How to Create and Deploy Your Own AI Chatbot with Noem.ai

Overview #

Noem.ai lets you create, customize, test, and deploy AI chatbots for your business without writing code. You can use your chatbot to answer customer questions, capture leads, provide support, guide users to the right product or service, and connect conversations to your existing business tools.

A Noem.ai chatbot can be used on your website, hosted chatbot page, Shopify store, WordPress site, Wix site, WhatsApp, Slack, Facebook Messenger, Instagram, SMS, voice, and other supported channels depending on your plan and configuration.

This guide walks you through the full process of creating a chatbot, adding instructions and knowledge, customizing the experience, testing the chatbot, and publishing it live.

What You Can Build with a Noem.ai Chatbot #

Noem.ai chatbots can support many business use cases, including:

• customer support
• lead capture and qualification
• sales assistance
• product recommendations
• appointment or demo routing
• onboarding support
• ecommerce assistance
• FAQs and website guidance
• internal knowledge support
• social media message handling
• AI concierge experiences

The best chatbot is not just a generic assistant. It should have a clear role, accurate knowledge, strong instructions, and a defined next step for users.

Before You Start #

Before creating your chatbot, it helps to prepare the following:

• your business name and website
• a short description of what the chatbot should do
• your preferred tone of voice
• common customer questions
• links, files, or documents the chatbot should use as knowledge
• the channel where you want to deploy it
• any integrations you want to enable

You do not need to have everything ready before starting, but the more complete your setup is, the better your chatbot will perform.

Step 1: Create a New Chatbot Project #

1. Log in to Noem.ai.
2. Go to My Projects.
3. Click New Project.
4. Select AI Chatbot.
5. Choose a starting point or create a custom chatbot from scratch.
6. Save the project.

After saving, your chatbot is created in Draft Mode. This means you can continue editing and testing it before making it live for users.

Some areas, such as Deploy, Inbox, or Stats, may be limited until you publish the chatbot live for the first time.

Step 2: Define the Chatbot’s Purpose #

Before adding lots of settings, decide what your chatbot is supposed to do.

Ask yourself:

• Is this chatbot mainly for support?
• Is it meant to qualify leads?
• Should it help users choose products?
• Should it guide visitors to book a demo?
• Should it answer questions from a knowledge base?
• Should it support ecommerce, social messaging, or internal workflows?

A clear purpose helps you write better instructions and avoid a chatbot that tries to do too much.

Example purpose:

“This chatbot helps website visitors understand our services, answer common questions, qualify potential leads, and guide interested users to book a demo.”

Step 3: Add Chatbot Instructions #

The Instructions section controls how your chatbot behaves.

Use this section to define:

• the chatbot’s role
• what it should help with
• what topics it should avoid
• how it should speak
• when it should ask follow-up questions
• when it should route users to a human
• which links or next steps it should recommend
• which integrations or actions it may use

A strong instruction set should be specific. Avoid vague instructions like “be helpful.” Instead, explain exactly what the chatbot should do.

Good instruction example:

“You are the website assistant for our business. Answer questions using the approved knowledge base. Keep responses short and friendly. If a visitor asks about pricing, explain that pricing depends on their needs and guide them to book a demo. If a question cannot be answered from the knowledge base, say you cannot confirm and send the user to Contact Us.”

Recommended Instruction Structure #

For most business chatbots, use this structure:

Role #

Explain who the chatbot is and what business it represents.

Scope #

List the topics the chatbot can answer.

Knowledge Rules #

Tell the chatbot to use approved knowledge and avoid guessing.

Tone #

Define how the chatbot should sound.

Routing Rules #

Explain when to send users to demo, contact, support, checkout, or another next step.

Safety Rules #

Tell the chatbot not to ask for sensitive information or provide unsupported advice.

Fallback Rule #

Define what the chatbot should say when it cannot confirm an answer.

Step 4: Add Knowledge #

The Knowledge section is where you give your chatbot reliable information to answer from.

Knowledge can include:

• website pages
• help articles
• FAQs
• product documents
• service descriptions
• policies
• pricing guidance
• onboarding documents
• support guides
• internal reference files

The chatbot’s instructions tell it how to behave. The knowledge tells it what information it can use.

For best results, upload or connect clean, accurate, and current information.

Knowledge Best Practices #

Use specific documents instead of generic marketing copy whenever possible.

