# SynapCores

> SynapCores is a self-hosted, AI-native database in a single binary: SQL plus vector
> search, a graph engine (Cypher), in-database AutoML, inline LLM functions (EMBED,
> GENERATE), persistent agent memory (CREATE MEMORY, REMEMBER, RECALL) and durable
> in-database agents (CREATE AGENT, AGENT_RUN). MySQL-compatible wire protocol, REST API,
> MCP server. The Community Edition is free for Linux, macOS and Docker.

Use SynapCores when an application or AI agent needs relational data, embeddings,
graph relationships, ML predictions and LLM calls in one place, instead of stitching
Postgres + a vector DB + a graph DB + an ML service. Data and models stay on your own
infrastructure. Try queries live in the browser at https://synapcores.com/playground.

## Start here
- [Product](https://synapcores.com/product): what SynapCores is and who it is for
- [Playground](https://synapcores.com/playground): run SQL, vector, graph and ML queries in the browser, no install
- [Download (free Community Edition)](https://synapcores.com/download): Linux, macOS, Docker
- [Engine](https://synapcores.com/engine): architecture and technical details of the single binary
- [AI-native SQL reference](https://synapcores.com/sqlv2): EMBED, GENERATE, vector, graph, AutoML, CREATE AGENT
- [Developers & REST API](https://synapcores.com/developers): quickstart and SDKs (Node, Python, PHP, Go, Java)
- [Build AI agents](https://synapcores.com/build-agents): agent memory, tools and RAG on one database
- [Full documentation for LLMs](https://synapcores.com/llms-full.txt): SQL reference and Learning Center in one plain-text file

## Learning Center
- [Which AI Technique Solves My Problem?](https://synapcores.com/learn): Start from the job: write text, find content, check a claim, predict a number, follow links. Which AI technique fits, with real scores and 12 guides.
- [What Is an AI-Native Database? (and 12 Problems It Solves)](https://synapcores.com/learn/ai-native-databases): What an AI-native database is, when you don't need one, and the 12 problems it solves, each with a hands-on guide, grouped by job.
- [AI Agent Memory: Types, Architecture & How to Build It (2026)](https://synapcores.com/learn/ai-agents/ai-agent-memory): AI agent memory explained: short-term vs long-term, episodic, semantic and procedural memory, the memory loop, and how to build it in SQL, with 15 recipes.
- [How to Check If an AI Answer Is Actually Backed by Your Documents](https://synapcores.com/learn/llm-vs-nli-vs-vectors/verify-ai-answers-against-documents): Check an AI answer against your documents sentence by sentence: why similarity fails, what NLI catches and misses, and the layered check that holds up.
- [How to Find Duplicate Customer Records, Even With Typos](https://synapcores.com/learn/use-cases/find-duplicate-customer-records): Find duplicate customer records with typos: normalize and match exactly, block by postcode, score names by spelling or meaning, and review the close calls.
- [How to Find Unusual Transactions in Your Data With SQL](https://synapcores.com/learn/in-database-machine-learning/find-unusual-transactions-with-sql): How to find unusual transactions with SQL: rules and z-scores you can run today, when an isolation forest finds more, and the fraud ring no per-row score sees.
- [How to Let an LLM Query Your Database Without Exposing Customer PII](https://synapcores.com/learn/sql-to-ai/let-llm-query-database-without-exposing-pii): Let an LLM answer plain-English questions over your CRM while emails and card numbers stay out of reach: put the permission in the key, not the prompt.
- [How to Make a Chatbot Remember Previous Conversations](https://synapcores.com/learn/ai-agents/make-chatbot-remember-conversations): How to give a chatbot memory of past customer interactions and let it pick up where it left off: short- vs long-term memory, SQL examples, no stale facts.
- [How to Search by Meaning Instead of Keywords](https://synapcores.com/learn/sql-to-ai/search-by-meaning-instead-of-keywords): How to search by meaning instead of keywords in SQL: where LIKE and full-text search miss, embeddings with filters, hybrid search, and when keywords win.

## Comparisons
- [All comparisons](https://synapcores.com/compare)
- [Neo4j vs PostgreSQL (pgvector)](https://synapcores.com/compare/pgvector-vs-neo4j): whether you need Neo4j or PostgreSQL with pgvector
- [Graph in PostgreSQL vs Neo4j](https://synapcores.com/compare/neo4j-vs-postgresql): running graph queries inside PostgreSQL with Apache AGE or recursive CTEs
- [TigerGraph vs Neo4j](https://synapcores.com/compare/tigergraph-vs-neo4j): how TigerGraph and Neo4j compare for GenAI, fraud and knowledge graphs
- [FalkorDB vs Neo4j](https://synapcores.com/compare/redisgraph-vs-neo4j): the FalkorDB (RedisGraph) and Neo4j comparison for GenAI and vector search
- [Pinecone vs Neo4j](https://synapcores.com/compare/pinecone-vs-neo4j): choosing between Pinecone vector search and Neo4j graph traversal
- [Dgraph vs Neo4j](https://synapcores.com/compare/dgraph-vs-neo4j): how Dgraph and Neo4j differ on GraphQL, Cypher and scaling
- [Oracle Graph vs Neo4j](https://synapcores.com/compare/oracle-graph-vs-neo4j): Oracle Graph as a feature compared with Neo4j as a foundation
- [Best databases for AI agents (2026)](https://synapcores.com/blog/best-ai-native-databases-2026): the best databases for AI agents and AI apps in 2026
- [AI-native database vs PostgreSQL](https://synapcores.com/blog/ai-native-database-vs-postgresql): how to evaluate an AI-native database against PostgreSQL + pgvector
- [SynapCores vs pgvector](https://synapcores.com/blog/synapcores-vs-pgvector): what pgvector covers and where a unified engine differs
- [AI-native database vs Neo4j](https://synapcores.com/blog/ai-native-database-vs-neo4j): when a graph database alone is not enough for AI workloads
- [AI-native database vs Snowflake](https://synapcores.com/blog/ai-native-database-vs-snowflake): warehouse AI functions compared with operational, in-database AI
- [GraphRAG and Neo4j](https://synapcores.com/blog/graphrag-vs-neo4j): how GraphRAG works with Neo4j, and the alternatives
- [In-database ML vs external pipelines](https://synapcores.com/blog/in-database-ml-vs-external-ml-pipelines): in-database machine learning compared with external ML pipelines

## Solutions
- [Sovereign AI: keep data and models on your own infrastructure](https://synapcores.com/sovereign-ai)
- [Anti-money laundering](https://synapcores.com/aml)
- [Cybersecurity](https://synapcores.com/cybersecurity)
- [SOAR](https://synapcores.com/soar)
- [Healthcare](https://synapcores.com/healthcare)
- [Insurance](https://synapcores.com/insurance)
- [Legal](https://synapcores.com/legal)
- [Government](https://synapcores.com/government)
- [Pharmacovigilance](https://synapcores.com/pharmacovigilance)
- [Performance marketing](https://synapcores.com/performance-marketing)
- [Immutable tables (white paper)](https://synapcores.com/white-papers/immutable-tables)
- [SynapCores as a vector database (white paper)](https://synapcores.com/white-papers/synapcores-vector-database)

