[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fxch6ciu1ah614hhkh55qJnMx7eT1O_giw6ds79_vGOg":3},{"item":4},{"id":5,"idKnowledge":6,"slug":7,"title":8,"description":9,"bodyMarkdown":10,"bodyHtml":11,"author":12,"date":13,"createdAt":14,"topics":15,"image":19,"hasDownload":20,"fileName":21},"23","9049C8F8-76AB-234D-9528-B67BE198E1AD","gemini-embedding-2-geef-je-filemaker-oplossing-een-ai-geheugen","Gemini Embedding 2: Give Your FileMaker Solution an AI Memory","AI has long since stopped being just about chatting. The real power emerges when an AI model understands and can retrieve your own knowledge, documents, manuals, emails, or project information. That's where embeddings come in.","# Gemini Embedding 2: Give Your FileMaker Solution an AI Memory\n\nAI has long since stopped being just about chatting. The real power emerges when an AI model understands and can retrieve your own knowledge, documents, manuals, emails, or project information. That's where embeddings come in.\n\nWith the introduction of Gemini Embedding 2, Google has made one of the most powerful embedding models available today. But what exactly are embeddings, why do they matter, and how can you use them within a FileMaker solution?\n\n---\n\n## What are embeddings?\n\nAn embedding can be seen as a numerical representation of text.\n\nWhen you pass a document, manual, or email through an embedding model, it is converted into a long list of numbers (a vector). Those numbers do not represent the exact words, but rather the meaning of the text.\n\nAs a result, two texts that share the same meaning can lie close together in what is called the vector space, even if entirely different words are used.\n\n### Example\n\nDocument\n\n> How do I create an invoice in FileMaker?\n\nUser question\n\n> What is the way to generate a new bill?\n\nTo a traditional search engine, these look like two different texts. To an embedding model, they are virtually identical.\n\nThis is what makes semantic search possible.\n\n---\n\n## Why Gemini Embedding 2?\n\nGoogle developed Gemini Embedding 2 as the successor to earlier embedding models.\n\nKey advantages include:\n\n- Higher accuracy in semantic search\n- Strong performance on business documents\n- Support for multiple languages\n- Suitable for large knowledge bases\n- Lower cost than repeatedly sending complete documents to a Large Language Model\n- Fast processing of large volumes of data\n\nFor organisations that want to combine AI with their own knowledge, this is an essential building block.\n\n---\n\n## How does it work?\n\nThe process consists of four steps.\n\n### Step 1: Gather your knowledge\n\nThink of:\n\n- Manuals\n- Procedures\n- Project documentation\n- Client files\n- Support tickets\n- Contracts\n- Product information\n- Safety documentation\n\n### Step 2: Create embeddings\n\nEach piece of text is sent to Gemini Embedding 2.\n\nThe model returns a vector such as:\n\njson [   0.0245,   -0.8121,   0.4431,   ... ] \n\nThis vector is stored in a vector database.\n\n### Step 3: User asks a question\n\nFor example:\n\n> How do I add a user to the system?\n\nThis question is also converted into an embedding.\n\n### Step 4: Search for similar knowledge\n\nVector search is used to find which documents are closest to the question.\n\nThe retrieved documents are then sent along to a Large Language Model such as Gemini, Claude, or ChatGPT.\n\nWe call this process:\n\n## Retrieval Augmented Generation (RAG)\n\nWith RAG, an AI model can provide answers based on your own knowledge rather than relying solely on its training data.\n\n---\n\n## Why is this interesting for FileMaker?\n\nMany organisations store their business knowledge in FileMaker.\n\nThink of:\n\n- CRM data\n- Tickets\n- Project files\n- Knowledge bases\n- Product documentation\n- Safety procedures\n- Contract information\n- Internal manuals\n\nTraditionally, an employee has to search using keywords.\n\nWith embeddings, a user can simply ask a question:\n\n> What procedure applies when an employee reports a near-miss?\n\nThe AI then automatically searches for the most relevant documents and generates an answer based on the information found.\n\n---\n\n## Practical example: HSE platform\n\nSuppose you manage an HSE platform containing:\n\n- Toolbox meetings\n- Safety instructions\n- Risk analyses\n- Incident reports\n- Procedures\n- Work permits\n\nBy embedding all documents, an intelligent knowledge assistant is created.\n\nUsers can ask questions such as:\n\n- What should I do in the event of a fall incident?