NxAgent User Guide

NxAgent is an AI-powered assistant embedded directly into the Koodisi workspace. You can ask it questions, have it build workflows and mappings, generate documentation, and investigate what went wrong — without leaving the canvas.

You remain responsible for reviewing generated content, testing changes, and deciding what enters an application.

Where to start

Use NxAgent chat

Open the panel, choose a mode, attach context, and review what it proposes.

Start here →

Build workflows and mappings

Generate a workflow or a data mapping from a plain-language description.

Build something →

Investigate health

Ask why something is failing and get real figures from live data.

Investigate →

Two kinds of AI feature

NxAgent's features reach you in two different ways, and knowing which is which saves you looking in the wrong place.

Kind	How you use it
Chat features	You ask for them in the NxAgent panel
In-place features	They live on their own control in the screen where they apply

In the chat panel

Task	Guide
Open the panel, choose a mode, review changes	Using NxAgent Chat
Generate a workflow or a data mapping	Building Workflows and Mappings
Create, rename, or delete applications and variables	Managing Applications and Variables
Create decision, lookup, and data tables	Working with Tables
Set up alerts and notification channels	Configuring Alerts and Notifications
Deploy, check status, and test a workflow	Publishing, Deploying, and Testing
Investigate failures and performance	Investigating Integration Health
Ask how Koodisi itself works	Asking About Koodisi
Apply your organization's AI skills	Using AI Skills

On their own control

Task	Guide
Generate application, manifest, and workflow descriptions	Generating Descriptions
Review what a publish or deploy would change	Reviewing Changes Before They Ship
Build resources and ID providers	Building Resources and ID Providers
Schema checks and activity generation that run on their own	Automatic AI Assistance

Keep people in control

• Review Preview and Diff before accepting a generated change.
• Test generated workflows and schemas with representative, non-sensitive data.
• Keep Auto-accept changes off for shared applications, unfamiliar tasks, and release-critical work.
• Treat generated explanations as guidance. Verify operational and security decisions against the application and your organization's policies.
• Use the smallest useful context. Remove unrelated workflows and schemas before sending a request.

warning

Do not include passwords, tokens, keys, or customer data in a prompt. Store sensitive values as secret variables instead.

Related

• Governing AI — organization AI skills, usage analytics, credit limits, and the AI kill switch.
• AI Integrations, Tools, and Authorization — runtime boundaries for AI-callable operations.