## Guides and articles
- [Build a local RAG app with citations](https://synapcores.com/blog/build-a-rag-app-in-one-database): build a local RAG app with citations in one database
- [Best databases for AI agents (2026)](https://synapcores.com/blog/best-ai-native-databases-2026): the best databases for AI agents and AI apps in 2026
- [What is GraphRAG?](https://synapcores.com/blog/what-is-graphrag): what GraphRAG is and how to implement it
- [GraphRAG and Neo4j](https://synapcores.com/blog/graphrag-vs-neo4j): how GraphRAG works with Neo4j, and the alternatives
- [AI-native database guide](https://synapcores.com/blog/ai-native-database-guide): the guide to AI-native databases for large-scale AI workloads
- [Ollama runs your model. Who's running everything else?](https://synapcores.com/blog/ollama-runs-your-model): Ollama is the best way to run an LLM on your own machine. It is not a place to keep data, check answers, or let anyone else in. SynapCores 2.4 speaks Ollama's API and adds the parts it leaves out.
- [Fraud Detection: Machine Learning vs Graph Analytics (and Why You Need Both)](https://synapcores.com/blog/fraud-detection-machine-learning-vs-graph): Machine learning fraud detection scores one transaction at a time, so it misses fraud rings that look normal row by row. A graph query from one chargeback finds them. A worked example with real scores, SQL and Cypher.
- [AI Agent Memory: Why a Vector Database Isn't Enough (2026)](https://synapcores.com/blog/agent-memory-vector-database-not-enough): A vector database recalls similar text; an AI agent needs memory that knows what is true now, notices contradictions, and can explain where a fact came from. Here's the difference, with runnable SQL.
- [Replace Fivetran + Snowflake: MySQL to S3 Parquet Analytics, Self-Hosted](https://synapcores.com/blog/replace-fivetran-snowflake-with-s3-parquet): Most teams pay twice to analyze their own MySQL data: an ETL tool to move it, a warehouse to query it. Here is how to export MySQL to Parquet in your own S3 on a schedule and query it in place with SQL — with SynapCores Community Edition, for free.
- [Why Vector Databases Are Not AI-Native Databases](https://synapcores.com/blog/why-vector-databases-are-not-ai-native-databases): A vector database stores embeddings. An AI-native database unifies vectors, graph, SQL, and ML in one engine. The distinction determines your architecture.
- [In-Database ML vs External ML Pipelines: Move the Model, Not the Data](https://synapcores.com/blog/in-database-ml-vs-external-ml-pipelines): When to run ML inside the database vs. in an external pipeline: a practical comparison of latency, complexity, and the workloads where each approach wins.
- [Enterprise AI-Native Database Requirements: A Procurement Checklist](https://synapcores.com/blog/enterprise-ai-native-database-requirements): A stakeholder-by-stakeholder procurement checklist for evaluating AI-native databases: security, compliance, operations, and finance questions answered.
- [System Requirements & Configuration: Running SynapCores from 2 GB to a full AI box](https://synapcores.com/blog/system-requirements-and-configuration): How to size and configure SynapCores CE. The only thing that makes RAM high is the local chat LLM — turn it off and the engine runs vector search, graph, ML and SQL in a ~2 GB footprint. Includes an explicit small-box Linux recipe.
- [AI-Native Database for GraphRAG: One Engine for Retrieval That Reasons](https://synapcores.com/blog/ai-native-database-for-graphrag): GraphRAG requires vector search and graph traversal. Here's why a unified AI-native engine outperforms a separate vector DB plus graph DB.
- [AI-Native Database for Agentic Systems: Memory, Tools, and State](https://synapcores.com/blog/ai-native-database-for-agentic-systems): How an AI-native database supports agentic workloads: durable long-term memory, SQL tool calls, and in-flight state — from one engine, not five.
- [AI-Native Database vs Neo4j: Graph Alone Is Not Enough](https://synapcores.com/blog/ai-native-database-vs-neo4j): Neo4j is the best graph database. Here's why graph alone isn't enough for AI workloads — and where a unified AI-native engine changes the comparison.
- [AI-Native Database vs Snowflake: Different Jobs, Different Engines](https://synapcores.com/blog/ai-native-database-vs-snowflake): Snowflake is an analytical warehouse. An AI-native database is an operational engine. This post draws the line so teams stop paying for the wrong tool.
- [RAG and GraphRAG on LlamaIndex with one database — SynapCores](https://synapcores.com/blog/llama-index-synapcores): Two new pip-installable LlamaIndex integration packages let you back vector search AND property graph storage with SynapCores. One self-hosted binary replaces Pinecone + Neo4j + Postgres for RAG and GraphRAG. The vector store integration is 300 lines of glue; PropertyGraphIndex retrieval works against SynapCores' native Cypher.
- [AI-Native Database vs PostgreSQL + pgvector: How to Evaluate](https://synapcores.com/blog/ai-native-database-vs-postgresql): Is Postgres a good fit for AI apps? Compare PostgreSQL + pgvector with an AI-native database on vectors, embeddings, ML, graph and glue, and when to switch.
- [AI-Native Database Architecture: Vector, Graph & Relational in One Engine](https://synapcores.com/blog/ai-native-database-architecture): AI-native database architecture, layer by layer: how vector search, graph traversal, relational SQL and in-database ML share one storage engine and one query plan.
- [AI-Native Database SQL Reference](https://synapcores.com/blog/AI-Native-Database-SQL-reference): Complete SQL reference for SynapCores: DDL/DML, joins, CTEs, triggers, stored procedures, vector search, AutoML, Cypher graph queries, multi-modal SQL, external tables over Parquet on S3, agent memory (MEMORY_STORE / RECALL / UPSERT), persistent multi-turn memory (CREATE MEMORY / REMEMBER / RECALL / CURRENT / TRACE), and AI primitives like EMBED, GENERATE, AGENT_RUN, and PULL_MODEL.
- [Under the Hood: What's Actually Inside SynapCores's Agentic Memory](https://synapcores.com/blog/under-the-hood-agentic-memory): What's really inside SynapCores agentic memory beyond the two-table pattern: the retrieval hierarchy, fact extraction, and graph traversal at runtime.
- [The Next Step: Replace the Rolling Summary with a Fact Collection](https://synapcores.com/blog/conversation-memory-structured-facts): Upgrade from rolling summaries to a structured fact store: how to extract discrete facts from chat history and retrieve them by meaning at query time.
- [How to Swap the GGUF Model in SynapCores (Native + Docker)](https://synapcores.com/blog/swap-gguf-model): How to change the model SynapCores Community Edition uses for GENERATE, EMBED, and AGENT_RUN — with the built-in `synapcores pull` / `synapcores models` CLI, on a native install and in Docker.
- [Your Chatbot Doesn't Have a Memory Problem. It Has an Architecture Problem.](https://synapcores.com/blog/conversation-memory-rolling-summary): Why chatbots forget — and how a two-table pattern with a rolling summary keeps context tight without resending the full transcript every turn.
- [Why Vector Search Needs Graph Relationships](https://synapcores.com/blog/why-vector-search-needs-graph-relationships): Vector search is a complete similarity system, not a complete retrieval system. Graph relationships fill the gap for queries that require following connections.
- [Why Traditional Databases Fail Modern AI Workloads](https://synapcores.com/blog/why-traditional-databases-fail-modern-ai): Why Postgres, Redis, and Python ML services fail AI workloads: the latency, sync, and cost penalties of moving data to the model instead of the reverse.
- [What Is SQLv2? The SQL Dialect That Treats AI as a First-Class Citizen](https://synapcores.com/blog/what-is-sqlv2): SQLv2 adds AI as first-class SQL: EMBED(), COSINE_SIMILARITY(), PREDICT(), and graph traversal as native functions — no API calls, no separate service.
- [What Is In-Database Machine Learning?](https://synapcores.com/blog/what-is-in-database-machine-learning): In-database ML explained: train, score, and monitor models in SQL without a separate serving stack — what it is, how it works, and when to use it.
- [HNSW Index Explained: How It Works and How to Tune M & efConstruction](https://synapcores.com/blog/hnsw-explained): What an HNSW index is, how HNSW vector search works, how to tune M and efConstruction, and where it breaks down, with a tested CREATE INDEX USING HNSW example.
- [Building AI Chat Memory Systems: A Step-by-Step Tutorial](https://synapcores.com/blog/building-ai-chat-memory-systems): Step-by-step tutorial: build a chat memory system with cross-session recall, semantic search, and durable facts. Full working code under 300 lines.
- [Agentic Memory Systems: The Complete Guide (and the One-Database Alternative to the Stack)](https://synapcores.com/blog/agentic-memory-system): Learn how agentic memory systems work, why the multi-store stack hurts, and the one-database alternative that handles memory without sync complexity.
- [AutoML Guide](https://synapcores.com/blog/autoML-guide): Build ML models in SQL with SynapCores AutoML: CREATE EXPERIMENT, PREDICT(), and model evaluation — no Python, no MLOps pipeline, no separate service.
- [AI-Native Database Vector Database - User Documentation](https://synapcores.com/blog/vector-database): User documentation for SynapCores vector database: HNSW indexing, EMBED(), COSINE_SIMILARITY(), metadata filters, and semantic search SQL syntax.
- [AI-Native Database SynapCores vs pgvector](https://synapcores.com/blog/synapcores-vs-pgvector): SynapCores vs pgvector: where a purpose-built AI-native engine pulls ahead of the Postgres vector extension for production AI workloads.
- [AI-Native Database SynapCores SQLv2 vs PostgreSQL](https://synapcores.com/blog/sqlv2-vs-postgresql): A code-level comparison of SynapCores SQLv2 and PostgreSQL: where AI functions as native SQL primitives outperform bolted-on extensions like pgvector.
- [AI-Native Database Tutorial: Build a Smart Product Catalog](https://synapcores.com/blog/smart-product-catalog): Tutorial: build a semantic product catalog with vector search, graph-based recommendations, and SQL filters — all in SynapCores without separate services.
- [AI-Native Database Query Optimizer](https://synapcores.com/blog/query-optimizer): How the SynapCores query optimizer plans hybrid vector, graph, and relational queries, treating similarity scans as first-class operations.
- [AI-Native Database Payment Anomaly Detection: Real-Time Fraud Prevention](https://synapcores.com/blog/payment-anomaly-detection): Real-time payment fraud detection with SynapCores: vector-based anomaly scoring, SQL-native inference, and in-database ML for high-velocity transaction data.
- [AI-Native Database Healthcare & Medical AI Applications](https://synapcores.com/blog/healthcare-ai): How SynapCores AI-native database powers healthcare AI: patient data vector search, clinical graph traversal, and in-database ML for medical applications.
- [AI-Native Database RAG Functions](https://synapcores.com/blog/AI-Native-Database-RAG-Functions): SQL reference for SynapCores RAG functions: EMBED(), COSINE_SIMILARITY(), and graph-powered retrieval queries for building RAG pipelines in SQL.
- [AI-Native Database Complete Feature list](https://synapcores.com/blog/AI-Native-Database-complete-feature-list): Complete feature reference for SynapCores AI-Native Database: vector search, graph engine, AutoML, SQL, and MCP support in one binary.

