Mendix 11 – The dawn of the AI-first era in low-code

Mendix 11 marks a new direction in the world of low-code applications, introducing a set of GenAI components and tools that enable the creation of intelligent, conversational, and contextual solutions. It’s no longer just a suggestion maker – now we have a fully-fledged AI ecosystem that:

• conducts conversations,
• searches for information semantically,
• analyzes data in context,
• performs microflows as actions,
• and even controls the computer environment.

GenAI Building Blocks in the Mendix Marketplace

With ready-made components from the Mendix Marketplace, implementing AI into business applications becomes quick and predictable. The main building blocks are:

• GenAI Commons – the foundation for integration with AI models

Provides ready-made actions for chat completions, RAG (Retrieval Augmented Generation), function calling, and token monitoring.

Business value: allows you to easily switch between model providers (OpenAI, Bedrock, Mendix Cloud GenAI) while maintaining a consistent interface.

• Conversational UI – a ready-made chat interface

Offers several UI variants: full-screen, sidebar, and modal. 

Integration with Atlas 4 and React Client sometimes requires minor tweaks, but provides the immediate ability to implement an AI agent in your application.

Business value: Rapidly launch an intelligent chatbot without the cost of developing your own UI.

• Agent Commons (Agents Kit) + Agent Builder Starter App

A tool for visually building AI agents: defining prompts, selecting models, connecting knowledge bases, and adding microflows as “tools.”

The Starter App allows you to quickly start building an AI assistant that talks, searches, and performs actions.

Business value: enables the creation of HR assistants, helpdesks, and sales agents – without extensive development costs.

• Application Starters (AI Bot, Support Assistant, RFP Assistant)

Ready-made templates based on RAGs and function calling. 

Business value: Accelerate the implementation of chatbots for customer service, service support, and RFP process automation.

• Knowledge Bases (PgVector, Azure AI Search, Mendix Cloud KB)

Mendix supports various knowledge base engines, enabling the construction of Semantic Search and RAGs.

Business value: AI responds based on company data, not “generic knowledge” of the model → greater precision and regulatory compliance.

• Function Calling and Tool Choice

AI can trigger microflows, and the developer determines which tools are available.

Business value: Controlled automation – the AI ​​agent not only responds but also performs actions (e.g., creates orders, checks statuses).

• Semantic Search – the foundation of modern Mendix applications

The added value of Mendix 11 is that GenAI components are seamlessly integrated with Semantic Search. This allows you to:

• search regulations, instructions, and contracts in natural language,
• analyze service requests and report trends,
• support salespeople searching for customer and offer data,
• manage HR and compliance processes, ensuring quick access to knowledge.

Pretius Semantic Search: A detailed look at the possibilities offered by integration with Mendix

Let’s take a more detailed look at that last building block. Its business value can’t be overstated, and Mendix offers plenty of flexibility in terms of its implementation. To function, the Mendix Semantic Search component must be integrated with an AI model. Here’s what you can achieve by using it together with our Pretius AI Semantic Search tool.

Semantic Search – intelligent data retrieval

Pretius Semantic Search allows you to search data not by keywords, but by meaning. Thanks to integration with Mendix, we can leverage this in virtually any business process.

Use cases:

• Customer service – the consultant enters a question in natural language (i.e. “What are the conditions for terminating the contract?”), and the system immediately returns the relevant fragment of the document.
  • Microflow: UserInput → SemanticSearch → ReturnSnippet → ShowInChat.

• E-commerce – the customer searches for “a phone with a good night camera”, and the system suggests smartphones with a large matrix and night mode, even if the description does not contain these words.
  • Microflow: SearchQuery → SemanticSearch(products) → FilterAvailable → SortBySimilarity → DisplayResults.

• Service request analysis – searching for similar incidents in history shortens the time to diagnose the problem.
  • Microflow: NewTicket → SemanticSearch(Incidents) → SuggestSolutions → PreFillTicket.

• Onboarding new employees – a new employee asks: “How do I report a vacation?” and Semantic Search suggests a fragment of the related HR procedure.
  • Microflow: EmployeeQuestion → SemanticSearch(HRDocs) → ReturnProcedure.

