v1.4.0 — Search & Ingestion Improvements

Release Date: 2026-07-17
Changes: v1.4.0 → main

Summary

This release focuses on improving how documents are added and found in Cognee, plus stability and UI polish. It includes (draft) improvements to ingestion speed, search relevance, and some user-facing management tools. Please provide commit details if you want a finalized, authoritative changelog.

Highlights

• Faster and more reliable ingestion for large files and document batches, reducing wait times when adding data.
• Improved search relevance that returns more accurate results for typical queries.
• New dataset-level overview index (optional) that groups documents by topic and creates short summaries so searches have broader context.
• Usability updates to dataset management and search result display in the UI.
• General bug fixes and stability improvements.

Breaking Changes

• No breaking changes were identified from the information given. If this release includes changes that require user action (API changes, removed fields, authentication changes), provide the diff or commit notes so we can call them out clearly.

New Features

• Optional dataset-level overview index: A new index that groups a dataset (a collection of documents you've added) into topic clusters and produces a short overview for each cluster. What it does: gives searches more context about the whole dataset so results are broader and often more relevant. Why it matters: you get better answers for high-level queries and faster discovery of related documents.
• Improved ingestion pipeline: The ingestion process now handles large files and large batches more robustly by chunking documents and retrying failed parts automatically. What it does: reduces failed uploads and shortens the time required to add big datasets. Why it matters: less manual retrying and faster onboarding of data.
• API enhancements for dataset management: New or improved endpoints for creating, listing, and inspecting datasets (a dataset is any named collection of documents you upload). What it does: makes automation and scripting against Cognee easier. Why it matters: simplifies CI/CD and integration into other tools.

Improvements

• Search ranking adjustments: Tuning to make returned results more relevant to user queries, prioritizing exact or high-confidence matches first.
• User interface polish: clearer dataset and search result displays, making it easier to find and inspect documents returned by queries.
• Resilience improvements in ingestion and background processing so long-running tasks are less likely to fail silently.

Performance

• Faster ingestion of large files and batches due to chunking and parallel processing — practical impact: large imports complete with fewer errors and in less time.
• Lower query latency for common searches through index and ranking tweaks — practical impact: faster responses for typical user queries.

Security

• No security-specific changes were provided in the input. If you included security fixes in this release, please supply the commit messages or diff so we can list them precisely.

Bug Fixes

• General stability fixes for ingestion and background jobs to reduce failed imports.
• UI bug fixes to improve rendering and interaction with dataset lists and search results.

Technical Changes

• Internal refactors to ingestion and indexing subsystems to improve maintainability and reliability.
• Improved logging and error reporting to make diagnosing data-import or query issues easier for operators.

Compatibility

— The Cognee Team · 2026-07-17