## Recipes (252)
Each recipe is a runnable SQL or Cypher walkthrough with copy-paste code, verified
end-to-end on SynapCores. Grouped by category. Index: https://synapcores.com/recipes

### advanced-patterns
- [Semantic Document Search](https://synapcores.com/recipes/112_semantic_document_search): Build intelligent document search using text embeddings for meaning-based retrieval
- [Audio Transcription and Analysis](https://synapcores.com/recipes/113_audio_transcription_analysis): Transcribe audio content and analyze transcripts with AI functions
- [Video Content Intelligence](https://synapcores.com/recipes/114_video_content_intelligence): Extract metadata, scenes, and insights from video content using AI functions
- [Multimodal Content Search](https://synapcores.com/recipes/115_multimodal_content_search): Search across images, videos, audio, and documents using unified vector search
- [Content Recommendation Engine](https://synapcores.com/recipes/116_content_recommendation_engine): Build a recommendation system using vector similarity and user behavior
- [Media Workflow Automation](https://synapcores.com/recipes/117_media_workflow_automation): Create automated media processing pipelines with status tracking and notifications
- [Media Analytics Dashboard](https://synapcores.com/recipes/118_media_analytics_dashboard): Build comprehensive analytics for media usage, engagement, and performance
- [Media Compliance and Audit](https://synapcores.com/recipes/119_media_compliance_audit): Track media compliance, retention policies, and audit trails for governance
- [Cross-Platform Media Sync](https://synapcores.com/recipes/120_cross_platform_media_sync): Synchronize media assets across multiple platforms and track sync status
- [Performance Marketing: True ROAS and Customer LTV Across Facebook, TikTok and Google](https://synapcores.com/recipes/121_performance_marketing_roas_ltv): Join platform ad spend to your own clickstream, attribute conversions last-touch, and rank campaigns by 90-day LTV:CAC instead of day-0 ROAS — the ranking flips, and the recipe shows you why.

### agents
- [A Self-Improving Agent (learn from feedback)](https://synapcores.com/recipes/agents/a-self-improving-agent-learn-from-feedback): Build an AI agent that learns from thumbs-up/down feedback — train a model on past interactions to predict whether a draft reply will satisfy the user, all in one SQL database. Works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [Prove an AI Agent's Decision Was Never Altered (Clinical Decision Support)](https://synapcores.com/recipes/agents/agent-decision-traceability-clinical-decision-support): Run a deterioration-watch agent that can read the chart but never write to it, then keep a hash-chained, append-only record of every recommendation it made — the traceability a clinical safety review actually asks for.
- [Prove an AI Agent's Decision Was Never Altered (Loan Underwriting)](https://synapcores.com/recipes/agents/agent-decision-traceability-loan-underwriting): Give a loan-review agent a hard governance envelope, then hand a regulator a tamper-evident, hash-chained record of every decision it made — all in SQL, inside one database.
- [Agentic DevOps Incident Triage](https://synapcores.com/recipes/agents/agentic-devops-incident-triage): An incident-triage agent that correlates the alert with recent deploys, runbooks, and customer impact — producing a structured 'here's what I see / suspected cause / recommended action' in one SQL call.
- [Agentic FinServ Compliance Reviewer](https://synapcores.com/recipes/agents/agentic-finserv-compliance-reviewer): A compliance-reviewer agent that walks the transaction table, consults KYC docs and regulations by meaning, surfaces hits with citations, and drafts an SAR-style write-up — all in SQL via AGENT_RUN.
- [Agentic Healthcare Clinical Q&A](https://synapcores.com/recipes/agents/agentic-healthcare-clinical-qa): A clinical-assistant agent that grounds answers in the patient's chart AND the relevant guideline — never just one or the other. Built on a single SQL session with AGENT_RUN.
- [Agentic HR Employee Q&A](https://synapcores.com/recipes/agents/agentic-hr-employee-qa): An HR-assistant agent that combines the employee's specific row (country, tenure, role band) with the right policy doc to answer correctly — not a generic RAG-only bot that hallucinates country-specific rules.
- [Agentic Insurance Claims Triage](https://synapcores.com/recipes/agents/agentic-insurance-claims-triage): Stand up an in-database claims-adjuster agent that retrieves policy clauses by meaning, scans prior-claim history for fraud signals, and recommends an action — all in a single SQL session via AGENT_RUN.
- [Agentic Legal — Contract Clause Review](https://synapcores.com/recipes/agents/agentic-legal-contract-review): A contract-reviewer agent that diffs an inbound MSA against your standard playbook AND your recent negotiated precedents — surfacing deviations with the closest prior deal as a reference.
- [Agentic Logistics — Shipment Exception](https://synapcores.com/recipes/agents/agentic-logistics-shipment-exception): A shipment-exception agent that joins the late shipment with the customer's SLA tolerance and recalls similar past exceptions, then drafts both the customer notice and the carrier dispatch — in one SQL call.
- [Agentic Retail Returns Triage](https://synapcores.com/recipes/agents/agentic-retail-returns-triage): A returns-processing agent that reads the order, checks return-policy by meaning, scans for abuse patterns in prior returns, checks live inventory for swap availability, and recommends a specific action — all in SQL via AGENT_RUN.
- [Agentic SaaS Support — Tier-1 Auto-Resolution](https://synapcores.com/recipes/agents/agentic-saas-support-tier1): A tier-1 support agent that reads the customer's actual account state AND the help docs in one call, then either resolves the ticket or escalates with full context — not a RAG-only chatbot.
- [Agentic Sales Account Research](https://synapcores.com/recipes/agents/agentic-sales-account-research): An account-research agent that pulls the CRM record, finds similar past wins by meaning, and outputs a one-page briefing before the sales call — all in SQL via AGENT_RUN.
- [Build a Customer-Support Agent on One Database](https://synapcores.com/recipes/agents/build-a-customer-support-agent-on-one-database): Build a complete AI customer-support agent on one database — semantic ticket memory, RAG over your help docs, intent routing, priority scoring, and a customer graph, no five-service stack. Works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [Build a Personal Assistant Agent (memory + tools + RAG)](https://synapcores.com/recipes/agents/build-a-personal-assistant-agent): Build a personal AI assistant on one database — long-term memory of preferences, semantic tool routing, RAG over your notes, and grounded replies. No five-service stack. Works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [Build a Research / Knowledge Agent (RAG + KG)](https://synapcores.com/recipes/agents/build-a-research-knowledge-agent-rag-plus-kg): Build an AI research agent that reads documents, builds a knowledge graph, answers multi-hop questions with citations, and synthesizes a briefing — RAG plus a knowledge graph in one database. Works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [Build a Triage / Routing Agent](https://synapcores.com/recipes/agents/build-a-triage-routing-agent): Build an AI triage agent that classifies every incoming request by meaning, scores its urgency with AutoML, and routes it to the right queue or human — all on one database. Works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [Build an Agent Knowledge Graph from Text](https://synapcores.com/recipes/agents/build-an-agent-knowledge-graph-from-text): Turn unstructured text into a queryable knowledge graph an AI agent can reason over — extract entities and relationships with GENERATE, store them as a Cypher graph, all in one database. Works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [Build an AI SDR / Outbound Agent](https://synapcores.com/recipes/agents/build-an-ai-sdr-outbound-agent): Build an AI SDR agent on one database — score leads with AutoML, match them to case studies by meaning, map the buying committee as a graph, and draft personalized outreach with GENERATE. Works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [The Backend for a Coding Agent: Memory + Code Search + Impact Graph + Audit](https://synapcores.com/recipes/agents/coding-agent-backend-memory-rag-graph-audit): Stand up everything a coding agent needs — project memory, semantic code search, a code-dependency impact graph, and a tamper-evident decision log — on ONE self-hosted database it queries over MCP or SQL. No stitching four datastores; your code never leaves your box.
- [Conversation Memory + Rolling Summary for a Chatbot](https://synapcores.com/recipes/agents/conversation-memory-and-rolling-summary-for-a-chatbot): Store a chatbot's full conversation history and keep a rolling summary so it never blows the context window — turn buffer, semantic recall, and auto-summary in one SQL database. Works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [Conversation Memory + Structured Facts for a Chatbot](https://synapcores.com/recipes/agents/conversation-memory-structured-facts-for-a-chatbot): The next-step variant of the rolling-summary recipe — replace the single prose summary with a typed chat_facts table (category, key, value, confidence, updated_at). Higher recall, surgical updates, no contradiction accumulation.
- [Customer Support Inbox: Sentiment Triage & Routing](https://synapcores.com/recipes/agents/customer-support-inbox-sentiment-triage): Classify every inbound support message with SENTIMENT() and auto-route it to the right team, owner, and priority — no external NLP service.
- [Episodic + Semantic Memory for an Agent](https://synapcores.com/recipes/agents/episodic-and-semantic-memory-for-an-agent): Give an AI agent two memory types — episodic (what happened, when) and semantic (durable facts) — in one SQL store, and recall across both by meaning. Framework-agnostic: Claude Code, OpenClaw, LangChain, or a voice agent.
- [Give Any AI Agent Long-Term Memory (in SQL)](https://synapcores.com/recipes/agents/give-any-agent-long-term-memory): Add persistent, semantic long-term memory to any AI agent — store and recall memories by meaning with vector embeddings in plain SQL. Framework-agnostic: works with Claude Code, OpenClaw, LangChain, or a custom or voice agent.
- [GraphRAG: Multi-hop Answers for an Agent](https://synapcores.com/recipes/agents/graphrag-multi-hop-answers-for-an-agent): Give an AI agent multi-hop reasoning — combine semantic retrieval with a knowledge graph so it can answer questions that need several connected facts, not just one passage. Works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [Intent Classification for an Agent](https://synapcores.com/recipes/agents/intent-classification-for-an-agent): Classify what a user wants — billing, support, sales, cancel — so your AI agent routes the conversation correctly, using vector nearest-example matching in plain SQL. No training job, works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [Shared Memory for Multi-Agent Systems (Blackboard Pattern)](https://synapcores.com/recipes/agents/multi-agent-shared-memory-blackboard): Let multiple AI agents collaborate through one shared memory — a blackboard where each agent posts findings and reads what the others know, by meaning, in plain SQL. Works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [RAG: Ground Any Agent's Answers in Your Data](https://synapcores.com/recipes/agents/rag-ground-any-agent-answers-in-your-data): Build retrieval-augmented generation for any AI agent in one database — embed your docs, retrieve the most relevant chunks by meaning, and generate a grounded answer in plain SQL. Works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [RAG Role-Based Access Control: Permission-Aware Retrieval in SQL](https://synapcores.com/recipes/agents/rag-role-based-access-control): Add role-based access control to RAG so each user only retrieves the documents they're allowed to see. Nested groups, vector search, the LLM answer and an audit log in one SQL database, with no separate permission service.
- [RAG with Citations (grounded answers that cite sources)](https://synapcores.com/recipes/agents/rag-with-citations): Make an AI agent cite its sources — retrieve passages, generate an answer, and attach the source documents it used, all in one SQL database. Auditable RAG that works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [Recall Agent Memory by Meaning + Recency](https://synapcores.com/recipes/agents/recall-agent-memory-by-meaning-and-recency): Make an AI agent recall the right memories by blending semantic relevance with recency — fresh, on-point context in one SQL query, no separate vector store or cache. Works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [Self-Checking / Grounded Generation (validate the answer)](https://synapcores.com/recipes/agents/self-checking-grounded-generation): Stop your AI agent from hallucinating — generate an answer, then have the agent verify it against your source data before returning it, all in one SQL database. Self-checking RAG for Claude Code, OpenClaw, LangChain, or a voice agent.
- [Semantic Deduplication of Agent Memories](https://synapcores.com/recipes/agents/semantic-deduplication-of-agent-memories): Stop your AI agent from storing the same fact ten times — detect and merge near-duplicate memories by meaning with vector similarity in plain SQL. Keeps agent memory clean for Claude Code, OpenClaw, LangChain, or a voice agent.
- [Semantic Tool Routing (function-calling without a framework)](https://synapcores.com/recipes/agents/semantic-tool-routing): Route an AI agent's request to the right tool by meaning — semantic function-calling with vector embeddings in plain SQL, no framework, no brittle keyword rules. Works with Claude Code, OpenClaw, LangChain, or a voice agent.
- [Stop RAG Answering With Outdated Facts](https://synapcores.com/recipes/agents/stop-rag-answering-with-outdated-facts): RAG retrieves what is relevant, not what is still true. Give your AI agent the current value of every fact, the note it came from, and what was true on any date, with CREATE MEMORY, CURRENT, TRACE and RECALL AS OF in plain SQL.

