Skip to main content
Back to BotKit SDK Overview This guide walks through an automated, production-ready BotKit deployment on AWS using an Auto Scaling Group, Application Load Balancer, and optional ElastiCache for Redis.

​
BotKit Overview

BotKit extends the Platform’s conversational AI by letting you:
  • Implement custom business logic in conversation flows via webhooks.
  • Integrate with unsupported messaging or voice channels.
  • Handle conversation lifecycle events for complex workflows.
  • Connect to external services not covered by the standard API node.
Example use cases:
Use CaseDescription
Channel integrationsAdd support for custom mobile SDKs or IoT devices
Custom NLU modelsInject external NLP models tuned on proprietary data
Conversation lifecycle handlersTrigger custom logic on events like conversation start/close
Non-standard integrationsConnect via webhooks to on-premise or legacy systems
Custom analyticsCapture KPIs not available out-of-the-box; send to third-party analytics
References: Using the BotKit SDK · SDK Events · SDK Functions

​
Deployment Steps

​
Step 1 — Create a Golden Image

  1. Launch a new EC2 instance with one of the following base OS:
    • Amazon Linux 2
    • CentOS / Red Hat 7.x or above
    Recommended instance type: m5.xlarge (4 vCPU, 16 GB RAM, 20 GB OS disk, 50 GB data disk)
  2. Update OS packages: 
    yum update
  3. Install Node.js (supported version).
  4. Install Nginx (latest).
  5. Verify both installations: 
    node -v
nginx --version
  6. Create a deploy script to pull the build from your central repository and extract it to the data disk.
  7. Create startup scripts to bring up BotKit and Nginx services.
  8. Configure Nginx to listen on port 80 and proxy to the BotKit instance.
  9. Create the Golden Image (AMI) from this instance.

​
Step 2 — Set Up an Application Load Balancer

  1. Create an external (internet-facing) Application Load Balancer (ALB).
  2. Add a listener on port 443 with an SSL certificate (use ACM).
  3. Create a target group targeting backend instances on port 80 and assign it to the listener.

​
Step 3 — Configure DNS

  1. Create an A or CNAME record in your external domain pointing to the ALB.

​
Step 4 — Set Up ElastiCache for Redis (Optional)

If BotKit uses Redis for async payload storage:
  1. Launch an AWS ElastiCache cluster for Redis with appropriate capacity.

​
Step 5 — Create a Launch Template

  1. Create a Launch Template using the Golden AMI.
  2. Configure VPC settings: subnets, security groups, IAM role.
  3. In User Data, invoke the deploy script to pull the latest build and start services.
  4. Set this version as the default.

​
Step 6 — Configure an Auto Scaling Group

  1. Create an Auto Scaling Group (ASG) using the Launch Template.
  2. Configure all required settings per your internal policies.
  3. Attach the target group so instances register with the ALB automatically.
  4. Enable SNS notifications for instance launch/termination events.
  5. Set a CPU utilization threshold for scale-out per your internal policies.

​
Step 7 — Maintain a Configuration Repository

  1. Store all BotKit and Nginx configuration files in a central repository (Git, S3, etc.) separate from the application code.

​
Step 8 — Implement a CI/CD Pipeline

Use Jenkins or equivalent automation:
  1. Check out the BotKit codebase.
  2. Check out configuration files from the configuration repository.
  3. (Optional) Run npm install for dependencies.
  4. Create an archive and push it to S3 or Artifactory.
  5. Invoke the deploy script using a rolling deployment:
    • Remove one server from the ALB.
    • Deploy the build and restart services.
    • Reattach the server to the ALB.
    • Repeat for remaining servers.
  6. (Optional) Add a vulnerability scan step and gate the pipeline on the scan result.

​
Security Patching

  1. Apply patches to the Golden Image.
  2. Launch a standalone server from the updated image and validate the application.
  3. On successful validation, launch new servers via the ASG and terminate the old ones.