Assistants

Assistants are the core of TeamWeb AI. Each assistant is an AI agent with its own personality, behavior, and capabilities. You can create multiple assistants for different purposes — a customer support agent, a content writer, a research assistant, and more.

Creating an Assistant

There are several ways to create an assistant:

From scratch — Click New Assistant on the Assistants page and fill in the form
From a template — Click Templates to pick a pre-configured starter (customer support, research, writing, Q&A)
By duplicating — Open an existing assistant and click Duplicate to clone it with all settings
By importing — Click Import to upload a previously exported JSON configuration

For the from-scratch route, you’ll need to provide:

Name — A descriptive name for the assistant (e.g. “Content Writer”, “Support Agent”)
Project — Optionally link the assistant to a project for access to knowledge and guidelines
Mode — How the assistant should behave in conversations (see Modes)
Personality — The personality prompt that defines the assistant’s character and behavior

See Templates & Import/Export for details on templates, duplicating, and sharing assistant configurations.

The Personality Prompt

The personality prompt is the most important part of your assistant configuration. This is the instruction set that tells the AI who it is and how to behave. Write it in natural language, describing:

The assistant’s role and expertise
How it should communicate (tone, style, formality)
What topics it should focus on
Any constraints or boundaries

You can use template tags in your personality prompt to dynamically insert context:

Tag	Description
`{assistant_name}`	The assistant’s name
`{user_name}`	The current user’s name
`{user_role}`	The current user’s role
`{user_bio}`	The current user’s bio
`{tools}`	Available tool descriptions
`{skills}`	Enabled skill descriptions
`{knowledge}`	Knowledge source information
`{guidelines}`	Content guidelines
`{project}`	Project context
`{deliverables}`	Recent deliverables
`{user_memories}`	What the assistant remembers about the current person
`{mode}`	Mode-specific instructions

Any section tags you don’t include in your prompt will be automatically appended, so your assistant always has access to all relevant context.

Technical Details

System prompt assembly — The system prompt sent to the LLM is assembled from multiple parts: the assistant’s personality prompt, tool descriptions and usage instructions, mode-specific behavior rules, the available skills catalog, project context, core knowledge, content guidelines, available agents for delegation, recent deliverables, user context (name, role, bio), and user memories. Each of these sections corresponds to a template tag.

Template tag resolution — When the personality prompt contains a tag like {tools} or {skills}, TeamWeb AI replaces it with the rendered content for that section at that exact position. If a section tag is not present in the personality prompt, its content is automatically appended at the end. This means the assistant always receives full context regardless of how you write the prompt — the tags just let you control placement.

System Instructions override — The System Instructions field on the Prompts tab replaces the entire prompt assembly pipeline. When set, it is used as-is with no auto-appended sections. This gives full control over prompt structure for advanced use cases, but means you must include any section tags you need explicitly.

Model override — Each assistant can specify a different LLM model than the global default (configured in Administration > Settings). When set, that model is used for all LLM calls involving this assistant. This allows you to use a more capable model for complex tasks and a faster or cheaper model for simpler ones.

Greeting Message

You can set a greeting message that the assistant automatically sends when a new conversation starts. This is a great way to introduce the assistant and set expectations. The greeting supports the same template tags as the personality prompt.

Notes Tab

The Notes tab on the assistant detail page shows all notes the assistant has saved — both general assistant notes and per-user memories.

Use the filter pills to switch between views:

Filter	Shows
All	Every note (general and per-user)
General	Only general assistant notes (not tied to a person)
Per-User	Only user-scoped memories — enter a person’s name to search

Per-user notes display a name badge showing which person the memory is about. When filtered to a specific person, a Clear All button lets you bulk-delete all memories for that person.