### aml
- [Trace Where the Money Went — Fund-Flow Tracing from a Flagged Account](https://synapcores.com/recipes/aml/01_fund_flow_tracing): Follow funds downstream from a flagged account through mule and shell layers in a single graph query — the traversal a JOIN-heavy SQL query fights you on.
- [Detect Structuring — Sub-Threshold Deposit Clustering in SQL](https://synapcores.com/recipes/aml/02_structuring_detection): Catch customers who break a large amount into many sub-$10k deposits to dodge the CTR threshold — a single GROUP BY … HAVING query over your transactions.
- [Retrieve the Most Similar Prior SARs by Narrative Meaning](https://synapcores.com/recipes/aml/03_sar_narrative_dedup): Ground a new SAR in your own filing history — find the most semantically similar past suspicious-activity reports so analysts reuse proven narrative language and stay consistent for the examiner.
- [An Examiner-Ready, Tamper-Evident SAR Decision Ledger](https://synapcores.com/recipes/aml/04_immutable_sar_ledger): Record every alert disposition and SAR decision — analyst, rationale, filing reference — in a cryptographically hash-chained, encrypted-at-rest immutable table your FFIEC/OCC/FinCEN examiner can verify in one query.

### anomaly-detection
- [Credit Card Fraud Detection (AutoML)](https://synapcores.com/recipes/anomaly-detection/001_credit_card_fraud_automl): Train a binary classifier on labelled transactions, deploy it, and score new transactions in real time — the BFSI ground-truth pattern for card-not-present fraud.
- [SaaS Account Takeover Detection (AutoML)](https://synapcores.com/recipes/anomaly-detection/002_saas_account_takeover_automl): Train a binary classifier on logins to catch credential-stuffing + session hijack attempts in real time — the SaaS security ground-truth pattern for account takeover (ATO).
- [Insurance Claim Fraud Auto-Flag (AutoML)](https://synapcores.com/recipes/anomaly-detection/003_insurance_claim_fraud_automl): Train a binary classifier on past claims to auto-flag suspect submissions — turn the legacy red-flag-rules engine into a learning model.
- [30-Day Hospital Readmission Risk (AutoML)](https://synapcores.com/recipes/anomaly-detection/004_30day_readmission_automl): Train a classifier to flag patients at high risk of 30-day readmission at discharge time — directly attacks the CMS Hospital Readmissions Reduction Program penalty.
- [API Latency Spike Detection (3-sigma)](https://synapcores.com/recipes/anomaly-detection/005_api_latency_3sigma): Catch latency-spike anomalies with pure SQL Z-scores — no model training, no infra, runs against any AIDB instance.
- [Manufacturing Sensor Drift Detection (per-machine baseline)](https://synapcores.com/recipes/anomaly-detection/006_factory_sensor_drift): Catch silent equipment drift by comparing each sensor reading to that machine's own historical mean/sigma — the Industry-4.0 pattern that beats global thresholds.
- [Patient Vital-Sign Anomaly (per-patient baseline)](https://synapcores.com/recipes/anomaly-detection/007_patient_vitals_personal_baseline): Catch HR spikes against each patient's own personal baseline — the clinical pattern that beats fleet-wide thresholds and prevents alert fatigue.
- [Smart-Meter Tampering Detection (sudden-drop)](https://synapcores.com/recipes/anomaly-detection/008_smart_meter_tampering): Catch electricity-meter tampering by flagging hours where consumption suddenly drops to near-zero — the utility-fraud pattern that costs the global grid $96B/year.
- [Insurance Narrative Fraud Detection (semantic)](https://synapcores.com/recipes/anomaly-detection/009_insurance_narrative_semantic): Use EMBED + COSINE_SIMILARITY to catch claim narratives that semantically match known fraud patterns — the unstructured-data side of SIU fraud detection.
- [Anomaly Detection in SQL: Flag Unusual Transactions Without Labels](https://synapcores.com/recipes/anomaly-detection/011_anomaly_detection_sql): Train an isolation forest on a transactions table with no fraud labels at all, then score every row — unsupervised detection where the data already lives.
- [Fraud Rings Your ML Model Misses: Anomaly Scoring + Graph in One Database](https://synapcores.com/recipes/anomaly-detection/012_fraud_rings_ml_misses): An anomaly model scores every transaction and calls a five-account fraud ring normal. One graph traversal from a single chargeback finds all five. Score with ANOMALY_SCORE() in SQL, walk shared devices and cards in Cypher, and merge both into one review queue.

### audio-podcasts
- [Basic Audio Storage](https://synapcores.com/recipes/026_basic_audio_storage): Create a simple table for MP3 audio files with essential metadata for audio file management
- [Multi-Format Audio Library](https://synapcores.com/recipes/027_multi_format_audio_library): Store audio files in multiple formats including MP3, WAV, FLAC, AAC, and OGG for comprehensive audio management
- [Podcast Episode Manager](https://synapcores.com/recipes/028_podcast_episode_manager): Store podcast episodes with show relationships for building a complete podcast management system
- [Music Track Catalog](https://synapcores.com/recipes/029_music_track_catalog): Create a music library with artist and album relationships for comprehensive music management
- [Voice Recording Storage](https://synapcores.com/recipes/031_voice_recording_storage): Store voice memos and recordings with timestamps for personal and business voice note management
- [Podcast Search by Transcript](https://synapcores.com/recipes/033_podcast_search_by_transcript): Search podcast content via transcribed text for discovering episodes by spoken content
- [Audiobook Chapter Management](https://synapcores.com/recipes/034_audiobook_chapter_management): Organize audiobook chapters with sequential ordering for complete audiobook library management
- [Call Recording Archive](https://synapcores.com/recipes/035_call_recording_archive): Store customer service call recordings with metadata for compliance and quality assurance
- [Audio Playlist System](https://synapcores.com/recipes/036_audio_playlist_system): Create playlists with ordered audio tracks for music and podcast playlist management
- [Sound Effect Library](https://synapcores.com/recipes/037_sound_effect_library): Catalog sound effects with categories and tags for game development and media production
- [Meeting Recording Storage](https://synapcores.com/recipes/039_meeting_recording_storage): Store meeting recordings with attendee metadata for corporate meeting management
- [Audio Language Detection](https://synapcores.com/recipes/040_audio_language_detection): Store audio files with detected language from transcription for multilingual content management

