June 2025 Summaries

The text discusses the implementation of a multi-model Retrieval-Augmented Generation (RAG) system using SurrealDB and LangChain to create a GenAI chatbot, highlighting enhancements in AI capabilities for generating more accurate responses by retrieving relevant information from a knowledge base. This system leverages GraphRAG, which uses structured knowledge graphs to improve the contextual understanding of information. The process involves ingesting data about health symptoms and treatments, storing it in both a vector store and a graph store, and then querying these stores based on user input to provide tailored medical advice. LangChain components such as SurrealDBVectorStore and SurrealDBGraph are used to facilitate vector similarity searches and graph queries, while the ChatOllama model generates natural language responses. The system aims to improve the accuracy and coherence of AI-generated responses by reducing hallucinations and integrating various medical practices and treatments for symptoms like nasal congestion, dizziness, and sore throat.

At a pivotal moment in AI development, the emergence of Agentic AI marks a significant evolution from generative AI by enabling systems to act with autonomy and coordination, transforming extracted information into independent actions. This shift necessitates a rethinking of data architectures, as traditional models with their reliance on ETLs and pipelines are inadequate for the dynamic and context-aware demands of Agentic AI. The concept of "Agentic Memory" becomes crucial, requiring a real-time, interconnected knowledge base that allows agents to make informed, context-sensitive decisions. SurrealDB is presented as a solution designed to support this new paradigm with its multi-modal, real-time database capabilities, effectively addressing issues of data fragmentation and enabling seamless interaction among agents. The future success in this AI era will depend on the ability to develop composable decision systems driven by adaptive, goal-oriented execution, highlighting the need for databases to evolve into central components of agentic systems.

The blog post details a method for implementing semantic search using SurrealDB and OpenAI embeddings through the Rust SDK. The guide explains how to set up an embedded SurrealDB database and store OpenAI-generated embeddings as SurrealDB vectors, emphasizing the use of Rust’s async-openai crate for generating these embeddings. It includes step-by-step instructions on creating a vector table, defining fields and indexes, and storing the embeddings in the database. The post also covers performing semantic searches using different algorithms like cosine similarity and HNSW indexing, providing sample code for embedding generation and database querying. It culminates with examples of semantic search queries, demonstrating how the system identifies and returns documents closest in meaning to a given input.

Surreal Cloud Enterprise offers a cutting-edge database solution with its multi-model SurrealDB, designed to enhance the application-building experience for developers by eliminating infrastructure management challenges. It provides enterprise-grade features such as dedicated fault-tolerant clusters, horizontal scalability without sharding, and extensive compliance measures including ISO 27001 and SOC 2 certification. The platform supports mission-critical applications by delivering high performance, resilience, and security at scale, and it is equipped with advanced security functionalities such as the option to bring your own encryption keys and AWS PrivateLink support. Upcoming features include enterprise SSO integration, comprehensive audit and access logs, and customizable log retention policies, which aim to meet strict governance standards. Surreal Cloud Enterprise promises full-stack flexibility for building and scaling next-generation applications and AI systems, and interested users are invited to join the Early Access Programme for further exploration.

SurrealDB has introduced new network capabilities in Surreal Cloud, enhancing security by allowing users to define granular network access rules for their databases. These capabilities enable administrators to specify allowlists and denylists for outbound connections, ensuring that databases only communicate with approved network targets, which is crucial for preventing unauthorized access and data breaches. By employing pattern matching, users can control access to specific domains, IP ranges, or wildcards, providing flexibility in crafting security policies tailored to their infrastructure. The system's design allows for seamless integration with existing Surreal Cloud tools, offering a user-friendly interface to manage these settings and reducing the risk of misconfiguration. This feature is especially beneficial in environments where databases must interact with external services while maintaining strict security protocols, and its implementation aligns with real-world security needs, helping organizations maintain compliance and minimize potential vulnerabilities.

SurrealDB offers a revolutionary approach to building real-time AI pipelines by integrating diverse data models and AI functionalities directly within the database, eliminating the need for traditional Extract, Transform, Load (ETL) processes. As a multi-model database, SurrealDB is adept at handling various data structures, such as relational, document, graph, and geospatial data, making it particularly suited for developing sophisticated AI applications. It supports the creation of knowledge graphs and graph-based recommendations, which enhance the querying and inference capabilities within data sets. The database facilitates real-time AI features like embedding creation, vector and semantic search, full-text search, and hybrid search. It integrates with Large Language Models (LLMs) through structured prompt templates, enabling context-aware and responsive AI solutions. By leveraging SurrealDB's capabilities, developers can build intelligent applications that adapt to real-time data changes, fostering a more dynamic and context-rich environment.

