuccess();

[Message("API call failed")]
public partial void ApiError();

Attribute Operations

Set and check contact attributes:

// Set attribute
[SetAttribute("Language", "en-US")]
public partial void SetLanguage();

// Set from Lambda response
[SetAttribute("CustomerName", Source = AttributeSource.LambdaResponse, Key = "name")]
public partial void SetNameFromLambda();

// Check attribute
[CheckAttribute("VIPCustomer", CompareType.Equals, "true")]
[OnTrue(Target = nameof(VipPath))]
[OnFalse(Target = nameof(StandardPath))]
public partial void CheckVipStatus();

[TransferToQueue("VIPSupport")]
public partial void VipPath();

[TransferToQueue("GeneralSupport")]
public partial void StandardPath();

Prompts

Play audio files:

// S3 audio file
[PlayPrompt("s3://my-bucket/prompts/welcome.wav")]
public partial void PlayWelcome();

// Dynamic prompt
[PlayPrompt(ConfigKey = "holidayGreeting")]
public partial void PlayHolidayGreeting();

// With interruption handling
[PlayPrompt("s3://my-bucket/long-message.wav", AllowInterrupt = true)]
[OnInterrupt(Target = nameof(Interrupted))]
public partial void PlayLongMessage();

[Message("How can I help you?")]
public partial void Interrupted();

Flow Control

Control flow execution:

// Disconnect
[Disconnect]
public partial void EndCall();

// Loop
[Loop(Target = nameof(GetInput), MaxIterations = 3)]
public partial void RetryInput();

// Wait
[Wait(Seconds = 2)]
public partial void PauseBriefly();

// Transfer to another flow
[TransferToFlow("EscalationFlow")]
public partial void Escalate();

// Transfer to phone number
[TransferToPhoneNumber("+18005551234")]
public partial void TransferExternal();

Flow Variables and Attributes

System Attributes

Amazon Connect provides built-in attributes:

[CheckAttribute("System.CustomerNumber", CompareType.StartsWith, "+1")]
[OnTrue(Target = nameof(UsCustomer))]
public partial void CheckCountry();

// Other system attributes:
// - System.CustomerNumber
// - System.DialedNumber
// - System.CustomerEndpointAddress
// - System.LanguageCode
// - System.QueueName
// - System.QueueArn

Custom Attributes

Set your own attributes:

[SetAttribute("Reason", "BillingInquiry")]
public partial void SetReason();

[SetAttribute("Priority", "High")]
public partial void SetPriority();

[InvokeLambda("CustomerLookup")]
// Lambda sets attributes automatically
public partial void LookupCustomer();

// Use attributes in conditions
[CheckAttribute("AccountStatus", CompareType.Equals, "Active")]
public partial void CheckStatus();

Testing Flows

Unit Testing

using Xunit;
using NickSoftware.Switchboard;

public class WelcomeFlowTests
{
    [Fact]
    public void WelcomeFlow_ShouldHaveCorrectStructure()
    {
        // Arrange
        var flow = new WelcomeFlow();

        // Act
        var actions = flow.GetActions();

        // Assert
        Assert.Equal(2, actions.Count);
        Assert.IsType<MessageAction>(actions[0]);
        Assert.IsType<TransferToQueueAction>(actions[1]);
    }

    [Fact]
    public void WelcomeFlow_ShouldTransferToCorrectQueue()
    {
        // Arrange
        var flow = new WelcomeFlow();

        // Act
        var transferAction = flow.GetActions()
            .OfType<TransferToQueueAction>()
            .First();

        // Assert
        Assert.Equal("GeneralSupport", transferAction.QueueName);
    }
}

Integration Testing

using Amazon.CDK.Assertions;

public class FlowStackTests
{
    [Fact]
    public void Stack_ShouldCreateContactFlow()
    {
        // Arrange
        var app = new App();
        var stack = new SwitchboardStack(app, "TestStack");

        // Act
        var template = Template.FromStack(stack);

        // Assert
        template.HasResourceProperties("AWS::Connect::ContactFlow", new Dictionary<string, object>
        {
            ["Name"] = "WelcomeFlow",
            ["Type"] = "CONTACT_FLOW"
        });
    }
}

Best Practices

1. Keep Flows Focused

// ✅ Good: Single purpose
[ContactFlow("Authentication")]
public partial class AuthenticationFlow
{
    // Only authentication logic
}

[ContactFlow("Routing")]
public partial class RoutingFlow
{
    // Only routing logic
}

// ❌ Bad: Too many responsibilities
[ContactFlow("MegaFlow")]
public partial class MegaFlow
{
    // Auth + routing + IVR + reporting + ...
}

2. Use Meaningful Names

// ✅ Good
[ContactFlow("VipCustomerSupport")]
public partial class VipCustomerSupportFlow
{
    [Message("Welcome valued customer")]
    public partial void WelcomeVipCustomer();
}

// ❌ Bad
[ContactFlow("Flow1")]
public partial class Flow1
{
    [Message("Welcome")]
    public partial void Action1();
}

3. Handle Errors Gracefully

[ContactFlow("RobustFlow")]
public partial class RobustFlow
{
    [InvokeLambda("ExternalAPI")]
    [OnSuccess(Target = nameof(Success))]
    [OnError(Target = nameof(HandleError))]
    public partial void CallApi();

    [Message("Request successful")]
    public partial void Success();

    [Message("We're experiencing technical difficulties")]
    [TransferToQueue("Support")]  // Don't just disconnect
    public partial void HandleError();
}

4. Provide Clear Messages

// ✅ Good: Clear, helpful
[Message("For sales, press 1. For support, press 2. For billing, press 3. To hear this again, press 9.")]
public partial void MainMenu();

// ❌ Bad: Unclear
[Message("Press a number")]
public partial void Menu();

5. Set Reasonable Timeouts

// ✅ Good: Appropriate timeouts
[GetUserInput("Enter account number", MaxDigits = 8, Timeout = 10)]
public partial void GetAccount();

// ❌ Bad: Too short
[GetUserInput("Enter account number", Timeout = 2)]  // Not enough time
public partial void GetAccountTooFast();

Next Steps

• Fluent Builders - Programmatic flow building
• Attribute-Based Flows - Declarative approach
• Flow Validation - Ensure flow correctness
• Flow Patterns - Common flow patterns

Related Examples

• Basic Call Center - Simple flow example
• IVR Menu - Menu system pattern
• Authentication - Customer verification