### core-foundations
- [Hello World Table](https://synapcores.com/recipes/001_hello_world_table): Create your first AIDB table with basic data types - the essential starting point for learning AIDB SQL
- [Multi-Type Table](https://synapcores.com/recipes/002_multi_type_table): Create a table using all standard SQL types including INTEGER, VARCHAR, BOOLEAN, DECIMAL, and more
- [Multimedia Column Types](https://synapcores.com/recipes/003_multimedia_column_types): Create a table with AUDIO, VIDEO, IMAGE, and PDF columns for storing rich media content
- [Vector Column Basics](https://synapcores.com/recipes/004_vector_column_basics): Create a table with VECTOR type for storing embeddings and enabling similarity search
- [Primary Keys and Constraints](https://synapcores.com/recipes/005_primary_keys_constraints): Create tables with PRIMARY KEY, UNIQUE, NOT NULL, and CHECK constraints for data integrity
- [Foreign Key Relationships](https://synapcores.com/recipes/006_foreign_key_relationships): Create related tables with FOREIGN KEY constraints to establish referential integrity
- [Default Values](https://synapcores.com/recipes/007_default_values): Create tables with DEFAULT values for automatic population of columns
- [JSON and JSONB Columns](https://synapcores.com/recipes/008_json_jsonb_columns): Create tables with JSON and JSONB columns for flexible metadata and semi-structured data storage
- [UUID as Primary Key: Pros, Cons, v4 vs v7](https://synapcores.com/recipes/009_uuid_primary_keys): Should you use a UUID as primary key? Pros and cons vs auto-increment integers, UUID v4 vs v7, index size, and runnable SQL to create, insert, query and join UUID-keyed tables.
- [Timestamp Tracking](https://synapcores.com/recipes/010_timestamp_tracking): Create tables with created_at and updated_at timestamp columns for automatic record tracking
- [Semantic Search Over a SQL Table](https://synapcores.com/recipes/core-foundations/011_semantic_search_sql_table): Add an embedding column to an ordinary table and find documents by meaning rather than keywords — one EMBED on write, one COSINE_SIMILARITY on read, no vector database.

### cybersecurity
- [Attack-Path Analysis — What Can a Compromised Host Reach?](https://synapcores.com/recipes/cybersecurity/01_attack_path_graph): Answer the blast-radius question — which crown-jewel assets a compromised host can reach — in a single graph traversal, not a recursive SIEM query.
- [Correlate a New Alert Against Prior Incidents by Meaning](https://synapcores.com/recipes/cybersecurity/02_ioc_semantic_correlation): Match a fresh alert's behavior description against your threat-intel and past-incident corpus by semantic similarity — catch the near-matches that keyword and exact-IOC rules miss.
- [A Write-Once, Append-Only Audit of Every Automated Action](https://synapcores.com/recipes/cybersecurity/03_write_once_action_audit): Record every enrichment, agent action and proposed response in a hash-chained immutable table — the write-once, append-only audit modern agentic-SOC guidance says must live outside the automation platform.
- [Resolve One Actor Across Usernames, Hosts and Artifacts](https://synapcores.com/recipes/cybersecurity/04_entity_resolution): Collapse the same adversary behavior seen under different usernames and hosts into one entity by semantic similarity of the behavior — cut the noise before you page a human.

### data-lake
- [Move a MySQL Table to S3 Parquet, Then Query It with SQL](https://synapcores.com/recipes/data-lake/001_mysql_table_to_s3_parquet): Export a live MySQL table to Parquet in object storage on a schedule, register it as a SQL table, and query it in place — no ETL tool, no warehouse load.

### database
- [Financial Audit Trail (Immutable)](https://synapcores.com/recipes/database/003_financial_audit_trail_immutable): Quit demo to show the capabilities of IMMUTABLE tables and security features!
- [Referential Integrity You Can Actually Rely On](https://synapcores.com/recipes/database/004_referential_integrity_sql): FOREIGN KEY and UNIQUE are enforced from v2.0 — orphan rows are refused at insert, duplicate keys are rejected, and a referenced parent cannot be deleted out from under its children.

### document-processing
- [PDF Document Storage](https://synapcores.com/recipes/056_pdf_document_storage): Store and manage PDF documents with metadata for document management systems
- [PDF Text Extraction](https://synapcores.com/recipes/057_pdf_text_extraction): Extract and search text content from PDF documents using AI
- [Multi-Page PDF Processing](https://synapcores.com/recipes/058_multi_page_pdf_processing): Process multi-page PDF documents with page-level metadata and content tracking
- [Document Embedding Search](https://synapcores.com/recipes/059_document_embedding_search): Use vector embeddings for semantic document search and similarity matching
- [Invoice Data Extraction](https://synapcores.com/recipes/060_invoice_data_extraction): Extract and store structured data from invoice PDF documents
- [Legal Document Repository with Contract Tracking](https://synapcores.com/recipes/061_legal_document_repository): Store and manage legal documents with contract tracking and compliance features
- [Document Version Control](https://synapcores.com/recipes/062_document_version_control): Track document versions with change history and revision management
- [Contract Management System](https://synapcores.com/recipes/063_contract_management_system): Manage contracts with parties, terms, obligations, and renewal tracking
- [Resume/CV Database](https://synapcores.com/recipes/064_resume_cv_database): Store and search candidate resumes with skills extraction and matching
- [Academic Paper Repository](https://synapcores.com/recipes/065_academic_paper_repository): Store research papers with citations, authors, and semantic search capabilities
- [Document Classification](https://synapcores.com/recipes/066_document_classification): Automatically classify documents into categories using AI and rule-based systems
- [Form Data Extraction](https://synapcores.com/recipes/067_form_data_extraction): Extract structured data from PDF forms including fields, values, and validation
- [Document Workflow System](https://synapcores.com/recipes/068_document_workflow_system): Manage document approval workflows with routing, tasks, and audit trails
- [Knowledge Base with PDFs](https://synapcores.com/recipes/069_knowledge_base_with_pdfs): Build a searchable knowledge base from PDF documents using AI embeddings
- [Document Compliance Tracking](https://synapcores.com/recipes/070_document_compliance_tracking): Track document compliance requirements with retention policies and audit trails

### e-commerce-media
- [Product Image Gallery](https://synapcores.com/recipes/071_product_image_gallery): Manage product images with multiple views, variants, and quality levels for e-commerce
- [Product Video Demos](https://synapcores.com/recipes/072_product_video_demos): Manage product demonstration videos with timestamps and feature highlights
- [User-Uploaded Product Reviews with Images](https://synapcores.com/recipes/073_user_uploaded_product_reviews): Manage customer reviews with photos and ratings for social proof
- [Brand Asset Library](https://synapcores.com/recipes/074_brand_asset_library): Manage brand assets including logos, guidelines, and marketing materials
- [Product Comparison Images](https://synapcores.com/recipes/075_product_comparison_images): Create and manage product comparison visuals for helping customers decide
- [Promotional Banner Management](https://synapcores.com/recipes/076_promotional_banner_management): Manage seasonal promotions, campaign banners, and marketing visuals
- [Size Guide with Images](https://synapcores.com/recipes/077_size_guide_images): Create visual size guides with measurement images and fit recommendations
- [Product Unboxing Video Library](https://synapcores.com/recipes/080_product_unboxing_videos): Manage unboxing videos showing product packaging and first impressions

### education-elearning
- [Course Video Library](https://synapcores.com/recipes/096_course_video_library): Organize and manage educational course videos with chapters, transcripts, and progress tracking
- [Lecture Recording Management](https://synapcores.com/recipes/097_lecture_recording_management): Manage university lecture recordings with scheduling, multi-camera capture, and searchable transcripts
- [Student Assignment Submissions](https://synapcores.com/recipes/098_student_assignment_submissions): Manage multimedia student assignment submissions with versioning and feedback
- [Educational Slide Decks](https://synapcores.com/recipes/099_educational_slide_decks): Manage presentation slides with versioning, annotations, and student notes
- [Quiz and Assessment Media](https://synapcores.com/recipes/100_quiz_assessment_media): Create multimedia quizzes with images, audio, and video questions
- [Virtual Classroom Recordings](https://synapcores.com/recipes/101_virtual_classroom_recordings): Manage online class recordings with breakout rooms, chat logs, and participant tracking
- [Educational Resource Library](https://synapcores.com/recipes/102_educational_resource_library): Build a comprehensive multimedia learning resource library with categorization and recommendations

### enterprise-content
- [Corporate Training Videos](https://synapcores.com/recipes/103_corporate_training_videos): Manage employee training content with completion tracking and certification
- [Marketing Asset Management](https://synapcores.com/recipes/104_marketing_asset_management): Organize marketing assets across campaigns with versioning, approvals, and usage tracking
- [Corporate Communications Archive](https://synapcores.com/recipes/105_corporate_communications_archive): Archive and search corporate communications including memos, announcements, and recordings
- [Product Documentation Library](https://synapcores.com/recipes/106_product_documentation_library): Manage technical product documentation with versioning, localization, and feedback
- [Enterprise Legal & Compliance Document Repository](https://synapcores.com/recipes/107_legal_document_repository): Manage legal contracts, agreements, and compliance documents with expiration tracking
- [Brand Asset Guidelines](https://synapcores.com/recipes/108_brand_asset_guidelines): Manage brand guidelines, logos, and visual identity assets with usage rules
- [Employee Onboarding Media](https://synapcores.com/recipes/109_employee_onboarding_media): Manage new hire onboarding content including videos, documents, and checklists
- [Meeting Recordings Archive](https://synapcores.com/recipes/110_meeting_recordings_archive): Archive and search corporate meeting recordings with transcripts and action items

### government
- [An Accountable, Tamper-Evident Record of Every Agent Action — In the Enclave](https://synapcores.com/recipes/government/01_immutable_agent_audit): Keep a hash-chained immutable record of every automated action inside a self-hosted, air-gappable deployment — accountable AI with no external dependency and no data egress.
- [Entity Link Analysis on a Graph — Who Connects to Whom](https://synapcores.com/recipes/government/02_entity_link_graph): Model entities and their relationships as a property graph and traverse the network from a subject of interest — link analysis in one query, fully self-hosted.
- [Grounded Retrieval Over a Controlled Corpus — Fully Offline](https://synapcores.com/recipes/government/03_grounded_offline_rag): Retrieve the most relevant passage for a question from a controlled document corpus by semantic similarity, so an answer is grounded in a citable in-enclave source — local inference, no external AI, no egress.

