- 8 minutes to read

AI in Nodinite

Ask questions. Get answers. Solve problems faster.

Nodinite brings AI to your integration monitoring—making complex data simple and troubleshooting instant.

Start Here — Choose Your Setup Path

Choose the setup path that matches the experience you want. The in-product AI features and the external MCP client path are related, but they are not the same setup.

Goal Where What you do Result
Integrated AI Assistant in the Web Client AI System Parameter and AI Assistant Tick AI Chat on the AI Settings page, then configure the language model: Provider, Model, Endpoint, and API key The Web Client chat is available
Optional AI guidance in Diagram Assistance AI System Parameter and AI Assistant Add EnableDiagramAiAssistance and reuse the same provider configuration The AI Assistance tab is available in Diagram Assistance
External MCP client for Repository and Architecture Diagrams MCP Endpoint Connect the client to the IIS-hosted MCP Endpoint (OAuth), or run the NuGet package over stdio (Windows authentication) VS Code, Claude Desktop, Cline, or another MCP client can query Nodinite
BizTalk-specific MCP tools and BizTalk C4 generation AI System Parameter, MCP Endpoint, and BizTalk and MCP Endpoint Enable EnableAiForBizTalk, then use the external MCP setup for BizTalk analysis or the administrator BizTalk generation path BizTalk environments, BizTalk application analysis, and BizTalk C4 generation become available

Important

The AI Assistant and MCP Endpoint were released as part of Nodinite 7.4.0. External MCP setup is required for any external client, regardless of whether you want Repository questions, Architecture Diagrams, or BizTalk workflows. The AI System Parameter controls the in-product AI surfaces and BizTalk-specific AI gating, not external MCP client registration.

Full setup guide for AI Assistant and MCP Endpoint →

Three Powerful AI Capabilities

  • AI Assistant — Ask "Which Finance integrations failed today?" and get instant answers
  • AI Diagnostics — AI automatically scans logs and suggests fixes before you even notice problems
  • Diagram Assistance — Add optional AI guidance to C4 Diagrams while built-in suggestions remain deterministic and preview-first

Your Choice of AI

Provider Type Data Location Cost Best For AI Assistant AI Diagnostics
OLLAMA Local / Ollama Cloud Your infrastructure Free (self-hosted) Sensitive data, strict compliance ✅ ✅
Azure OpenAI Cloud Microsoft cloud Pay-per-token Enterprise Azure environments ✅ ✅
OpenAI Cloud OpenAI cloud Pay-per-token General purpose, broad model selection ✅ ✅
GitHub Models, through OpenAI-compatible endpoint Cloud GitHub cloud Free tier + paid Developer teams, GitHub-native workflows ✅ New 7.7.0

Note

OLLAMA is the only option where data never leaves your infrastructure. With all cloud providers, only query text is sent—never your full database.

AI in Nodinite
AI-powered insights directly in your Nodinite environment.

Why Use AI for Integration Monitoring?

The Old Way

You spend 30 minutes digging through screens, logs, and documentation to find what you need.

With AI

You ask a question and get an answer in 60 seconds. AI watches your logs 24/7 and tells you about problems before they impact users.

Real-World Example: Impact Analysis

Before

"If we take the Payment Gateway offline, what breaks?"
→ 45 minutes searching dependencies across multiple systems

With AI Assistant

→ 60 seconds: "Show me all integrations that depend on Payment Gateway"
→ AI lists dependencies, suggests maintenance window, offers to visualize in Mapify

Time savings: 97%

Real-World Example: Faster Problem Resolution

Before AI Diagnostics

Agent fails → Discovered 2-8 hours later → 30 minutes analyzing logs → Fix applied

With AI Diagnostics

Agent fails → AI detects in 60 seconds → Root cause + fix shown immediately → 5 minutes to resolve

MTTR reduction: 95%

Choose Your AI Capability

AI Assistant — Ask Questions, Get Answers

What is it?

A chat interface where you ask questions about your integration landscape in plain English.

What can you ask?

  • "Which Finance integrations failed today?"
  • "Show me all dependencies for the Payment Gateway"
  • "What Custom Metadata exists for SAP Orders?"
  • "List all Services in the Healthcare domain"

Who should use it?

  • Integration operators exploring the landscape
  • Business analysts investigating issues
  • Compliance teams generating reports
  • Anyone who wants answers without learning query syntax

Get started with AI Assistant →

AI Diagnostics — Automatic Problem Detection

What is it?

AI that watches your monitoring agent logs 24/7, detects issues, and tells you what's wrong and how to fix it.

What does it do?

  • Scans diagnostics logs continuously
  • Detects errors and warnings automatically
  • Explains root cause in plain language
  • Suggests specific fixes
  • Works while you sleep

Who should use it?