SurrealDB has announced official support for Airbyte, an open-source data integration platform, allowing developers to create customizable, self-hosted data pipelines that can transfer data from various sources directly into SurrealDB. This integration is aimed at engineering teams that prioritize flexibility and transparency, as it enables the inspection, versioning, and extension of connectors, which can be operated locally or in the cloud. By leveraging SurrealDB’s native support for dynamic schemas and real-time queries, the ingested data is immediately ready for transactional use, analytics, or AI inference. Together, SurrealDB and Airbyte provide a fully programmable data stack that is open, composable, and designed for modern applications, with the SurrealDB connector now available on the Airbyte platform, supported by an integration guide for configuration.

Surrealist 3.4 introduces several highly anticipated features, including multi-window support for desktop users, enhanced computed table views, and a redesigned Cloud deployment workflow, which collectively aim to improve user experience and workflow efficiency. Users can now open multiple independent windows to connect to different SurrealDB instances, enhancing the ability to work with multiple databases simultaneously. The support for computed table views has been significantly improved, allowing for better distinction and management in both Explorer and Designer views, while manual record creation in these views is disabled to maintain data integrity. Additionally, the Cloud deployment process is now more intuitive, featuring automatic instance name generation, options for initializing with sample datasets, and enhanced deployment within preferred organizations. The platform has also relocated to a new web address, app.surrealdb.com, and incorporates a modernized appearance, including improved visual consistency, updated icons, and the addition of breadcrumbs for easier navigation.

Knowledge graphs have evolved from an academic concept into a critical component for modern AI systems, providing structured memory that enhances search, recommendation, and enterprise AI capabilities by grounding large-language models (LLMs) in reality. These graphs, which represent knowledge as networks of nodes connected by relationships, have become a priority for decision-makers seeking competitive advantages through improved data lineage, semantic search, and reasoning capabilities. The rise of LLMs, vector search, and regulatory demands for transparency have fueled the adoption of knowledge graphs as a context reservoir for AI agents. SurrealDB offers a unified, ACID-compliant approach to storing knowledge graphs alongside other data models like documents and vectors, allowing for efficient queries and real-time updates without the complexity of managing multiple specialized databases. By seamlessly integrating graph capabilities with other data types, SurrealDB provides a scalable and operationally efficient solution for organizations aiming to leverage knowledge graphs for enhanced AI performance and strategic insights.

The focus of AI development has shifted from building bigger models to enhancing database efficiency, as highlighted by the increasing number of database acquisitions in the AI space and insights from Kolawole Samuel Adebayo’s Forbes article. The current bottleneck for AI is not computational power but rather the management of data, storage, and memory, which are essential for real-time access and reasoning. Many AI initiatives are failing due to data unavailability when needed, caused by fragmented sources, slow pipelines, and lack of a memory layer. Traditional databases are insufficient for the real-time, consistent, and recall-based demands of AI agents, which require a memory architecture that is semantically searchable, relationally traversable, and transactionally safe. The industry is gradually recognizing that databases are not merely infrastructure but fundamental to scalable cognition, prompting architectural changes at major tech companies. SurrealDB advocates for a storage and memory engine tailored for AI agents, integrating unstructured and structured data storage with advanced search and reasoning capabilities to provide a solid foundation beyond simple caching.

This blog post provides a concise overview of various techniques for optimizing database schemas using SurrealDB. It highlights ten tips that enhance schema design, such as utilizing sets to prevent duplicate entries, defining arrays and nested indexes with specific types and constraints, and using the FLEXIBLE keyword in SCHEMAFULL tables to accommodate free-form objects. The post also covers advanced features like customizing error messages with ASSERT clauses, employing time formatters for datetime precision, and leveraging the double exclamation operator (!!) to validate non-empty values. Additional strategies include defining reusable parameters for cleaner code, using literal types for detailed error handling, and integrating graph queries within the schema to manage complex relationships like family trees. These insights aim to empower developers to create more efficient and flexible database schemas with SurrealDB.

Cyril Scetbon explores the integration of SurrealDB and Rig.rs to create a knowledge agent capable of embedding and querying word definitions efficiently. SurrealDB, a Rust-based database combining document, graph, relational, and vector models, complements Rig.rs, a toolkit for building language model-native agents. The process involves setting up a project, ingesting and embedding documents in SurrealDB, and using Rig to create an agent that applies OpenAI's models for natural language processing. This setup allows the agent to respond to queries based on stored embeddings while avoiding fabricating answers. The article emphasizes the advantages of using Rust-native tools like SurrealDB for flexible and efficient data handling in language model applications, encouraging developers to streamline their workflows by leveraging the comprehensive capabilities of Rig and SurrealDB.