\n- Which PPE is mandatory for working at height?\n- What does our procedure say about reporting a near miss?\n- What training is required for these activities?\n\nThe AI searches directly across all available documentation and presents an answer including source references.\n\n---\n\n## Combining Gemini Embedding 2 with FileMaker\n\nA possible architecture looks like this:\n\ntext FileMaker     │     ├── Documents     ├── Tickets     ├── Procedures     └── CRM data             │             ▼     Gemini Embedding 2             │             ▼       Vector Database  (Pinecone \u002F Weaviate \u002F Qdrant \u002F pgvector)             │             ▼  Gemini \u002F Claude \u002F ChatGPT             │             ▼         Answer \n\nNew documents can be processed automatically as soon as they are saved in FileMaker.\n\nThis keeps the knowledge base up to date at all times.\n\n---\n\n## When do you use embeddings?\n\nEmbeddings are valuable when you:\n\n- Have more than a few hundred documents\n- Want to support natural language use\n- Want AI to answer based on your own knowledge\n- Want to build intelligent search functionality\n- Want to feed chatbots with business information\n- Want to automate support processes\n- Want to unlock knowledge currently hidden within documents\n\nFor simple keyword search, embeddings are often not necessary.\n\nFor AI-driven knowledge platforms, they have become virtually indispensable.\n\n---\n\n## Practical applications\n\nWithin FileMaker, we are seeing applications like the following increasingly often:\n\n### AI Helpdesk\n\nAllow users to ask questions about manuals, procedures, and work instructions.\n\n### Smart document search\n\nFind documents by meaning rather than by keywords.\n\n### Customer support\n\nAutomatically retrieve relevant support tickets from the past.\n\n### Safety management\n\nMake safety procedures directly accessible to employees on site.\n\n### CRM Intelligence\n\nLet AI gather relevant customer information from thousands of notes and contact moments.\n\n### Project assistant\n\nQuickly find previous project experiences, solutions, and documentation.\n\n---\n\n## Conclusion\n\nGemini Embedding 2 forms an essential building block for modern AI solutions.\n\nBy converting business knowledge into vectors, it becomes possible to retrieve information not just by words, but above all by meaning.\n\nFor FileMaker developers, this opens the door to:\n\n- Intelligent knowledge bases\n- AI assistants\n- Semantic search engines\n- Advanced RAG solutions\n- Smarter helpdesks\n- Knowledge-driven chatbots\n\nWhere Large Language Models handle the generation of answers, embeddings ensure that the right knowledge is found.\n\nAnd it is precisely that combination that makes business AI truly valuable.\n\n---\n\n## Want to know more?\n\nWould you like to know how to combine Gemini Embedding 2 with FileMaker, a vector database, and an AI model such as Gemini, Claude, or ChatGPT?\n\nGet in touch with Loggix. We help organisations build AI solutions that leverage their own data, processes, and knowledge.","\u003Ch1>Gemini Embedding 2: Give Your FileMaker Solution an AI Memory\u003C\u002Fh1>\n\u003Cp>AI has long since stopped being just about chatting. The real power emerges when an AI model understands and can retrieve your own knowledge, documents, manuals, emails, or project information. That's where embeddings come in.\u003C\u002Fp>\n\u003Cp>With the introduction of Gemini Embedding 2, Google has made one of the most powerful embedding models available today. But what exactly are embeddings, why do they matter, and how can you use them within a FileMaker solution?\u003C\u002Fp>\n\u003Chr>\n\u003Ch2>What are embeddings?\u003C\u002Fh2>\n\u003Cp>An embedding can be seen as a numerical representation of text.\u003C\u002Fp>\n\u003Cp>When you pass a document, manual, or email through an embedding model, it is converted into a long list of numbers (a vector). Those numbers do not represent the exact words, but rather the meaning of the text.\u003C\u002Fp>\n\u003Cp>As a result, two texts that share the same meaning can lie close together in what is called the vector space, even if entirely different words are used.\u003C\u002Fp>\n\u003Ch3>Example\u003C\u002Fh3>\n\u003Cp>Document\u003C\u002Fp>\n\u003Cblockquote>\n\u003Cp>How do I create an invoice in FileMaker?\u003C\u002Fp>\n\u003C\u002Fblockquote>\n\u003Cp>User question\u003C\u002Fp>\n\u003Cblockquote>\n\u003Cp>What is the way to generate a new bill?