April 2026 Summaries

DeadReckoning, developed during the LangChain × SurrealDB London Hackathon, is a knowledge graph agent designed to enhance codebase navigation by enabling queries in plain English. It addresses the limitations of Retrieval-Augmented Generation (RAG) systems, which often return semantically similar but incorrect results, by employing a hybrid search strategy that combines vector and keyword searches using Reciprocal Rank Fusion (RRF). This approach is implemented within SurrealDB, allowing for efficient, database-native operations without the need for middleware or score normalization. The system further enriches search results by providing context through graph traversal, detailing the class, sibling functions, and call relationships, thereby transforming accurate answers into useful insights. DeadReckoning operates on small local models, making it suitable for environments with limited resources or stringent data privacy requirements, and highlights the importance of context engineering over model complexity.

The integration of Kreuzberg and SurrealDB offers a streamlined solution for building document search and retrieval systems by combining the document intelligence framework of Kreuzberg with SurrealDB's multi-model database capabilities. This partnership facilitates the extraction, chunking, and embedding of over 88 document formats, while SurrealDB manages the storage, indexing, and search functionalities, including both keyword and semantic search options. The kreuzberg-surrealdb connector simplifies the ingestion workflow with automatic schema setup and content deduplication, enabling immediate readiness for document search. By integrating these processes into a single system, it eliminates the need for separate tools and supports both BM25 keyword search and advanced hybrid search methods using HNSW vector indexes and Reciprocal Rank Fusion.

Synapse is a memory-first reflection agent designed to transform journal entries into a persistent knowledge graph structured around therapeutic frameworks such as CBT, DBT, IFS, and Schema Therapy. Developed for the London LangChain x SurrealDB Hackathon, it helps users track emotional and behavioral patterns over time by linking new reflections to existing data and extracting insights. The system uses a LangGraph pipeline to enhance memory compounding, allowing the chat agent to provide informed answers by leveraging a structured graph of patterns, emotions, and relationships. To ensure effectiveness, Synapse undergoes rigorous evaluation for extraction quality, graph integrity, chat grounding, and pipeline performance. Key features include crisis detection, non-diagnostic language, and voice input via Telegram integration. Despite challenges like latency, improvements were made through parallel processing, batching, and SSE streaming, with the Anthropic claude-sonnet-4-6 model chosen for its deep extraction capabilities. The development process emphasized orchestration and prompt design, facilitated by tools such as LangChain, FastAPI, and SurrealDB.

Surrealist 3.8 introduces several enhancements, including a redesigned navigation experience and improved query formatting, aimed at providing a more intuitive user interface. The update consolidates the organisations page into the overview page, enhancing accessibility and information visibility with a new sidebar navigation. Quality of life improvements include advanced query formatting options, a request timer, and streamlined notifications management, alongside the ability to export query results in JSON or CSV formats. The release also addresses numerous bug fixes, enhancing stability and usability across macOS and other platforms. Users are encouraged to provide feedback and engage with the community to further refine future updates.

The blog post explores the implementation of a hybrid fusion search model for SurrealDB documentation, combining full-text and vector search techniques to enhance the relevance of search results. Full-text search involves splitting and modifying text to match query terms, while vector search utilizes OpenAI's embeddings to capture semantic meanings. The search functionality integrates both methods using Reciprocal Rank Fusion (RRF) to provide a comprehensive search experience. The implementation is detailed in the SurrealDB documentation repository, and it allows users to locally deploy and test the search feature. Additionally, the post provides a simplified example to demonstrate the hybrid search logic, encouraging readers to explore the approach for their own applications.