📖Reference: AI Chat

Introduction

AI Chat in Motion provides a conversational way to interact with your workspace. Instead of navigating menus or searching manually, you can ask natural-language questions, request summaries, or generate drafts directly from within Chat. AI Chat respects existing workspace permissions and is designed to be a safe, assistive layer that helps you move faster without losing control of your data.

This reference article explains how AI Chat works, what it can and cannot do, and the safeguards in place to ensure secure and predictable use.

At a Glance

• Core Functionality → Ask questions, generate drafts, and summarize content in natural language.

• Limitations → AI Chat cannot publish, delete, or make system-level changes without explicit user action.

• Safety & Control → Suggestions only; actions require user approval. AI Chat respects workspace RBAC.

---

Core functionalities

This section provides a detailed overview of the wide range of capabilities AI can perform, particularly within the Motion platform and integrated services.

Core AI Capabilities

• Basic Text Generation: Generate coherent and contextually relevant text using large language models (LLMs).

• Instruction Processing: Analyze and categorize user instructions for various tasks.

• Data Extraction: Build and execute JSONPath queries to extract specific information from structured data.

• Web Research: Conduct internet research using grounding-enabled models to provide accurate information.

• Search Optimization: Identify and optimize search queries for external search engines like Google.

• Person Research: Gather background information on individuals using specialized search tools.

Motion Platform AI Actions

• Create and Update Tasks: Create new tasks or update existing ones within projects and workspaces based on user instructions.

• Create and Update Projects: Manage projects by creating or updating attributes such as name, assignee, dates, status, and custom fields.

• Search Tasks and Projects: Search for tasks and projects based on various attributes within the workspace.

• Calendar Event Search: Retrieve and filter calendar events within specified date ranges (read-only).

• Document Search: Search and select documents, notes, and transcripts within user permissions.

Email Management AI Actions

• Write and Draft Emails: Generate email responses based on history or create drafts for review.

• Send Emails: Send emails or responses to specified addresses or incoming emails.

• Search and Label Emails: Search inbox messages, classify emails, apply or create labels (except label removal).

Collaboration Platform AI Actions

• Slack: Find users and channels, send messages to public channels or direct messages, create public channels, and manage user groups.

• Microsoft Teams: Find users by email or display name, send direct messages (no channel posting).

Salesforce AI Actions

• Salesforce Assistant: Retrieve schema, query, create, update records, and list available record types for Salesforce objects.

Limitations

• Docs: Can't edit document content or titles directly.

• Templates/events: Can't create/update project templates or calendar events.

• Files/audio: Can't delete items, upload/view files, or transcribe audio.

• Email: Can't send emails, messages, or contact people on your behalf.

• Scheduling: Can't schedule meetings or set reminders outside Motion tasks.

• Automation: Can't run recurring/automated actions without you triggering them.

• Voice/AI Employees: Can't use voice commands or operate "AI employees" skills.

Safety & Control

AI Chat is designed with built-in guardrails so that all generated output remains transparent, reversible, and under user control.

Key Behaviors:

• No silent actions → Drafts, suggestions, or insights appear in AI Chat first; they only become real tasks, projects, or docs if you choose to create or insert them.

• Reversible workflow → Any AI-generated draft can be edited or discarded before being committed to your workspace.

• Transparency → When an AI-generated suggestion references workspace data, the source (Doc, Task, Project) is displayed so you know where the context came from.

Practical Example:

• Asking AI Chat: “Summarize the latest Support Playbook doc.” → AI Chat returns a summary based only on the text you have permission to view.

• Asking AI Chat: “Create a new project for Marketing Launch.” → AI Chat suggests a draft outline, but nothing is published until you confirm creation.