MakerAI 3.1 Suite — The AI Operating System for Delphi

🌐 Official Website: https://makerai.cimamaker.com

MakerAI 3.1 is the definitive AI orchestration framework for Delphi developers.

Now featuring support for GPT-5.1, Gemini 3.0, and Claude 4.5, MakerAI allows you to design and deploy intelligent, self-learning, and context-aware systems natively in Delphi with full support for RAG, Knowledge Graphs, MCP Servers, and autonomous agents.

🚀 What's New in MakerAI 3.1.001 (Dec-10-2025)

MakerAI 3.1 represents a massive leap forward in native AI capabilities:

• 🧠 Next-Gen Models: Full support for OpenAI GPT-5.1 & Sora 2 (Video), Gemini 3.0 (NanoBanana & Veo 3), and Claude 4.5.
• 🎨 Professional FMX UI: New set of FireMonkey visual components for building modern, multimodal chat interfaces (Text + Voice + Vision) effortlessly.
• 🎯 High-Precision RAG: Added Embedding Rerank functionality to drastically improve search relevance in large document sets.
• 🕸️ RagGraph Engine: A new graph-based knowledge engine that goes beyond vector similarity.
• 🔗 MCP Server Framework: Create your own MCP Servers natively using SSE or DataSnap.

📊 Feature Support Matrix

Overview of capabilities across supported providers in MakerAI 3.1.

Legend: ✅ = Full Support (Sync & Async) | ⚠️ = Partial Support (Sync Only) | ❌ = Not Supported by Provider/Driver

🛠️ MakerAI 3.1 Core Modules & Components

The suite is composed of non-visual and visual components designed for maximum modularity.

🌐 Universal Connectivity & LLMs (AiConnection V3)

🎨 Visual Multimodal Interface (FMX Only)

🧠 Knowledge & Memory Engines (RAG)

📞 Tools & Protocol (MCP & Functions)

📦 Detailed Installation Guide (Modular V3.0)

Follow these steps carefully to ensure a correct installation of the core packages.

Step 1: Get the Source Code

git clone https://github.com/gustavoeenriquez/MakerAi.git

Step 2: Configure Paths in the Delphi IDE

Add the following subfolders to your Library Path (Tools > Options > Language > Delphi > Library):

...\MakerAi\Source\Agents
...\MakerAi\Source\Chat
...\MakerAi\Source\ChatUI       <-- (FMX Visual Components)
...\MakerAi\Source\Core
...\MakerAi\Source\Design
...\MakerAi\Source\MCPClient
...\MakerAi\Source\MCPServer
...\MakerAi\Source\Packages
...\MakerAi\Source\RAG
...\MakerAi\Source\Resources
...\MakerAi\Source\Tools
...\MakerAi\Source\Utils

Step 3: Compile and Install the Packages

Compile and Install in this specific order:

1. MakerAi.dpk (Runtime Core)
2. MakerAi.RAG.Drivers.dpk (Database Connectors)
3. MakerAi.UI.dpk (FMX Visual Components)
4. MakerAiDsg.dpk (Design-Time Editors)

💡 Usage Examples

Example 1: Simple Text Generation with GPT-5.1

procedure TForm1.Button1Click(Sender: TObject);
var
  Response: string;
begin
  // Ensure DriverName is 'OpenAI' and Model is 'gpt-5.1-turbo'
  AiConnection1.DriverName := 'OpenAI';
  AiConnection1.Model := 'gpt-5.1-turbo';
  
  Response := AiConnection1.AddMessageAndRun('Explain the theory of relativity in one sentence.', 'user', []);
  Memo1.Lines.Add('GPT-5.1: ' + Response);
end;

Example 2: Multimodal Input Using TAiChatConnection (Universal Connector)

procedure TForm18.BtnFromAiConnectionClick(Sender: TObject);
var
  AiChat: TAiChatConnection;
  MF: TAiMediaFile;
  Prompt, Res: String;
begin
  AiChat := TAiChatConnection.Create(nil);
  try
    AiChat.DriverName := 'OpenAi';
    AiChat.Model := 'gpt-5.1';
    AiChat.Params.Values['NativeInputFiles'] := '[Tfc_Image, tfc_pdf]';
    AiChat.Params.Values['ChatMediaSupports'] := '[Tcm_Image, tcm_pdf]';
    