• B2B Sales – the salesperson asks: “What offers did we make to customer X in the last year?” and the system returns similar records and offer documents.
  • Microflow: Query → SemanticSearch(Offers) → AttachDocs → DisplaySummary.

• Compliance and risk – a lawyer searches for clauses in hundreds of contract pages. Semantic Search instantly identifies relevant provisions.
  • Microflow: LawyerQuery → SemanticSearch(Contracts) → HighlightClause.

• Trend reporting – The manager asks, “What were the most common reasons for complaints in Q1 2024?” Semantic Search groups similar reports and returns a report.
  • Microflow: ManagerQuery → SemanticSearch(Complaints) → Cluster → Summarize → GeneratePDF.

Agent Builder – creating your own AI agents

With Agent Builder in Mendix, we can quickly create intelligent assistants that use Pretius Semantic Search as a source of knowledge.

Use cases:

• HR assistant – Answers employee questions about benefits and procedures using a knowledge base.
• Sales agent – ​​Generates brand-style product descriptions based on Semantic Search and ERP data.
• Helpdesk AI – Analyzes customer requests and proposes solutions based on previous cases.

Microflow: AgentInput → SemanticSearch(DomainKB) → RAG Prompt → LLM Answer → Action/Display.

Multi-Agent Patterns – a team of virtual experts

Sometimes one agent isn’t enough. Multi-Agent Patterns allow you to build a system in which multiple agents collaborate as a team of experts.

Use cases:

• Complaint handling
  • Agent 1 analyzes the complaint,
  • Agent 2 checks similar cases in Semantic Search,
  • Agent 3 generates a response consistent with company policy.
• Creating business reports
  • Agent A retrieves historical data,
  • Agent B analyzes and draws conclusions,
  • Agent C summarizes the entire process in a PDF report for management.

Microflow: UserQuery → AgentA (Search) → AgentB (Analysis) → AgentC (Summarize) → ReturnPDF.

Function Calling – AI that performs actions

Thanks to Function Calling AI in Mendix, it not only responds but also executes microflows.

Use cases:

• The customer writes in the chat: “I want to order 3 pieces of product X” and AI automatically calls createOrder().
• Employee types: “Show me my leave status” and agent calls getLeaveRequestStatus().
• Consultant writes: “Find the data of the customer with PESEL number 123…” and AI queries the CRM system.

Microflow: ChatInput → IntentDetection → If(action) → CallMicroflow(FunctionX) → ReturnResult.

RAG – Retrieval Augmented Generation

RAG is a combination of Semantic Search and LLM. First, we find relevant fragments, then AI generates the answer.

Use cases:

• Document Chat – an employee uploads a contract and asks questions in natural language. The response comes from the document.
• Compliance and Law – AI interprets regulatory provisions.
• Finance – an analyst asks: “What were the most common causes of payment delays in 2023?” and the system generates a summary.

Microflow: UserQuestion → SemanticSearch(Documents) → BuildRAGPrompt → LLM Answer → Display.

Summary: Transform your apps with the Mendix AI toolkit

Mendix 11 revolutionizes low-code development by introducing an AI-first ecosystem with ready-made components like GenAI Commons, Conversational UI, and Agents Kit, enabling the creation of intelligent, conversational, and contextual solutions. This platform allows businesses to leverage AI driven by their own corporate data through knowledge bases and semantic search, transforming AI into an active participant in processes that can respond, analyze, report, and trigger microflows. 

Furthermore, the integration of Semantic Search with Generative AI in Mendix unlocks vast possibilities across various domains, including customer service, sales, e-commerce, HR, compliance, and analytics, positioning Mendix as a comprehensive platform where AI actively drives business actions and supports real-time processes beyond just information retrieval.

If you’re interested in AI Semantic Search, please visit our dedicated page or contact us at hello@pretius.com. For more of this kind of content, check the following articles on our blog:

1. Semantic search with LLMs: Building a better search experience
2. AI knowledge base: How to unify company information
3. How AI semantic search with LLMs is redefining enterprise search
4. Computer vision applications in manufacturing: GAN-driven defect detection for automated quality control
5. Computer vision in retail: Optimizing store layouts