Quickstart
From start to finish, create a conversational AI application with NLX in minutes
Create an account
You'll need an account to begin building conversational AI with NLX (it's free!).
Choose template
Regardless of the template you choose, a full set of flows for greetings, goodbyes, error handling, and unknown inputs are provided along with instructional notes and required integration set up to help you get the application up and running.
The following templates are available during onboarding and are easy to implement:
🎧 Hands-free AI browsing: Optimized for voice, this template can navigate a website for users and fill out forms
📅 Appointment scheduler: Optimized for voice, this template automates booking and creating a calendar event
🌐 Museum plan-your-day: Optimized for chat, this template can handle common FAQs, give information on special exhibits, and address plan-your-day information
🔎Research agent: Optimized for email chat, this template sets up an async worker application that can receive requests for tasks that involve searching the web; generating text, images, or PDFs; and emailing when the request is complete
✨ MCP Server: Describe your application and expose it via Model Context Protocol to extend the capabilities to Claude Desktop, Cursor, or Postman
Explore
If using the Research agent template for a purely agentic app, follow the Async worker page.
Each NLX workspace comes equipped with a built-in NLP engine managed by NLX; no external setup required. You may connect a third-party provider if a phone channel demands it, or you prefer using a specific engine.
Once you've imported a template, you'll notice a series of Flows in your workspace resources. Flows define the structured path a conversation can take, guiding how the application responds to users, collects information, makes decisions, and connects to external services to complete tasks.
Each flow contains instructional tips and best practices for you to fully customize or refine your application. Review each flow to ensure the logic, generative prompts, and messaging follow your intended process, or follow along to re-create your own from scratch. Swap for any custom or managed API integrations you need to finalize as well as flows you may want to add or remove.
Select Resources in workspace menu > Choose Flows > Choose a flow to open the flow builder
Make adjustment to flows as needed
Click Save
Repeat for any additional flows your application will help automate.
🧠 Looking for more? See Intro to flows & variables
Deploy
If using the Research agent template for a purely agentic app, follow the Async worker page.
Now you'll update your conversational AI application with any changes made.
Select Applications from workspace menu > Choose application from template
Go to the application's Flows tab, and make sure any newly added flows are attached so your application can access them
Select Default behavior tab of application
Ensure the correct flow is assigned to each of the three default behaviors
Welcome,Fallback,Unknown
Select Channels tab of application > Expand API option > Click Edit channel
Click + Add domain under Whitelisted domains
Add both your webpage's URL and https://developers.nlx.ai if not already done so
Click Save
Click Deployment tab of application > Select Review & build
Wait for validation to complete > Select Create build
Click Deploy
🧠 Looking for more? See Add channels
Install
If using the Research agent template for a purely agentic app, follow the Async worker page.
To make your application available through your frontend and adjust the look, feel, and experience with theming, custom UI components, and Voice input support, install Touchpoint:
From Deployment tab of application, select a deployed build
Choose Deployment tab in modal > Expand API under Setup instructions
Scroll down and select Open Touchpoint configurator
You can try your application live and make adjustments to the theming
Once satisfied, copy Setup snippet > Install on your site
🧠 Looking for more? See Developer docs
Last updated