\u003C\u002Fp>\n\u003C\u002Fblockquote>\n\u003Cp>To a traditional search engine, these look like two different texts. To an embedding model, they are virtually identical.\u003C\u002Fp>\n\u003Cp>This is what makes semantic search possible.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2>Why Gemini Embedding 2?\u003C\u002Fh2>\n\u003Cp>Google developed Gemini Embedding 2 as the successor to earlier embedding models.\u003C\u002Fp>\n\u003Cp>Key advantages include:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>Higher accuracy in semantic search\u003C\u002Fli>\n\u003Cli>Strong performance on business documents\u003C\u002Fli>\n\u003Cli>Support for multiple languages\u003C\u002Fli>\n\u003Cli>Suitable for large knowledge bases\u003C\u002Fli>\n\u003Cli>Lower cost than repeatedly sending complete documents to a Large Language Model\u003C\u002Fli>\n\u003Cli>Fast processing of large volumes of data\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>For organisations that want to combine AI with their own knowledge, this is an essential building block.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2>How does it work?\u003C\u002Fh2>\n\u003Cp>The process consists of four steps.\u003C\u002Fp>\n\u003Ch3>Step 1: Gather your knowledge\u003C\u002Fh3>\n\u003Cp>Think of:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>Manuals\u003C\u002Fli>\n\u003Cli>Procedures\u003C\u002Fli>\n\u003Cli>Project documentation\u003C\u002Fli>\n\u003Cli>Client files\u003C\u002Fli>\n\u003Cli>Support tickets\u003C\u002Fli>\n\u003Cli>Contracts\u003C\u002Fli>\n\u003Cli>Product information\u003C\u002Fli>\n\u003Cli>Safety documentation\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3>Step 2: Create embeddings\u003C\u002Fh3>\n\u003Cp>Each piece of text is sent to Gemini Embedding 2.\u003C\u002Fp>\n\u003Cp>The model returns a vector such as:\u003C\u002Fp>\n\u003Cp>json [   0.0245,   -0.8121,   0.4431,   ... ] \u003C\u002Fp>\n\u003Cp>This vector is stored in a vector database.\u003C\u002Fp>\n\u003Ch3>Step 3: User asks a question\u003C\u002Fh3>\n\u003Cp>For example:\u003C\u002Fp>\n\u003Cblockquote>\n\u003Cp>How do I add a user to the system?\u003C\u002Fp>\n\u003C\u002Fblockquote>\n\u003Cp>This question is also converted into an embedding.\u003C\u002Fp>\n\u003Ch3>Step 4: Search for similar knowledge\u003C\u002Fh3>\n\u003Cp>Vector search is used to find which documents are closest to the question.\u003C\u002Fp>\n\u003Cp>The retrieved documents are then sent along to a Large Language Model such as Gemini, Claude, or ChatGPT.\u003C\u002Fp>\n\u003Cp>We call this process:\u003C\u002Fp>\n\u003Ch2>Retrieval Augmented Generation (RAG)\u003C\u002Fh2>\n\u003Cp>With RAG, an AI model can provide answers based on your own knowledge rather than relying solely on its training data.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2>Why is this interesting for FileMaker?\u003C\u002Fh2>\n\u003Cp>Many organisations store their business knowledge in FileMaker.\u003C\u002Fp>\n\u003Cp>Think of:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>CRM data\u003C\u002Fli>\n\u003Cli>Tickets\u003C\u002Fli>\n\u003Cli>Project files\u003C\u002Fli>\n\u003Cli>Knowledge bases\u003C\u002Fli>\n\u003Cli>Product documentation\u003C\u002Fli>\n\u003Cli>Safety procedures\u003C\u002Fli>\n\u003Cli>Contract information\u003C\u002Fli>\n\u003Cli>Internal manuals\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>Traditionally, an employee has to search using keywords.\u003C\u002Fp>\n\u003Cp>With embeddings, a user can simply ask a question:\u003C\u002Fp>\n\u003Cblockquote>\n\u003Cp>What procedure applies when an employee reports a near-miss?\u003C\u002Fp>\n\u003C\u002Fblockquote>\n\u003Cp>The AI then automatically searches for the most relevant documents and generates an answer based on the information found.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2>Practical example: HSE platform\u003C\u002Fh2>\n\u003Cp>Suppose you manage an HSE platform containing:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>Toolbox meetings\u003C\u002Fli>\n\u003Cli>Safety instructions\u003C\u002Fli>\n\u003Cli>Risk analyses\u003C\u002Fli>\n\u003Cli>Incident reports\u003C\u002Fli>\n\u003Cli>Procedures\u003C\u002Fli>\n\u003Cli>Work permits\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>By embedding all documents, an intelligent knowledge assistant is created.\u003C\u002Fp>\n\u003Cp>Users can ask questions such as:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>What should I do in the event of a fall incident?