    MF := TAiMediaFile.Create;
    try
      MF.LoadFromFile('MyFile.png');
      Prompt := 'Describe this image';
      Res := AiChat.AddMessageAndRun(Prompt, 'user', [MF]);
      ShowMessage(Res);
    finally
      MF.Free;
    end;
  finally
    AiChat.Free;
  end;
end;

Example 3: Multimodal Input Using TAiOpenChat (OpenAI-Specific Driver)

procedure TForm18.BtnFromOpenAIClick(Sender: TObject);
var
  OpenAiChat: TAiOpenChat;
  MF: TAiMediaFile;
  Prompt, Res: String;
begin
  OpenAiChat := TAiOpenChat.Create(nil);
  try
    OpenAiChat.Model := 'gpt-5.1';
    OpenAiChat.NativeInputFiles := [Tfc_Image, tfc_pdf];
    OpenAiChat.ChatMediaSupports := [Tcm_Image, tcm_pdf];
    
    MF := TAiMediaFile.Create;
    try
      MF.LoadFromFile('MyFile.png');
      Prompt := 'Describe this image';
      Res := OpenAiChat.AddMessageAndRun(Prompt, 'user', [MF]);
      ShowMessage(Res);
    finally
      MF.Free;
    end;
  finally
    OpenAiChat.Free;
  end;
end;

Note:

• Example 2 uses the universal TAiChatConnection component, which allows you to easily switch between providers (OpenAI, Claude, Gemini, etc.) by changing the DriverName property.
• Example 3 uses the OpenAI-specific TAiOpenChat component, providing direct access to OpenAI-native properties and features.
• Both examples demonstrate multimodal input support through the TAiMediaFile class, enabling you to send images, PDFs, and other supported file types directly to the AI model.

⚠️ Known Issues & Limitations

Please be aware of the following behaviors in version 3.1.0:

1. MCP SSE Server: The Server-Sent Events (SSE) implementation for the MCP Server framework is currently experimental. Connectivity is not yet 100% functional, and you may experience intermittent drops.
2. Linux Compilation: Compiling for Linux currently requires significant manual adjustments to library paths and dependencies. Seamless support is planned for a future patch.
3. Delphi Compatibility:
   • Full Support: Delphi 11 Alexandria, 12 Athens, and 13 Florence.
   • Limited Support: Delphi 10.4 Sydney (some advanced features or new FMX components may not function as expected).

🔄 Changelog

Version 3.1.0 (December 3, 2025) - OFFICIAL RELEASE

• New Model Support:
  • OpenAI: Added full support for Responses API, gpt-5.1 models, and Sora 2 (Video).
  • Gemini: Added support for Gemini 3.0 series (NanoBanana, Veo 3).
  • Claude: Added support for Claude 4.5 API.
• Visual Interface (FMX):
  • Added professional-grade FMX visual components: TAIChatList and TAIMultimodalInput for rich text/voice/video interaction.
• RAG Engine:
  • Introduced Rerank capabilities in embeddings to improved search precision.
  • Introduced RAGGraph for Knowledge Graph based retrieval.
• New Drivers:
  • Added TAIChatLMStudio for local inference compatibility.
  • Added TAIChatKimi (Kimi 2).
  • Updated TAIChatOllama to support the new native local API.
• MCP Protocol:
  • Added MCP Server framework with support for SSE (Server-Sent Events) and MakerAI+DataSnap protocols.

Version 2.6.0 (August 2025)

• Added Delphi 11 Alexandria official support.
• Preliminary support for Ollama async calls.

🤝 Contributing

Contributions are welcome. Please open an issue or submit a pull request on GitHub.

💬 Community & Support

• Official Website: https://makerai.cimamaker.com
• Email: gustavoeenriquez@gmail.com

📜 License

This project is licensed under the MIT License. See the LICENSE.txt file for details.