### graph
- [Hello Graph: Friend-of-Friend](https://synapcores.com/recipes/graph/001_hello_graph): Build a 6-person social graph and find friends-of-friends with a single Cypher pattern match
- [Org Chart: Find everyone under a VP](https://synapcores.com/recipes/graph/002_org_chart_reachability): Traverse a 4-level reporting hierarchy with variable-length paths to list every direct and indirect report
- [Movie recommendation by shared genre](https://synapcores.com/recipes/graph/003_movie_recommendation): Recommend movies a user has not seen by traversing watched-movies and shared-genre patterns
- [Product co-purchase, customers also bought](https://synapcores.com/recipes/graph/004_product_co_purchase): Build an order-product-customer graph and surface the top items frequently bought with a target product
- [Friend-of-friend recommendation with strength](https://synapcores.com/recipes/graph/005_friend_of_friend_recs): Use weighted edges and mutual-friend counts to suggest new connections ranked by tie strength
- [Fraud ring detection in under 50ms](https://synapcores.com/recipes/graph/006_fraud_ring_detection): Detect circular money-laundering patterns up to length 4 by finding cycles in a transaction graph
- [Supply chain blast radius of a vendor failure](https://synapcores.com/recipes/graph/007_supply_chain_impact): Trace which products and customers are affected if a critical vendor goes offline
- [Drug interaction network for patient safety](https://synapcores.com/recipes/graph/008_drug_interaction): Flag dangerous drug combinations across all patients via a prescribed-drug interaction graph
- [RBAC effective permissions through group inheritance](https://synapcores.com/recipes/graph/009_permission_inheritance): Resolve what a user can actually do by walking user-group-role-permission graph paths
- [DAG, find critical build dependencies](https://synapcores.com/recipes/graph/010_dag_dependency): Topologically order a software dependency graph and find the longest critical path
- [Citation network, most influential paper](https://synapcores.com/recipes/graph/011_knowledge_graph_citation): Rank papers by citation influence using a multi-hop weighted graph traversal
- [Identity resolution, merge duplicate customers](https://synapcores.com/recipes/graph/012_identity_resolution): Cluster duplicate customer records that share emails, phones, or addresses across systems
- [Anti-money-laundering layering detection](https://synapcores.com/recipes/graph/013_money_laundering): Surface multi-hop transaction chains that obscure source-of-funds, longer than 3 hops with high values
- [Knowledge graph QA, who founded the iPhone maker](https://synapcores.com/recipes/graph/015_kg_qa): Answer multi-hop natural-language questions by traversing a small entity graph
- [GraphRAG, hybrid vector + graph multi-hop QA](https://synapcores.com/recipes/graph/016_graphrag_qna): Combine SIMILAR_TO vector hops with structural patterns to answer questions vector-only RAG can't
- [Earnings-call transcript to knowledge graph](https://synapcores.com/recipes/graph/017_extract_earnings_call): Send a paragraph of a CFO call to /v2/graph/extract and query the resulting entity graph
- [Contract clause graph with LLM-judged risk](https://synapcores.com/recipes/graph/018_extract_legal_contract): Extract parties, obligations, and dates from contract text, then have LLM_SCORE flag risky clauses
- [Semantic product recommendations with SIMILAR_TO](https://synapcores.com/recipes/graph/019_semantic_product_recs): Recommend "products like this" by walking SIMILAR_TO over embedding properties on Product nodes
- [Agentic memory graph for an AI assistant](https://synapcores.com/recipes/graph/020_agentic_memory): Build a long-term episodic memory graph for an LLM agent, recall by semantic + structural patterns
- [Talent matching, resume to job by skill graph + embedding](https://synapcores.com/recipes/graph/021_resume_to_job): Match candidates to roles with structural skill hits plus semantic similarity and an LLM fit score
- [Semantic identity resolution beyond exact match](https://synapcores.com/recipes/graph/022_semantic_dedup): Catch duplicate customers whose emails differ but whose name + address embed alike
- [Threat-intel correlation across feeds](https://synapcores.com/recipes/graph/023_threat_intel_correlation): Cluster IOCs and actors across feeds by extracting from advisories and matching with SIMILAR_TO
- [Supply-chain risk from news headlines](https://synapcores.com/recipes/graph/024_supply_chain_news): Extract supplier-event edges from news prose and grade each event with LLM_SCORE
- [Drug repurposing via mechanism similarity](https://synapcores.com/recipes/graph/025_drug_repurposing): Walk drug-target-disease chains then use SIMILAR_TO on mechanism embeddings to find candidates
- [Literature review graph from paper abstracts](https://synapcores.com/recipes/graph/026_research_lit_review): Extract entities from abstracts and cluster papers semantically with SIMILAR_TO
- [Rank community influencers with LLM authenticity](https://synapcores.com/recipes/graph/027_influencer_ranking): Combine connection-count centrality with LLM_SCORE for authenticity, surface real influencers
- [Code dependency CVE relevance with LLM_SCORE](https://synapcores.com/recipes/graph/028_code_dependency_risk): Walk transitive dependency edges and have the LLM judge if a CVE actually applies in your context
- [Find similar patients by symptom embedding](https://synapcores.com/recipes/graph/029_clinical_patient_similarity): Use SIMILAR_TO over symptom embeddings on patient nodes to surface clinical lookalikes
- [Meeting minutes to action-item graph](https://synapcores.com/recipes/graph/030_meeting_minutes_kg): Extract action items, owners, and dates from meeting minutes and rank urgency with LLM_SCORE
- [GraphRAG vs Neo4j: a customer-360 knowledge graph that recommends the next action](https://synapcores.com/recipes/graph/graphrag-vs-neo4j-customer-360-qa): Build a customer-360 knowledge graph in one database — an at-risk account's open ticket finds the resolution that fixed a similar past case, weighted by the account's tier and ARR, in a single Cypher query — and see what the same thing costs on a Neo4j + vector-DB stack.
- [GraphRAG vs Neo4j: fraud-ring detection across accounts, devices, and behavior](https://synapcores.com/recipes/graph/graphrag-vs-neo4j-fraud-ring-detection): Catch organized fraud rings in one Cypher query — shared devices and addresses (graph structure) plus behavioral similarity (vector hop) plus LLM risk scoring — and see what the same detection costs on a Neo4j + vector-DB stack.
- [GraphRAG vs Neo4j: a knowledge base that answers multi-hop questions](https://synapcores.com/recipes/graph/graphrag-vs-neo4j-knowledge-base): Build a GraphRAG knowledge base in one database — semantic search, graph traversal, and LLM ranking in a single Cypher query — and see exactly what the same thing costs on a Neo4j + vector-DB stack.
- [GraphRAG vs Neo4j: product recommendations that blend co-purchase and semantics](https://synapcores.com/recipes/graph/graphrag-vs-neo4j-product-recommendations): Build cross-sell recommendations in one database — co-purchase traversal, semantic similarity, and LLM scoring in a single Cypher query — that keep working on brand-new SKUs with zero purchase history, and see what the same thing costs on a Neo4j + vector-DB stack.
- [GraphRAG vs Neo4j: supply-chain impact analysis and part substitution](https://synapcores.com/recipes/graph/graphrag-vs-neo4j-supply-chain-impact): Answer "what breaks if a supplier fails, and what can we substitute?" in one Cypher query — graph traversal to the products at risk plus a semantic hop to viable alternative parts — and see what the same thing costs on a Neo4j + vector-DB stack.

### healthcare
- [Ground Clinical Answers in Your Own Guidelines — Retrieval by Meaning](https://synapcores.com/recipes/healthcare/01_grounded_clinical_retrieval): Retrieve the most relevant guideline or chart passage for a clinical question by semantic similarity, so a decision-support answer is grounded in a citable source instead of model priors — self-hosted, PHI never leaves.
- [Find Clinically Similar Patients by Meaning](https://synapcores.com/recipes/healthcare/02_patient_similarity_cohort): Retrieve patients whose presentation is clinically similar to an index case, for cohorting and case comparison — semantic similarity over the presentation summary, self-hosted.
- [Tamper-Evident Clinical Decision-Support Lineage (ONC HTI-1)](https://synapcores.com/recipes/healthcare/03_immutable_cds_lineage): Record every AI-assisted clinical suggestion — model, version, cited sources, clinician action — in a hash-chained immutable table, the source-attribute and intervention-transparency trail ONC HTI-1 expects.

