C3 AI Documentation Home

Overview

C3 Code is an AI-powered application builder within the C3 Agentic AI Platform. You use natural language, file attachments, and visual editors to create, iterate on, and deploy enterprise applications without needing to write code from scratch.

What you can build

Common application types include:

  • Custom dashboards and monitoring views built on existing C3 AI vertical application data, such as asset health summaries, supply chain risk visualizations, and demand forecast comparisons.
  • Operational monitoring tools that surface real-time data from C3 Types, such as alert dashboards and equipment health trackers.
  • Analysis and planning applications, such as failure trend analysis tools and what-if scenario modelers.

You can develop an application from scratch, or customize a pre-built template using prompts and visual configuration.

How C3 Code works

C3 Code follows a prompt-and-iterate loop:

  1. Describe your application: Enter a natural language description of what you want to build. Optionally attach CSV files, images, or specification documents to provide additional context. You can also select a template from the C3 Code template gallery as a starting point instead of writing a description from scratch.
  2. C3 Code generates your application: The AI creates application code, a data model, and a UI. The generated application is committed to the workspace's Git repository.
  3. Preview and iterate: Open the application in Preview mode to see it running. Refine it through follow-up prompts, visual configuration in Configure mode, or direct code edits in Code mode.
  4. Deploy: Deploy the application to a staging or production environment with a single action.

Each iteration builds on the previous state. You do not need to regenerate the application from scratch after each change.

Applications and workspaces

In C3 Code, an application is the product you are building. It has a name, a set of collaborators, and a codebase that can be deployed to staging or production. A workspace is an isolated cloud development environment within that application where you actively build and iterate. C3 Code supports multiple workspaces when you configure and link a Git repository to your application.

Each workspace runs its own services and maintains its own Git branch. Changes in one workspace do not affect other workspaces until you merge them. You might use multiple workspaces to:

  • Work on separate features in parallel while keeping a stable workspace available for demos or deployment.
  • Experiment with a major change (for example, a different data model or UI direction) without affecting your working version.
  • Let team members each work independently on the same application.

The default quota is 5 workspaces per user across all applications. If you reach the limit, contact your administrator to increase it.

Workspace modes

The C3 Code workspace provides three modes for reviewing and refining your application:

  • Preview: Runs the live application in the workspace. Use this to test workflows, validate data display, and share a working demo before deployment.
  • Configure: A visual editor for adjusting data bindings, form fields, chart configurations, and UI components without editing code directly.
  • Code: A full browser-based code editor. Use this to inspect generated code, make targeted edits, or debug issues that cannot be resolved through prompts or Configure mode.

For a full breakdown of each mode and all workspace controls, see Workspace.

Who uses C3 Code

Four types of roles use C3 Code:

  • Business analysts and domain experts: Build data applications and internal tools without relying on an engineering team.
  • Application developers: Generate a working starting point, then inspect and extend the code directly through Code mode or the C3 VS Code Extension.
  • Data scientists: Build user-facing dashboards, analytics, and algorithms.
  • C3 Studio and environment administrators: Ensure governance and guardrails on the applications built on C3 Code.

File and data inputs

C3 Code accepts file attachments directly in the prompt interface. Attaching a file provides the AI with additional context when generating or modifying the application.

Supported file types:

  • Spreadsheets and data files: XLSX, CSV. Use to seed data models and generate data-backed UIs.
  • Images: PNG, JPG, JPEG. Use to provide mockups, wireframes, or screenshots as visual references.
  • Documents: PDF, DOCX, TXT, MD. Use to provide requirements, specifications, or reference material.

File size limits: 20 MB per file, 100 MB total per message.

C3 Code also connects to existing C3 Types and data models defined in your C3 AI environment. Applications can read from and write to those types directly.

Version control

Every C3 Code workspace is backed by a Git repository. C3 Code automatically commits generated code as you build and iterate. The workspace header shows the current branch and the number of uncommitted changes.

C3 Code connects to GitHub through an OAuth authorization flow.

To configure the source control connection, see Settings.

Deployment

Applications deploy from the workspace header to Staging or Production environments with a single action. C3 Code handles the deployment pipeline without additional configuration. For information on deployment status and history, see Deployments.

Collaboration

Multiple users can work on the same application through standard Git workflows, using branches and pull requests. Two roles control access to an application:

  • Admin: Full access, including managing collaborators and application settings.
  • Collaborator: Access to workspaces, without collaborator management or deletion rights.

Manage collaborators from the Collaborators tab on the Applications page. See Applications.

AI assistant and MCP servers

The AI assistant is available in all workspace modes through the chat panel on the right side of the workspace. It accepts natural language prompts, file attachments, and pulls context from the currently open file in Code mode. The assistant generates and modifies application code, explains changes, and can diagnose and resolve build and runtime errors.

Extend the assistant's capabilities by adding Model Context Protocol (MCP) servers. MCP servers provide the assistant with access to external APIs, data sources, and services specific to your organization. See Settings.

In this guide

This guide covers the full C3 Code workflow, from your first prompt to a deployed application:

  • Getting Started: Three quick starts to get a working application running fast: building from a prompt, turning a CSV file into a dashboard, and starting from a template.
  • UI Reference: A page-by-page reference for every screen in C3 Code, including all tabs, controls, and panels.
  • Working with C3 Code: Deep dives on each core capability: writing effective prompts, navigating the workspace, integrating data, handling errors, managing versions, and deploying your application.

See also

Was this page helpful?