Creating Your First Chatbot

Creating a chatbot in Aimogen is a configuration task, not a coding task. You are not “training” a bot and you are not deploying a generic widget. You are defining a chatbot instance with a role, a brain (model), and rules about where and how it appears.

This guide walks through the minimum required steps to get a working chatbot online.

---

Step 1: Make Sure an AI Provider Is Connected #

Before creating a chatbot, at least one AI provider must be configured and working.

This means:

• an API key is entered in Aimogen → Settings → API Keys
• models are visible in model dropdowns

If no models are available, the chatbot cannot function.

---

Step 2: Open the Chatbot Manager #

In the WordPress admin, go to:

Aimogen → Chatbots

This is the central place where all chatbot instances are created and managed.

You can have one chatbot or many. There is no global “default bot”.

---

Step 3: Create a New Chatbot #

Click Add New Chatbot.

You are now creating a single chatbot instance. Everything you configure here applies only to this bot.

Nothing is live yet.

---

Step 4: Define the Chatbot Persona #

The persona defines who the chatbot is.

At minimum, you should set:

• a name (visible to users)
• a role or description (used as system context)

Examples:

• “Customer Support Assistant for an online store”
• “Helpful documentation assistant for developers”
• “Sales assistant answering pre-purchase questions”

This text becomes the system prompt that shapes how the chatbot behaves.

The clearer the persona, the better the responses.

---

Step 5: Select Provider and Model #

Choose:

• an AI provider
• a model for this chatbot

This choice affects:

• response quality
• speed
• cost
• streaming capability

This selection is per chatbot. Changing it does not affect other bots or other Aimogen features.

For a first chatbot, choose a:

• fast
• stable
• chat-oriented model

---

Step 6: Basic Behavior Settings #

At this stage, you can leave advanced options untouched.

The defaults are enough to:

• accept user messages
• generate AI responses
• maintain conversation context

You do not need workflows, triggers, or embeddings to get started.

---

Step 7: Choose Where the Chatbot Appears #

Decide where the chatbot should be visible.

Common first setups:

• site-wide on the front end
• specific pages only
• specific post types

If you enable it globally, the chatbot will appear across the site according to its placement settings.

Nothing appears until visibility rules allow it.

---

Step 8: Save the Chatbot #

Save the chatbot configuration.

At this point:

• the chatbot exists
• it is fully configured
• it is ready to answer messages

If visibility rules allow it, it is now live.

---

Step 9: Test the Chatbot #

Visit a page where the chatbot should appear and start a conversation.

Test:

• response quality
• tone and role adherence
• speed
• basic understanding

If responses feel wrong, adjust:

• the persona text
• the model
• the system instructions

Small changes here make big differences.

---

What You Do Not Need for the First Chatbot #

You do not need:

• embeddings
• document uploads
• workflows
• triggers
• lead collection
• real-time voice
• multiple personas
• fallback logic

All of those are optional and can be added later.

---

Safe Iteration Model #

Chatbots are easy to adjust:

• editing a chatbot does not break others
• changes apply immediately
• nothing is permanent
• you can duplicate chatbots to experiment

This encourages experimentation without risk.

---

Common First-Time Mistakes #

• leaving the persona vague
• using a slow or expensive model unnecessarily
• enabling the chatbot everywhere before testing
• assuming the chatbot “knows” your site without embeddings
• over-configuring before basic testing

Start simple.

---

Summary #

To create your first chatbot in Aimogen:

• connect an AI provider
• create a chatbot instance
• define a clear persona
• choose a model
• set visibility
• save and test

That’s it.

A working chatbot requires configuration, not complexity. Everything else—knowledge, rules, workflows, real-time features—can be layered on later, once the fundamentals are solid.