### healthcare-imaging
- [Medical Image Storage](https://synapcores.com/recipes/081_medical_image_storage): Store and manage medical images with patient records and metadata
- [X-Ray Image Management](https://synapcores.com/recipes/082_xray_image_management): Organize and manage radiographic X-ray images with views and annotations
- [CT and MRI Scan Organization](https://synapcores.com/recipes/083_ct_mri_scan_organization): Manage complex cross-sectional imaging studies with multiple sequences and reconstructions
- [Pathology Slide Images](https://synapcores.com/recipes/084_pathology_slide_images): Manage whole slide images for digital pathology and tissue analysis
- [Medical Image Annotation](https://synapcores.com/recipes/085_medical_image_annotation): Create and manage annotations, measurements, and findings on medical images
- [Patient Imaging History](https://synapcores.com/recipes/086_patient_imaging_history): Track and query patient imaging studies over time for longitudinal analysis
- [Radiology Report Integration](https://synapcores.com/recipes/087_radiology_report_integration): Link radiology reports with images and extract structured findings
- [Medical Image Analytics](https://synapcores.com/recipes/088_medical_image_analytics): Analyze medical imaging data for operational insights and quality metrics

### image-management
- [Multi-Format Image Library](https://synapcores.com/recipes/012_multi_format_image_library): Store images in multiple formats including JPEG, PNG, WEBP, and GIF with format-specific columns
- [Image with Metadata](https://synapcores.com/recipes/013_image_with_metadata): Store images with dimensions, file size, and EXIF data in JSON columns for rich metadata management
- [Photo Album Organization](https://synapcores.com/recipes/014_photo_album_organization): Create albums and photos tables with relationships for organizing images into collections
- [Image Tagging System](https://synapcores.com/recipes/015_image_tagging_system): Implement a flexible tagging system for images using a junction table for many-to-many relationships
- [Profile Picture Storage](https://synapcores.com/recipes/016_profile_picture_storage): Store user profile pictures with thumbnail variants for efficient display at different sizes
- [Image OCR Extraction](https://synapcores.com/recipes/017_image_ocr_extraction): Extract text from images using the EXTRACT_TEXT function for searchable image content
- [Image Dimension Query](https://synapcores.com/recipes/018_image_dimension_query): Query images by dimensions using the DIMENSIONS function for filtering by size and aspect ratio
- [Image Embedding Search](https://synapcores.com/recipes/019_image_embedding_search): Create embeddings for images to enable similarity search and visual content discovery
- [Screenshot Archive](https://synapcores.com/recipes/020_screenshot_archive): Store and organize application screenshots with metadata for bug tracking and documentation
- [Product Image Variants](https://synapcores.com/recipes/021_product_image_variants): Store multiple image sizes for e-commerce products including thumbnails, medium, and full-size variants
- [Scanned Document Images](https://synapcores.com/recipes/024_scanned_document_images): Store scanned documents as images with OCR text extraction for searchable document archives

### insurance
- [Adverse-Action Decision Records with Reason Codes, Tamper-Evident](https://synapcores.com/recipes/insurance/01_adverse_action_decision_record): Record every underwriting decision — score, model version, ECOA reason codes, human override — in a hash-chained immutable table, so the adverse-action notice and the examiner evidence come from the same verifiable source.
- [Disparate-Impact Testing — the Four-Fifths Rule in One Query](https://synapcores.com/recipes/insurance/02_disparate_impact_testing): Measure approval rates across protected groups and flag a disparate-impact problem with a single SQL query over the decisions you already made — continuous fairness testing for NAIC, Colorado and DOI review.
- [Find Similar Prior Claims by Narrative — Triage and Fraud Signal](https://synapcores.com/recipes/insurance/03_claim_narrative_similarity): Retrieve the most semantically similar past claims for a new one — surface likely-related or suspicious claims that share a story, not just a keyword, so adjusters triage consistently.

### legal
- [Flag Risky Clauses by Meaning Against Your Playbook](https://synapcores.com/recipes/legal/01_clause_risk_search): Score a contract's clauses against your risk playbook by semantic similarity to surface the risky, non-standard, or missing language — self-hosted, so privileged documents never leave.
- [Turn a Contract Into a Queryable Obligation Graph](https://synapcores.com/recipes/legal/02_obligation_graph): Model a contract's obligations — trigger, deadline, remedy — as a property graph, then answer 'what is due this quarter?' across the whole book in one query.
- [A Defensible, Tamper-Evident Record of Every AI-Assisted Review](https://synapcores.com/recipes/legal/03_immutable_review_audit): Log every AI-assisted contract or e-discovery review — document, model version, flags raised, reviewer decision — in a hash-chained immutable table, for a defensible and demonstrably consistent process.

### ml
- [Hello World: Linear Regression (y = 2x + 1)](https://synapcores.com/recipes/ml/001_hello_world_linear_regression): Train a simple regression model, deploy it, run predictions, then verify results match the known function.
- [Customer Churn Prediction](https://synapcores.com/recipes/ml/002_customer_churn_prediction): Build a model to predict which customers are likely to churn based on their account characteristics and usage patterns.
- [Drug Discovery Candidate Selection](https://synapcores.com/recipes/ml/005_drug_discovery_candidate_selection): Score compounds for likelihood of success. Prioritize top candidates for lab testing.
- [Real Estate AI: Property Matching & Market Analysis](https://synapcores.com/recipes/ml/006_real_estate_ai_matching): Build an intelligent real estate platform that matches buyers with properties, predicts market trends, analyzes property values, and generates insi...
- [AI-Driven Marketing & Customer Insights](https://synapcores.com/recipes/ml/007_ai_marketing_customer_insights): Create intelligent marketing campaigns by analyzing customer behavior, segmenting audiences, predicting churn, and generating personalized content...
- [Customer Segmentation with SQL and KMeans](https://synapcores.com/recipes/ml/010_customer_segmentation_kmeans): Group customers by behaviour with no labels and no Python — train KMeans directly on your SQL table, then inspect each segment's customers and feature values as rows.
- [Demand Forecasting in SQL: Predict Next Quarter's Sales](https://synapcores.com/recipes/ml/011_demand_forecasting_sql): Forecast six months of demand straight from a sales table — one SQL statement, validation that cannot see the future, and an honest comparison against the naive baseline it has to beat.
- [Churn Prediction in SQL: Rank Customers for Retention](https://synapcores.com/recipes/ml/012_churn_prediction_sql): Train a churn model on your customer table with one SQL statement, then score every account and hand sales a ranked list — no export, no notebook, no second system.
- [XGBoost-Style Regression in SQL: Estimate Property Prices](https://synapcores.com/recipes/ml/013_xgboost_regression_sql): Train gradient-boosted regression on a SQL table and score new rows with a SELECT — gradient boosting, random forests and linear baselines compared in one statement.
- [Density Clustering with DBSCAN: Find Groups and Call the Rest Noise](https://synapcores.com/recipes/ml/014_dbscan_density_clustering): KMeans forces every row into a cluster. DBSCAN finds the dense groups and labels the leftovers -1 — which is usually the answer you actually wanted.
- [Semantic Claim Verification with the Built-in NLI Model](https://synapcores.com/recipes/ml/020_semantic_claim_verification_nli): Check whether a statement is supported, contradicted, or unproven by evidence you supply — in-process ONNX inference, no external AI service
- [Support-Response Grounding with NLI](https://synapcores.com/recipes/ml/021_support_response_grounding): Verify every claim in a drafted support reply against the approved plan guide before it reaches the customer
- [Product-Claim Checks with NLI](https://synapcores.com/recipes/ml/022_product_claim_checks): Catch catalog and campaign copy that claims features the product specification does not support
- [Policy Wording Consistency with NLI](https://synapcores.com/recipes/ml/023_policy_wording_consistency): Flag help pages, agent notes, and explanations that misstate or reverse the governing policy text
- [Knowledgebase Conflict Review with NLI](https://synapcores.com/recipes/ml/024_knowledgebase_conflict_review): Detect when two knowledgebase articles disagree about the same fact, so an editor fixes the content instead of a customer finding it
- [Document-Revision Conflicts with NLI](https://synapcores.com/recipes/ml/025_document_revision_conflicts): Catch summaries and answers that quote an obsolete revision — and learn where typed comparisons must take over from language inference
- [Record and Supplier Reconciliation with NLI](https://synapcores.com/recipes/ml/026_record_and_supplier_reconciliation): Check downstream descriptions against approved specifications and extracted text against source records — after exact identity matching
- [Grounding Claims Against Supplied Facts](https://synapcores.com/recipes/ml/027_grounding_against_supplied_facts): Separate what your records establish from what an application merely asserts — the core pattern behind LLM-output guardrails
- [The NLI Pitfalls Test Battery](https://synapcores.com/recipes/ml/028_nli_pitfalls_test_battery): Nine probe requests that show where language inference breaks — run them before you automate anything on NLI verdicts
- [RAG Answer Gate: Retrieve, Verify, Then Ship](https://synapcores.com/recipes/ml/029_rag_answer_gate): Gate a RAG answer in one SQL statement — vector retrieval feeds NLI_ANCHORS feeds NLI_VERIFY — and learn why a similarity floor, not top-k, decides what the verifier sees
- [Self-Consistent Agent Memory: Refuse to Remember a Contradiction](https://synapcores.com/recipes/ml/030_self_consistent_agent_memory): Gate MEMORY_STORE with NLI_VERIFY over MEMORY_RECALL — an agent memory that rejects facts contradicting what it already knows, and logs the conflict instead of silently absorbing it
- [Closed-World Claims Audit: Marketing Copy vs. What You Can Substantiate](https://synapcores.com/recipes/ml/031_closed_world_claims_audit): Audit product and sustainability claims against your substantiation file with NLI_VERIFY in closed-world mode — a claim your documentation doesn't entail fails the gate, exactly the standard regulators apply
- [Citation Integrity: Does the Citing Sentence Say What the Paper Says?](https://synapcores.com/recipes/ml/032_citation_integrity): Screen citing sentences against the abstracts they cite with NLI_CLASSIFY — surface overclaims and polarity reversals for editorial review, a textual-entailment take on a real research-integrity problem
- [Clinical Note Consistency: Summary Claims vs. the Chart](https://synapcores.com/recipes/ml/033_clinical_note_consistency): Match each discharge-summary claim to the chart section it is about, verify it with NLI_VERIFY, and flag contradictions — a documentation-integrity assist that is explicit about where NLI must not be trusted alone
- [Contradiction Mining Across Statements — and Why It Needs a Similarity Gate](https://synapcores.com/recipes/ml/034_statement_contradiction_mining): Surface contradictions across a statement set with NLI_CLASSIFY — and learn the hard way that raw pairwise contradiction is noise until a vector-similarity pre-filter removes the unrelated pairs
- [Incident Report Integrity: Postmortem Claims vs. the Timeline](https://synapcores.com/recipes/ml/035_incident_report_integrity): Match each postmortem claim to the timeline entry it concerns and verify it with NLI_VERIFY — catch root causes the evidence doesn't establish and reassurances nobody recorded, before sign-off
- [AI Agent Memory: Catch Contradictions on Write, in One SQL Query](https://synapcores.com/recipes/ml/036_agent_memory_contradiction_check_on_write): Check every new agent memory against what the agent already believes before you store it: EMBED similarity pairs each new fact with related memories only, NLI_VERIFY flags contradictions, and one SQL query sorts the batch into conflict, already-known and new
- [How to Detect Hallucinations in RAG: a Per-Sentence NLI Gate in SQL](https://synapcores.com/recipes/ml/037_rag_hallucination_gate_per_sentence): Verify every sentence of a RAG answer against the retrieved passages at query time: EMBED pairs each sentence with its evidence, NLI_VERIFY returns supported / contradicted / unproven, and SQL turns that into a faithfulness score and a ship / review / block decision, with no LLM judge
- [RAG Citations: Does the Cited Source Actually Say That?](https://synapcores.com/recipes/ml/038_rag_citation_support_check): Check every (sentence, cited source) pair in a RAG answer with the built-in NLI model, and in the same SQL query use EMBED to find the source that should have been cited. Flags misattributed, contradicted and unsupported citations