The integration of SurrealDB and Amazon Bedrock offers a streamlined approach to building Retrieval-Augmented Generation (RAG) systems, which are essential for modern AI applications that require concise and reliable information retrieval. SurrealDB serves as a unified, ACID-compliant engine capable of managing JSON documents, graph edges, and vectors in a single database, while Amazon Bedrock provides managed foundation models for generating document embeddings and summaries. This combination simplifies RAG architectures by eliminating the need for multiple specialized services, thereby reducing complexity and cost. A practical application of this system is a Contract Copilot for legal teams, which indexes contract clauses as vectors and uses Bedrock to generate summaries and insights, drastically cutting down the time lawyers spend on document searches. This unified data and processing model accelerates system delivery, scales efficiently, and allows for the creation of robust, production-ready RAG systems with reduced overhead.

Financial fraud is a rapidly increasing global issue, significantly impacting both consumer and business sectors, with recent studies indicating that fraud now accounts for 6.5% of annual revenue losses for companies, totaling $359 billion in losses for a sample set, and $485 billion globally in 2023. Fraud is often a graph problem, as it involves complex relationships between entities such as shared emails and reused devices, which can be better understood through graph databases. SurrealDB offers a comprehensive platform to tackle fraud detection by integrating graph, document, and vector data, facilitating real-time alerts, event-based actions, and inline machine learning scoring, all within a single system. This unified approach simplifies the architecture by eliminating the need for multiple external services, thereby reducing maintenance complexity, latency, and costs. SurrealQL, the query language of SurrealDB, is designed to model frequent fraud patterns efficiently, allowing for rapid adoption by engineers familiar with SQL. The platform's real-time capabilities are enhanced by SurrealML, which enables the execution of machine learning models directly within the database, providing immediate fraud predictions and enabling a live feedback loop for continuous model improvement.

SurrealDB has announced its official support for Fivetran, allowing businesses to integrate and sync data from a wide range of sources directly into SurrealDB without requiring any manual coding. This new Fivetran connector automates essential processes such as schema evolution, data normalization, and incremental updates, reducing the need for maintaining ETL scripts and enabling teams to focus on utilizing live, trusted data for real-time decision-making and AI-powered systems. The integration aims to enhance the reliability, scalability, and user-friendliness of SurrealDB, making it a robust platform for application backends and other demanding real-time applications. Currently available in private preview, SurrealDB invites feedback on the integration experience to refine its offering further.

Aspire Comps, a UK-based retail prize competition company, transitioned from Firebase to SurrealDB to address performance issues and streamline operations, resulting in a scalable and efficient infrastructure supporting over 700,000 users. The migration consolidated five backend services into one, leveraging SurrealDB's capabilities to unify application logic and data with enhanced access control and support for serverless functions. This change reduced system complexity and improved performance by handling high concurrency without additional optimizations, allowing for flexible modeling and native delivery of search, access control, and event handling. SurrealDB's multi-model engine supports rapid iteration for new competition formats and customer flows, preparing Aspire Comps for future machine-learning workloads and advanced search capabilities.

The blog post explores the innovative use of Surrealist's graph view, a feature of SurrealDB, to visualize musical concepts such as the circle of fifths, alongside traditional data like train stations and international agreements. It delves into constructing a circle of notes and the circle of fifths using SurrealDB and LangChain, illustrating how these visualizations can aid musicians in identifying harmonious chord progressions and constructing major chords. The circle of fifths, a powerful musical tool, helps musicians transition seamlessly between chords by showing which chords naturally align sonically. The post further demonstrates how to create a database-wide parameter for musical notes and a function to identify perfect fifths, offering a practical application for both musicians and developers interested in data visualization and music theory. Additionally, it encourages readers to expand on the concept by incorporating minor scales to create a more comprehensive circle of fifths, enhancing the ability to transition between major and minor chords.

SurrealDB offers a versatile approach to data relationship modeling, featuring three primary methods: Record Links, Record References, and Graph Edges, each catering to different needs. Record Links provide a straightforward, efficient way to connect records directly without using JOINs or intermediate tables, though they are static and don't automatically update if linked records change. Record References, an experimental feature, allow for bidirectional relationships, enabling reverse visibility without manual joins, though they require enabling and carry certain limitations. Graph Edges, meanwhile, are robust records that store metadata and support complex relationships, offering forward, backward, and recursive traversal, ideal for scenarios requiring detailed metadata and visual schema representation. These tools collectively empower users to model data directly, flexibly, and with comprehensive graph awareness, providing options for simple to complex relational needs within SurrealDB.