Quick Start | LanSphere User Guide

Quick Start

This guide walks you through the minimal loop — create → orchestrate → debug → save → publish. It is for users who have logged in and are ready to build their first app. Before reading, make sure your role is Developer or above; Viewers have no creation entry.

The New Project Entry

After logging in, click “New Project” below “Account” in the left menu, then choose an app type in the pop-up: Workflow, Chatflow, or Agent.

Choosing Among the Three Types

What you need Recommended type
Automated jobs that run on a schedule or on demand — take input, produce output, no user interaction along the way Workflow
Conversational apps with multi-turn Q&A, such as a knowledge Q&A assistant or a business inquiry assistant Chatflow
Complex tasks where the LLM plans autonomously and flexibly calls plugins, Workflows, Knowledge Bases, MCP Services, and more Agent

Tip: Workflows cannot be published as Lansenger AI Assistants. If you want your app to reach end users as a Lansenger AI Assistant, choose Chatflow or Agent.

A simple rule of thumb:

  • Fixed flow, clear inputs and outputs → Workflow;
  • The flow advances dynamically with the conversation, but the overall path is fairly predictable → Chatflow;
  • Open-ended tasks that need the LLM to plan on its own and orchestrate tools flexibly → Agent.

Creation Methods

After choosing a type, there are three ways to create the app:

  1. Blank creation: Orchestrate from scratch on the canvas — for well-defined needs that call for full customization.
  2. Create from the Template Store: Build quickly on official templates. The Workflow home page shows 5 random templates; hover and click “Build Now” to start orchestrating from a template, click “Refresh” for another batch, or “More” to enter the Template Store and filter templates by name, type, and tags.
  3. Import a DSL file: Restore an app from an existing DSL file. Legacy Agent DSL files are rejected on import, and the page will direct you to the Agent module instead.

App Name, Description, and Avatar

When creating an app, fill in its basic info:

Field Rules
App name Required, 1–30 characters; Chinese characters, upper/lowercase letters, digits, and hyphens only
App description Optional, up to 300 characters
Avatar Custom upload supported

The same rules apply to the Agent creation pop-up.

Tip: The app name and description appear in the App Square and the Lansenger AI Assistant Marketplace. Use a name and description that clearly state the purpose so teammates can find and use the app.

Minimal-Loop Example: Create and Publish a Workflow

Let’s walk the full path from creation to publishing with “create a simple Workflow and publish it to the App Square”:

  1. Create the app and enter the orchestration canvas;
  2. Add and configure nodes on the canvas;
  3. Debug a run to verify;
  4. Save the draft;
  5. Publish to the App Square from the “Save” drop-down.

Step 1: Create the App

Click “New Project,” choose “Workflow” in the pop-up, pick blank creation, fill in the app name and description, and confirm to enter the orchestration canvas.

Step 2: Add Nodes on the Canvas

  • The node list is on the left of the canvas; it can be expanded/collapsed and searched by fuzzy name match.
  • Nodes are added by clicking only — drag-and-drop is not supported. Click a node’s name to add it to the canvas.
  • The “Input” node is globally unique: once it is on the canvas, it is hidden from the list automatically.
  • To call a plugin, a published Workflow, an MCP Service, or a Custom Tool, pick the corresponding tool-type node in the node list, then choose the specific tool in the pop-up tool list.
  • After adding nodes, connect them in business order and finish parameter settings in each node’s configuration panel.

Step 3: Debug

  • The canvas supports single-node debugging and end-to-end debugging: the former verifies each node’s configuration one by one; the latter verifies the complete flow.
  • While debugging, you can view each node’s input, output, and error info; past runs are in “Debug History.”
  • The platform provides a Variable Inspector that automatically checks variable definitions, type matching, and reference validity, helping you catch configuration issues before publishing.
  • The bottom of the canvas offers zoom in/out, undo/redo, and change history; undo/redo covers adding/removing nodes and edits inside nodes.

Tip: On publish, the platform automatically runs a Checklist to validate configuration completeness and logic. We recommend clearing all issues found during debugging first.

Step 4: Save

When orchestration and debugging are done, click “Save” at the top right of the canvas. The app enters the draft state (visible only to you).

Note: A draft cannot be published to the App Square. Save the latest content before publishing.

Step 5: Publish to the App Square

Hovering over the “Save” button always shows three options:

  • Save: Save current content as a draft.
  • Publish to App Square: Saves the latest content and lists the app in the App Square, visible and usable by everyone on the platform (including Viewers).
  • Publish as Component: Visible for the Workflow type only. After publishing, it becomes a Workflow component that syncs automatically to “Collaboration Components → Workflow” for other apps to call, and is not listed in the App Square.

Choose “Publish to App Square” to finish publishing. You can then see and use the app under “App Square” in the left menu.

Note: “Publish as Component” and “Publish to App Square” are mutually exclusive — an app can only take one of the two.

The Four App States

An app goes through four states from creation to going live:

State Description
Created (unsaved) Created but not saved yet; content may be lost
Saved (draft) A saved draft, visible only to you
Published Published to the App Square / as a component / to Lansenger — the three channels are independent
WebApp live Served externally through a web link; can coexist with the three publishing channels

Tip: This example showed “Publish to App Square.” To also open a web link or API access, or to publish a conversational app as a Lansenger AI Assistant, configure it in the “Publishing Channels” tab on the app details page — see “Publishing Channels.”

Next Steps

With the minimal loop done, dive deeper as needed:

  • For more complex automation, see “Workflow” and “Node Reference.”
  • For multi-turn conversational apps, see “Chatflow.”
  • To let the LLM orchestrate tools on its own, see “Agent.”
  • For API access or publishing as a Lansenger AI Assistant, see “Publishing Channels.”