Creating AI Assistants

Creating an AI Assistant in Aimogen means defining a persistent, reusable AI entity that encapsulates behavior, tools, and optional knowledge. Once created, the assistant can be used anywhere Aimogen accepts an AI engine: chatbots, OmniBlocks, content generators, and bulk workflows.

You are not creating a prompt. You are creating a long-lived AI component.

---

What You Create When You Create an Assistant #

An assistant consists of:

• instructions (core behavior)
• a selected AI model
• enabled tools (optional)
• attached files or knowledge (optional)
• persistent state handling

This definition is saved and reused. Every time the assistant is called, it behaves according to this configuration.

---

Where Assistants Are Created #

AI Assistants are created in the Aimogen admin area.

Path:
Aimogen → AI Assistants

From there, you can create, edit, duplicate, or remove assistants.

Assistants are global objects, not tied to a single chatbot or workflow.

---

Step 1: Define the Assistant’s Purpose #

Before touching settings, define one clear responsibility.

Good examples:

• documentation assistant
• WooCommerce support assistant
• SEO content assistant
• internal editorial helper
• product analysis assistant

Bad examples:

• “does everything”
• “general AI”
• “random helper”

Assistants work best when their scope is narrow.

---

Step 2: Write Assistant Instructions #

This is the assistant’s core system behavior.

Instructions define:

• what the assistant does
• what it avoids
• tone and style
• boundaries and constraints
• expected output format (if relevant)

These instructions are always active and do not need to be repeated in prompts.

They should be written as internal system rules, not as user messages.

Once saved, these instructions apply everywhere the assistant is used.

---

Step 3: Choose the Model #

Select the AI model the assistant will use.

This choice affects:

• reasoning ability
• speed and latency
• cost
• tool compatibility

The model is bound to the assistant. Changing it affects all usages of that assistant.

Choose models based on task type, not popularity.

---

Step 4: Enable Tools (Optional) #

Assistants can have tools enabled.

Depending on provider support, tools may include:

• file access
• file search
• code execution
• structured tool calls

If a tool is not enabled here, the assistant cannot use it later, even if prompted.

Only enable tools the assistant actually needs.

---

Step 5: Attach Files or Knowledge (Optional) #

You can attach files to an assistant to give it reference context.

These files:

• are available to the assistant automatically
• do not need reinjection on every call
• act as working knowledge, not training data

This is ideal for:

• documentation assistants
• policy-aware bots
• product knowledge assistants

Files can be updated without recreating the assistant.

---

Step 6: Save and Register the Assistant #

Once saved:

• the assistant becomes selectable throughout Aimogen
• it can replace raw models in chatbots
• it can be used as an AI step in OmniBlocks
• it can be tested in the backend Playground

No publishing step is required.

---

Using the Assistant After Creation #

After creation, the assistant can be used:

• as the engine behind a chatbot
• inside OmniBlocks AI steps
• in bulk content generators
• in single post generators
• in backend testing

The assistant behaves consistently everywhere.

---

Assistants vs Personas #

Important distinction:

• Assistant instructions → core reasoning and capabilities
• Chatbot personas → presentation, tone layering, role framing

When an assistant is used in a chatbot:

• assistant instructions run first
• persona rules are layered on top
• workflows and triggers still apply

They do not conflict if designed correctly.

---

Editing an Assistant #

When you edit an assistant:

• changes apply immediately
• all usages are affected
• no migration is needed

This makes assistants powerful but also risky on production sites.

Avoid editing assistants blindly if they are widely used.

---

Versioning Strategy (Recommended) #

For production setups:

• duplicate assistants instead of editing live ones
• test changes in the backend Playground
• swap assistants only after validation

Treat assistants like shared services.

---

What Creating an Assistant Does Not Do #

Creating an assistant does not:

• fine-tune a model
• train on your site data
• create frontend UI
• expose it publicly
• bypass usage limits
• store permanent memory

It defines behavior and capability, nothing more.

---

Common Mistakes #

• overloading assistants with too many responsibilities
• writing instructions like user prompts
• enabling unnecessary tools
• attaching irrelevant files
• editing live assistants without testing
• assuming assistants replace workflows

Assistants simplify reasoning, not system design.

---

Best Practices #

Define assistants narrowly, write explicit system instructions, enable only required tools, reuse assistants across features, and combine them with OmniBlocks for structure. Let assistants reason and transform; let workflows control execution.

---

Summary #

Creating AI Assistants in Aimogen means defining persistent, reusable AI entities with fixed behavior, tools, and optional knowledge. Assistants sit above raw models and below workflows, enabling consistent reasoning across chatbots, OmniBlocks, and content pipelines. Designed carefully, assistants reduce prompt complexity, increase reliability, and scale cleanly across complex AI systems.