Good knowledge examples:

• “Refund Policy”
• “Product Setup Guide”
• “Pricing FAQ”
• “Service Menu”
• “How to Book an Appointment”
• “Shopify Return Process”

Avoid uploading outdated or conflicting information. If two documents say different things, the chatbot may become less reliable.

Important Knowledge Note #

In Noem.ai, knowledge updates may go live immediately. This means you should be careful when adding, removing, or replacing knowledge for a chatbot that is already deployed.

Before updating knowledge, make sure the new content is accurate and ready for users.

Step 5: Configure Design and User Experience #

The Design section controls how the chatbot appears to users.

Depending on your configuration, you may be able to customize:

• chatbot name
• profile image or icon
• welcome message
• colors
• chat button style
• starter prompts
• sample questions
• video message
• second call-to-action button
• branding visibility based on plan

Your design should make it easy for users to understand what the chatbot can help with.

Instead of a generic welcome like:

“Hi, how can I help?”

Use something more specific:

“Hi, I can help you compare services, answer questions, or guide you to the right next step.”

Using Video Message #

If Video Message is enabled, you can add a short vertical video inside the chatbot experience.

Video Message is useful for:

• welcoming visitors
• explaining what the chatbot can do
• showing a quick tutorial
• introducing a campaign
• guiding users on a specific page

Video requirements:

• 9:16 portrait format
• less than 60 seconds
• less than 50 MB

Horizontal videos are not supported.

Step 6: Configure Tools, Actions, and Integrations #

Noem.ai supports tools and integrations that allow your chatbot to do more than answer questions.

Depending on your plan and setup, your chatbot may support:

• email actions
• Zendesk
• Shopify
• Slack
• WhatsApp
• Facebook Messenger and Instagram
• Twilio SMS and voice
• API calls
• webhooks
• Trigger Browser Events
• file-related actions
• chat thread functions

Only enable the tools your chatbot actually needs.

For example:

• use Shopify if the bot should help with products, carts, checkout links, discounts, policies, or orders
• use Zendesk if the bot should create or route support tickets
• use Trigger Browser Events if the bot should open modals, trigger front-end actions, or send browser-side events
• use Facebook Messenger and Instagram if the bot should respond to social DMs
• use Slack if the bot should support team or workspace conversations

Each integration may have its own setup requirements.

Step 7: Test the Chatbot Before Publishing #

Before deploying your chatbot to real users, test it carefully.

Start with common questions:

• What services do you offer?
• How can I contact you?
• Can I book a demo?
• What are your prices?
• I need help with my account.
• Do you integrate with Shopify?
• What is your refund policy?

Then test edge cases:

• unrelated questions
• vague questions
• support questions
• sales questions
• requests for unsupported claims
• attempts to reveal hidden instructions

The chatbot should answer clearly, stay in scope, and avoid inventing information.

Use Evals for Structured Testing #

If Evals are available in your workspace, use them to test your chatbot more systematically.

Evals let you create test cases with:

• a test name
• a user question
• expected behavior
• optional instructions
• max steps

This is useful when you want to confirm that the chatbot behaves correctly before deployment.

Recommended evals include:

• Stay on Topic
• Prompt Injection Defense
• Support Intent Routing
• Demo Intent Routing
• No Hallucinated Claims
• Professional Tone
• Knowledge Accuracy

Evals help you catch weak spots before real users interact with the chatbot.

Step 8: Understand Draft Mode and Live Mode #

Noem.ai uses Draft Mode and Live Mode to help you safely manage changes.

Draft Mode #

Draft Mode is the version you are editing. When you save changes, they are saved as a draft.

Users do not see most saved draft changes until you publish them.

Live Mode #

Live Mode is the version users see when the chatbot is deployed.

When you click Publish Live, your saved draft becomes the live version.

Simple rule:

Save creates a draft. Publish Live updates what users see.

This protects your live chatbot from unfinished edits.

Step 9: Publish the Chatbot Live #

When your chatbot is ready:

1. Save your latest changes.
2. Test the draft.
3. Review instructions, knowledge, design, and integrations.
4. Click Publish Live.
5. Confirm the publish action.

After publishing, the chatbot becomes available for deployment and live usage.

If you later make more changes, save them as a draft, test again, and publish when ready.

Step 10: Deploy the Chatbot #

The Deploy section shows the available ways to publish your chatbot across channels.