  • Operations teams monitoring agents
  • Integration experts managing production systems
  • Anyone tired of manually checking logs
  • Teams wanting to reduce incident response time

Get started with AI Diagnostics →

Diagram Assistance — Guided C4 Diagrams with Optional AI

What is it?

Diagram Assistance is the guidance workspace for Architecture Diagrams and C4 Diagrams. It uses Built-in Suggestions to surface Add from Set, Coverage, and Review Impacts for the current Diagram Set. When EnableDiagramAiAssistance is enabled, the same workspace also exposes an AI Assistance tab.

Who should use it?

  • Solution architects reviewing Diagram Sets
  • Integration teams building L1, L2, and L3 diagrams
  • Reviewers who need quick impact analysis before promoting diagram changes

Learn more about Diagram Assistance →

Quick Start

Get AI Assistant Running (5 minutes)

  1. Open Administration → Settings → System Parameters, and click the AI row to open AI Settings.
  2. In the Features card, tick AI Chat (the EnableMcpChat flag).
  3. In the Language model card, select your Provider.
  4. Enter Model, Endpoint, and API key (and Deployment name for Azure OpenAI).
  5. Save the configuration and open the AI Assistant widget.
  6. Ask a simple question such as "Show me all Services".

Full setup guide →

Get the MCP Endpoint Ready (5 minutes)

  1. Choose the way in: the IIS-hosted MCP Endpoint when the environment uses OAuth, or the NuGet package over stdio when it uses Windows authentication.
  2. Add the matching server entry to .vscode/mcp.json, as shown in Set up AI Assistant and MCP Endpoint.
  3. For the package, check the .NET 10 SDK, the NuGet --source, and a plain-text ConnectionStrings__NodiniteConfig.
  4. Reload your MCP client and run a simple repository question to confirm connectivity.

This same MCP setup applies whether you want simple Repository questions, Architecture Diagram and Diagram Set review, or BizTalk workflows. BizTalk-specific workflows additionally require EnableAiForBizTalk and the appropriate administrator access.

MCP setup details →

Enable AI Guidance in C4 Diagrams (optional)

  1. On the AI Settings page, tick C4 diagram assistance (the EnableDiagramAiAssistance flag of the AI System Parameter).
  2. Configure the language model once, on the same page.
  3. Open Diagram Assistance from Architecture Diagrams and verify that the AI Assistance tab is available.

Built-in suggestions in Diagram Assistance remain available even when AI is disabled.

Get AI Diagnostics Running (5 minutes)

  1. Open your Monitoring Agent configuration (Logic Apps, Message Queuing, etc.)
  2. Go to the AI Diagnostics tab
  3. Enable AI Diagnostics
  4. Choose your AI provider
  5. Set scan interval (recommended: 900 seconds)
  6. Save—AI starts monitoring automatically

Full setup guide →

Common Questions

Q: Can AI change my configuration?
A: The AI Assistant and AI Diagnostics analyze and suggest. The MCP Endpoint also has write and delete tools, for Articles, Snippets, Templates, Document Profiles, C4 Diagrams, and the Repository. It doesn't check Nodinite Roles: in OAuth mode the granted scopes limit the tools, in Windows mode every allowed user can call them, and the NuGet package has no authorization. See MCP Endpoint.

Q: What AI models work with Nodinite?
A: OLLAMA (phi3, llama2, etc.), Azure OpenAI (gpt-4o, gpt-3.5-turbo), OpenAI (gpt-4o, gpt-4o-mini, etc.), GitHub Models (GPT-4o, Mistral-large, Llama, etc.) through OpenAI-compatible endpoint, or any MCP-compatible AI client. GitHub Models support in AI Diagnostics is coming soon.

What You'll Save

Time Savings Example

Integration Operator at Mid-Sized Company

  • 100 incidents/month
  • Before AI: 30 min investigation each = 50 hours/month
  • With AI: 5 min investigation each = 8.3 hours/month
  • Savings: 41.7 hours/month

At $75/hour = $3,125/month ($37,500/year)

Add self-service queries, faster compliance reporting, and reduced escalations:

Estimated ROI: $50,000-$75,000/year

Tips for Success

For AI Assistant

  • Be specific: "Show me failed SAP integrations today" beats "Show me failures"
  • Use business terms: AI understands Domains, Services, and Custom Metadata
  • Ask follow-up questions: AI remembers context

For AI Diagnostics

  • Start with 900-second scan interval (15 minutes)
  • Use temperature 0.2 for focused diagnostics
  • Verify suggestions before applying fixes
  • Monitor API costs if using cloud AI

Next Steps

Start with the AI getting started guide, then choose the AI experience you want to use first:

Need help choosing? Contact Support