AUMO User Guide

AUMO is an enterprise AI platform that allows users to connect company documents and systems to create AI agents. These agents provide responses based on the connected data and are deployed to users via a Chat interface. The platform is structured into four primary areas:

• Organization: Global settings and configurations applicable to the entire company.
• Workspace: Settings and assets belonging to a specific team or project.
• Chat: The daily interface for interacting with AI agents.
• Agents: The area for creating and managing AI agents.

The Model Context Protocol (MCP) allows AI agents to access external tools. The integration process follows a three-tier hierarchy:

1. Organization Registration: A server is added to the organization's approved catalog.
2. Workspace Installation: The registered server is enabled for a specific workspace and configured.
3. Agent Assignment: The installed MCP is assigned to a specific AI Agent.

AUMO is an enterprise AI platform that allows users to connect company documents and systems to create AI agents. These agents provide responses based on the connected data and are deployed to users via a Chat interface. The platform is divided into four primary conceptual areas:

• Organization: Settings and configurations that apply to the entire company.
• Workspace: Resources, settings, and data belonging to a specific team or project.
• Chat: The daily interface for interacting with AI agents.
• Agents: The tools used to build and manage the AI agents.

Aumo organizes workflow nodes into the following functional groups:

• Input and Output (Entrada e saída): Mandatory endpoints for user forms and final results.
• LLM: Invokes AI models using prompts constructed from the workflow flow.
• Documents: Used for extracting text from files, querying Collections, and generating documents.
• Logic: Handles branching (diverting), looping over lists, and selecting between results.
• Integration: Connects to external systems via HTTP, email, or installed integrations.
• Code: Executes custom Python or Node.js scripts for specialized tasks not covered by built-in nodes.

Aumo provides four specialized nodes for handling documents within a workflow:

1. File Extractor (Extrator de arquivos): Converts files received at the Input into plain text. This is typically the first node in a document workflow.
2. CPF Extractor (Extrator de CPF): Locates, validates, and extracts CPF and CNPJ identifiers from previously processed text.
3. Vector Database (Banco de dados vetorial): Searches workspace Collections and returns relevant snippets, enabling Retrieval-Augmented Generation (RAG) on the canvas.
4. File Generator (Gerador de arquivos): Produces a final document for download in Markdown, DOCX, or PDF formats based on the workflow's output.

Every execution of a Workflow generates a 'run', which serves as a detailed audit trail of the process. Unlike Agents, which only provide a final response, Workflows record the entire path to the result. A run captures:

• Input: The values submitted via the form.
• Intermediate Steps: The input and output of every node in the order they were executed.
• Output: The final result displayed to the user.
• Errors: The specific node where the execution stopped and the timing of the failure.

The Chat module is the non-administrative side of the platform for workspace participants. It is divided into two primary interaction areas:

1. Chat: Used for open, message-by-message conversations with a selected Agent.
2. Agentes (Agents): Despite the name, this area is used to execute published Workflows via a form to achieve a specific result.

Users can switch between the Chat module and the Admin module without re-authenticating.

Quotas define monthly consumption ceilings for AI models. If no ceiling is configured, the scope consumes without limits. Quotas are managed in two primary locations: Organization > Quotas and Workspace > Quotas.

Quota Scopes:

• General: The total ceiling for the entire scope.
• Providers: Individual ceilings per provider (useful for managing high-cost providers).
• Users: Consumption limits per person.
• Project Keys: (Workspace scope only) Limits for programmatic integrations to prevent uncontrolled consumption.

Priority Rule: Organization quotas take priority over Workspace quotas. If the Organization ceiling is exhausted, all workspaces are blocked regardless of their individual remaining limits.