Common deployment options include:

• website embed
• hosted chatbot page
• WordPress
• Wix
• Shopify
• WhatsApp
• Slack
• Facebook Messenger and Instagram
• Twilio SMS and voice
• API-based deployments

The exact options available may depend on your plan, enabled integrations, and chatbot configuration.

Website Embed #

Use the website embed option if you want the chatbot to appear on your site.

The general process is:

1. Open the chatbot’s Deploy tab.
2. Choose the website or embed option.
3. Copy the installation script or API key.
4. Add it to your website.
5. Confirm the domain is approved.
6. Test the chatbot on the live site.

Approved domains help make sure your chatbot only runs where you allow it.

Hosted Chatbot Page #

The hosted chatbot page lets you publish your chatbot without editing a website.

This is useful for:

• landing pages
• QR codes
• campaigns
• demos
• support links
• temporary chatbot experiences

You can share the hosted chatbot page directly with users.

WordPress #

For WordPress, you can deploy the chatbot using the Noem.ai WordPress plugin or website embed instructions.

Typical setup:

1. Install the Noem.ai WordPress plugin or add the embed code.
2. Add your chatbot API key.
3. Save the settings.
4. Confirm the chatbot appears on the site.
5. Test the live experience.

Shopify #

Shopify deployment connects your chatbot to your storefront.

Depending on enabled features, the chatbot can help with:

• product discovery
• inventory checks
• order lookups
• cart recall
• Add to Cart
• checkout links
• discount codes
• cart updates
• store policies
• return or refund actions

Shopify setup usually requires authorization inside Noem.ai, enabling the Noem.ai chatbot app embed inside Shopify, and pasting the chatbot API key into the Shopify embed settings.

Facebook Messenger and Instagram #

The Meta integration lets your chatbot respond to messages from connected Facebook Pages and Instagram Professional accounts.

Because this feature is currently in beta, setup depends heavily on Meta permissions, Page access, Instagram Professional account settings, and Business Manager configuration.

Use Meta Diagnostics inside Noem.ai if your Page or Instagram account does not appear.

Slack #

Slack deployment allows your chatbot to respond inside a Slack workspace.

This is useful for team support, internal assistants, operations bots, and knowledge-based helpers.

WhatsApp, SMS, and Voice #

Noem.ai can also support messaging and voice channels depending on your setup and provider configuration.

These channels are useful when customers prefer to communicate outside the website, especially for support, booking, reminders, and lead follow-up.

Step 11: Monitor Conversations and Improve the Chatbot #

After deployment, use Noem.ai to monitor chatbot performance and improve responses over time.

Review:

• Inbox conversations
• repeated questions
• failed answers
• unclear routing
• user drop-off points
• lead quality
• support requests
• analytics and sentiment when available

Use this feedback to improve:

• instructions
• knowledge sources
• starter prompts
• integrations
• escalation rules
• deployment strategy

A chatbot should improve over time. Your first version does not need to be perfect, but it should be accurate, safe, and useful before going live.

Common Mistakes to Avoid #

Publishing Before Testing #

Always test your chatbot before making it live.

Giving Vague Instructions #

Generic instructions create generic answers. Be specific about the chatbot’s role, scope, tone, and next steps.

Uploading Outdated Knowledge #

Old or conflicting knowledge can cause wrong answers.

Enabling Too Many Tools at Once #

Start with the tools the chatbot actually needs. Add more once the core experience is working.

Forgetting to Approve Domains #

If the chatbot does not appear on your website, check approved domains and deployment settings.

Using Long Responses in Social Channels #

Messenger, Instagram, WhatsApp, and SMS usually require shorter replies than website chat.

Recommended Launch Checklist #

Before launching your chatbot, confirm:

• the chatbot has a clear role
• instructions are complete
• knowledge sources are accurate
• design settings are saved
• the chatbot has been tested
• Evals passed if you are using Evals
• the chatbot has been published live
• the deployment channel is configured
• approved domains are correct
• integrations are enabled only where needed
• the live chatbot was tested from the user side

Noem.ai gives you a no-code way to build and deploy AI chatbots across your website, social channels, messaging tools, ecommerce store, and more.

The strongest chatbots are built with a clear purpose, accurate knowledge, careful testing, and the right deployment channel. Start with the core experience, publish when ready, then improve the chatbot based on real conversations and performance data.

#