\u003C\u002Fli>\n\u003Cli>Which PPE is mandatory for working at height?\u003C\u002Fli>\n\u003Cli>What does our procedure say about reporting a near miss?\u003C\u002Fli>\n\u003Cli>What training is required for these activities?\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>The AI searches directly across all available documentation and presents an answer including source references.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2>Combining Gemini Embedding 2 with FileMaker\u003C\u002Fh2>\n\u003Cp>A possible architecture looks like this:\u003C\u002Fp>\n\u003Cp>text FileMaker     │     ├── Documents     ├── Tickets     ├── Procedures     └── CRM data             │             ▼     Gemini Embedding 2             │             ▼       Vector Database  (Pinecone \u002F Weaviate \u002F Qdrant \u002F pgvector)             │             ▼  Gemini \u002F Claude \u002F ChatGPT             │             ▼         Answer \u003C\u002Fp>\n\u003Cp>New documents can be processed automatically as soon as they are saved in FileMaker.\u003C\u002Fp>\n\u003Cp>This keeps the knowledge base up to date at all times.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2>When do you use embeddings?\u003C\u002Fh2>\n\u003Cp>Embeddings are valuable when you:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>Have more than a few hundred documents\u003C\u002Fli>\n\u003Cli>Want to support natural language use\u003C\u002Fli>\n\u003Cli>Want AI to answer based on your own knowledge\u003C\u002Fli>\n\u003Cli>Want to build intelligent search functionality\u003C\u002Fli>\n\u003Cli>Want to feed chatbots with business information\u003C\u002Fli>\n\u003Cli>Want to automate support processes\u003C\u002Fli>\n\u003Cli>Want to unlock knowledge currently hidden within documents\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>For simple keyword search, embeddings are often not necessary.\u003C\u002Fp>\n\u003Cp>For AI-driven knowledge platforms, they have become virtually indispensable.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2>Practical applications\u003C\u002Fh2>\n\u003Cp>Within FileMaker, we are seeing applications like the following increasingly often:\u003C\u002Fp>\n\u003Ch3>AI Helpdesk\u003C\u002Fh3>\n\u003Cp>Allow users to ask questions about manuals, procedures, and work instructions.\u003C\u002Fp>\n\u003Ch3>Smart document search\u003C\u002Fh3>\n\u003Cp>Find documents by meaning rather than by keywords.\u003C\u002Fp>\n\u003Ch3>Customer support\u003C\u002Fh3>\n\u003Cp>Automatically retrieve relevant support tickets from the past.\u003C\u002Fp>\n\u003Ch3>Safety management\u003C\u002Fh3>\n\u003Cp>Make safety procedures directly accessible to employees on site.\u003C\u002Fp>\n\u003Ch3>CRM Intelligence\u003C\u002Fh3>\n\u003Cp>Let AI gather relevant customer information from thousands of notes and contact moments.\u003C\u002Fp>\n\u003Ch3>Project assistant\u003C\u002Fh3>\n\u003Cp>Quickly find previous project experiences, solutions, and documentation.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2>Conclusion\u003C\u002Fh2>\n\u003Cp>Gemini Embedding 2 forms an essential building block for modern AI solutions.\u003C\u002Fp>\n\u003Cp>By converting business knowledge into vectors, it becomes possible to retrieve information not just by words, but above all by meaning.\u003C\u002Fp>\n\u003Cp>For FileMaker developers, this opens the door to:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>Intelligent knowledge bases\u003C\u002Fli>\n\u003Cli>AI assistants\u003C\u002Fli>\n\u003Cli>Semantic search engines\u003C\u002Fli>\n\u003Cli>Advanced RAG solutions\u003C\u002Fli>\n\u003Cli>Smarter helpdesks\u003C\u002Fli>\n\u003Cli>Knowledge-driven chatbots\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>Where Large Language Models handle the generation of answers, embeddings ensure that the right knowledge is found.\u003C\u002Fp>\n\u003Cp>And it is precisely that combination that makes business AI truly valuable.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2>Want to know more?\u003C\u002Fh2>\n\u003Cp>Would you like to know how to combine Gemini Embedding 2 with FileMaker, a vector database, and an AI model such as Gemini, Claude, or ChatGPT?\u003C\u002Fp>\n\u003Cp>Get in touch with Loggix. We help organisations build AI solutions that leverage their own data, processes, and knowledge.\u003C\u002Fp>\n","Jeroen","2026-06-09",1781001771000,[16,17,18],"RAG","Vectorstore","Ai","\u002Fapi\u002Fknowledge\u002Fimage\u002F23\u002F?v=da8a1247cd2b",false,""]