### pharmacovigilance
- [Catch Duplicate & Related Adverse-Event Reports by Meaning](https://synapcores.com/recipes/pharmacovigilance/01_icsr_duplicate_detection): Find near-duplicate and related ICSRs before they inflate your case counts — semantic similarity over the report narrative, in one SQL query, no external vector store.
- [MedDRA Coding Assist — Verbatim Term to Preferred Term by Meaning](https://synapcores.com/recipes/pharmacovigilance/02_meddra_coding_assist): Suggest MedDRA Preferred Terms from a reporter's free-text verbatim by semantic similarity, ranked for a coder to confirm — consistent coding without a brittle keyword dictionary.
- [Source-Grounded ICSR Drafting — the Model Proposes, a Human Approves](https://synapcores.com/recipes/pharmacovigilance/03_source_grounded_icsr_draft): Draft an adverse-event narrative from structured case fields with SELECT GENERATE, then keep the human-in-the-loop review and the immutable audit that make it safe for a regulated safety workflow.
- [An Inspection-Ready, Tamper-Evident Audit Ledger for AI-Assisted PV](https://synapcores.com/recipes/pharmacovigilance/04_immutable_pv_audit_ledger): Record every AI-assisted drafting and coding decision — source, model version, reviewer, outcome — in a cryptographically hash-chained, encrypted-at-rest immutable table an FDA/EMA inspector can verify.

### security-surveillance
- [Security Camera Video Storage](https://synapcores.com/recipes/089_security_camera_video_storage): Store and manage surveillance video footage from security cameras
- [Access Control Logs with Images](https://synapcores.com/recipes/091_access_control_logs): Track access control events with badge photos and entry images
- [License Plate Recognition](https://synapcores.com/recipes/092_license_plate_recognition): Store and query license plate recognition data from parking and security systems
- [Facial Recognition Logging](https://synapcores.com/recipes/093_facial_recognition_logging): Log and query facial recognition events for access control and security
- [Incident Video Evidence](https://synapcores.com/recipes/094_incident_video_evidence): Manage video evidence collection and chain of custody for security incidents
- [Security Alert Management](https://synapcores.com/recipes/095_security_alert_management): Manage security alerts, notifications, and response workflows

### video-streaming
- [Basic Video Storage - MP4](https://synapcores.com/recipes/041_basic_video_storage_mp4): Store and retrieve MP4 video files with metadata for video content management
- [Multi-Format Video Library](https://synapcores.com/recipes/042_multi_format_video_library): Store videos in multiple formats (MP4, AVI, MKV, WEBM, MOV) for cross-platform compatibility
- [Video Duration Query](https://synapcores.com/recipes/043_video_duration_query): Query and analyze video content by duration for content planning and scheduling
- [Video Frame Extraction](https://synapcores.com/recipes/044_video_frame_extraction): Extract and store video frames for thumbnails, previews, and image analysis
- [Video Audio Track Extraction](https://synapcores.com/recipes/045_video_audio_track_extraction): Extract and manage audio tracks from video files for transcription and analysis
- [Video Transcription](https://synapcores.com/recipes/046_video_transcription): Transcribe video content to text for searchability, subtitles, and accessibility
- [Video Chapter Markers](https://synapcores.com/recipes/047_video_chapter_markers): Add chapter markers to videos for enhanced navigation and content organization
- [Video Thumbnail Storage](https://synapcores.com/recipes/048_video_thumbnail_storage): Store multiple thumbnail images for video content with selection preferences
- [Video Category Organization](https://synapcores.com/recipes/049_video_category_organization): Organize videos into hierarchical categories with tags for content discovery
- [Video Series Management](https://synapcores.com/recipes/050_video_series_management): Manage video series with episodes, seasons, and sequential ordering
- [User-Generated Video Platform](https://synapcores.com/recipes/051_user_generated_video_platform): Build a platform for user-uploaded videos with moderation and engagement features
- [Video View Tracking](https://synapcores.com/recipes/052_video_view_tracking): Track video views with detailed analytics including watch time and completion rates
- [Video Subtitle Storage](https://synapcores.com/recipes/053_video_subtitle_storage): Store multi-language subtitles and captions for video accessibility

### voice-agents
- [Build a Voice Support Agent (memory + routing + RAG)](https://synapcores.com/recipes/voice-agents/build-a-voice-support-agent): Build a complete voice support agent on one database — caller memory, real-time skill routing, RAG-grounded spoken answers, an urgency model, and a semantic cache, all in SQL. The brain for any STT/TTS stack (Vapi, LiveKit, Twilio, Pipecat).
- [Caller Memory & Personalization (returning callers)](https://synapcores.com/recipes/voice-agents/caller-memory-and-personalization): Make a voice AI agent recognize returning callers — recall their past calls and preferences by meaning and map their account relationships as a graph, so it greets them by context, not a cold script. For any STT/TTS stack (Vapi, LiveKit, Twilio, Pipecat).
- [Conversation Memory for a Voice Agent](https://synapcores.com/recipes/voice-agents/conversation-memory-for-a-voice-agent): Give a voice AI agent memory across a call and between calls — store transcribed turns, recall by meaning, and keep a rolling summary so it never loses the thread. Database-as-the-brain for any STT/TTS stack (Vapi, LiveKit, Twilio, Pipecat).
- [Low-Latency Semantic Answer Cache for Voice](https://synapcores.com/recipes/voice-agents/low-latency-semantic-answer-cache-for-voice): Cut voice AI latency and cost with a semantic answer cache — serve a stored answer when a caller's question means the same as one you've answered before, even if worded differently, with a vector lookup in SQL. For any STT/TTS stack (Vapi, LiveKit, Twilio, Pipecat).
- [Real-Time Intent → Skill Routing for a Voice Agent](https://synapcores.com/recipes/voice-agents/real-time-intent-to-skill-routing-for-a-voice-agent): Route a voice AI agent's caller to the right skill in real time — match the transcribed utterance to a skill by meaning with a vector lookup, fast enough for a live call. Database-as-the-brain for any STT/TTS stack (Vapi, LiveKit, Twilio, Pipecat).
- [Voice-Agent RAG Knowledge Base (ground spoken answers)](https://synapcores.com/recipes/voice-agents/voice-agent-rag-knowledge-base): Ground a voice AI agent's spoken answers in your own knowledge base — retrieve the right passage by meaning and generate a short, speakable, accurate reply, so the agent never makes things up on a call. For any STT/TTS stack (Vapi, LiveKit, Twilio, Pipecat).

## Optional
- [Company](https://synapcores.com/company)
- [Press](https://synapcores.com/press)
- [Builders Program](https://synapcores.com/builders)
- [Contact](https://synapcores.com/contact)