Skip to main content

Key Concepts

Understanding these core concepts will help you get the most out of Hublvu.

Chat

The Chat interface is the primary way to interact with Hublvu. Through natural conversation, you can:

  • Ask questions and get answers from your knowledge base
  • Execute procedures and automation
  • Run diagnostic commands
  • Access custom agent capabilities

The AI assistant has access to your organization's knowledge base, connected tools, and custom agents, allowing it to provide contextual, actionable responses.

Conversations

Each chat session is a conversation that maintains context. You can:

  • Start new conversations for different topics
  • Return to previous conversations from the sidebar
  • Branch a conversation to explore alternative approaches

Steps

Steps are the fundamental building block of automation in Hublvu. Both flows and guides are composed of steps.

Each step can be:

  • Fully automated — AI uses tools to execute the step without human involvement
  • Semi-automated — AI executes with human confirmation
  • Manual — Human performs the action and reports the result

How Steps Work

When a step executes:

  1. The AI model interprets what needs to be done
  2. It selects appropriate tools from connected outposts
  3. Tools execute and return results
  4. The model interprets results and determines next action

This approach brings simplicity (steps are easy to understand and compose) and robustness (the model adapts to variations in tool output).

Flows

Flows are automation for tasks that involve diagnosis and decision-making. They're ideal for:

  • Incident resolution
  • Troubleshooting scenarios
  • Any task where the path forward depends on what you discover

Flow Structure

Every flow contains:

  • Triage Steps — Investigation steps that gather information and narrow down the problem
  • Outcomes — The possible results of triage, each mapping to one or more root causes
  • Root Causes — The identified problems that the flow can address
  • Remediation Steps — Actions to resolve each root cause

Flow Lifecycle

StateDescription
DraftWork in progress, not available for general use
ActiveCurated and approved for team use
ArchivedDeprecated, kept for historical reference

Guides

Guides are automation for procedures and runbooks. They execute a sequence of steps to accomplish a task.

Use guides for:

  • Password rotation
  • Software upgrades
  • Network configuration
  • System integrations
  • Maintenance procedures
  • Onboarding processes
  • Any repeatable multi-step task

Like flows, guides can be entirely manual, entirely automated, or anywhere in between. The difference is structure: flows branch based on diagnosis, while guides execute steps in sequence.

Custom Agents

Custom agents extend the chat interface with specialized capabilities. They allow you to add new functionality that users can access through natural conversation.

Wrapping Guides Around Agents

A powerful pattern: wrap a guide around a custom agent. The guide defines what the agent can do (its steps and tools), and users invoke it by chatting. This makes it easy to add new capabilities without building custom integrations.

Example: Create a "Database Admin" agent with a guide that includes steps for common database operations. Users can then say "rotate the database credentials" and the agent executes the appropriate guide.

Insights

Insights are saved data explorations and visualizations. When you ask the AI to analyze data, you can save successful queries as insights for reuse.

Insights help you:

  • Track recurring metrics without rewriting queries
  • Share useful analyses with your team
  • Build a library of standard reports

Runs

A Run is an individual execution of a flow or guide. Each run tracks:

  • Start time and duration
  • Step-by-step progress
  • Tool executions and their results
  • Final outcome

The Runs section provides a unified view of all executions across flows and guides.

Ownership and Sharing

Flows, guides, and insights have ownership and sharing controls that determine who can access them.

Sharing Levels

LevelDescription
PersonalOnly you can view, edit, and run
TeamYour team members can view and run; team admins can edit
PublicAll users can view and run; only owner/admins can edit

Why Sharing Matters

  • Personal items let you develop and test without affecting others
  • Team sharing enables collaboration within your group
  • Public sharing scales best practices across the organization

When you create a flow or guide, it starts as Personal. As you refine it, you can share it more broadly. This progressive sharing model ensures quality while enabling collaboration.

Knowledge Base

The Knowledge Base contains your organization's documentation, indexed for AI-powered search.

TypeDescription
Institutional KnowledgeRunbooks, playbooks, and operational procedures
Product FilesVendor documentation, product manuals, and technical specs

When you upload documents, Hublvu processes them for semantic search. The AI assistant automatically references relevant knowledge when answering questions.

Outposts

Outposts are lightweight agents that execute tools in your environment. They connect to Hublvu using outbound-only connections, making them firewall-friendly.

Outposts enable Hublvu to:

  • Execute commands on your infrastructure
  • Query databases and APIs
  • Gather information from internal systems

MCP Servers

MCP (Model Context Protocol) Servers are tool containers that run on outposts. Each server provides a set of related tools—for example, a bash server for shell commands or a network server for connectivity tests.

Quick Actions

Quick Actions are pre-configured prompts that appear in the chat welcome panel. They provide one-click access to common operations, customized for your team's workflows.

Key Metrics

MTTR (Mean Time To Resolution)

The average time from start to completion. Hublvu tracks MTTR for flows to help you identify which automation delivers the fastest results.

Consistency Index

A reliability metric for flows, measuring how consistently a flow produces successful outcomes. Higher consistency indicates more predictable automation.

Success Rate

The percentage of runs that complete successfully without errors or manual intervention.

What's Next?