Updating according to the feedback

Author: ViktarStarastsenka

content/develop/redis-versions/6-2

@@ -23,10 +23,10 @@ Redis now supports indexing and querying multi-value attributes across all field
 The Redis Query Engine now enables suffix and infix wildcard searches for `TEXT` and `TAG` fields. This enhancement provides greater flexibility in data retrieval and filtering.
 
 ### t-digest: a new probabilistic data structure for quantile estimation
-Redis introduces t-digest, an advanced probabilistic data structure that efficiently estimates quantiles in large datasets or continuous data streams. This is particularly beneficial for analytics and monitoring applications where precise quantile calculations are required.
+Redis introduces t-digest, an advanced probabilistic data structure that efficiently estimates quantiles in large datasets or continuous data streams. This is particularly beneficial for analytics and monitoring applications where quantile calculations are required.
 
 ### Retrieve aggregation results for ongoing time series buckets
-A new feature allows users to retrieve the latest, still-open time series buckets during compaction. This is particularly useful when dealing with retention policies that limit access to raw historical data, enabling real-time insights into event counts over extended periods.
+A new feature allows users to retrieve the latest, still-open time series buckets during compaction.
 
 ### Time-weighted average aggregator for time series
 Redis now includes a time-weighted average aggregator, improving accuracy in average-over-time calculations. This feature is especially valuable for time series data with irregular sampling intervals.

content/develop/redis-versions/7-2

@@ -11,23 +11,21 @@ linkTitle: Redis 7.2
 weight: 9
 ---
 
-Redis version 7.2 introduces new capabilities, including improved geospatial queries, streamlined JSON data manipulation, and the public preview of Triggers and Functions for event-driven processing. Performance optimizations, client-side enhancements, and behavioral refinements further improve Redis' efficiency, security, and usability.
+Redis version 7.2 introduces new capabilities, including improved geospatial queries, streamlined JSON data manipulation, and the public preview of Triggers and Functions for event-driven processing (not included in 7.4 and later). Performance optimizations, client-side enhancements, and behavioral refinements further improve Redis' efficiency, security, and usability.
 Below is a detailed breakdown of these updates.
 
 ## New features
 
 ### Geospatial queries with polygon search
 Redis Query Engine now supports querying geospatial data using polygon search, enabling developers to efficiently filter and retrieve data within complex geographic boundaries.
 
-For example, in a weather monitoring system, Redis can store geospatial data for extreme weather events such as hurricanes, wildfires, or thunderstorms. When a user defines a geographic area of interest, Redis Query Engine quickly retrieves relevant weather alerts within that region, enabling real-time notifications and faster emergency response.
-
 ### Streamlined data manipulation in JSON
 JSON now includes two new commands for improved data handling:
 
 - `JSON.MERGE`: Merges a given JSON value into matching paths, allowing more flexible updates.
 - `JSON.MSET`: Sets or updates multiple JSON values simultaneously based on specified key-path-value triplets, improving efficiency when handling structured data.
 
-### Public preview: Triggers and Functions
+### Public preview: Triggers and Functions (not included in 7.4 and later)
 Redis introduces the public preview of Triggers and Functions, an event-driven engine for executing TypeScript/JavaScript code within Redis. This feature enables efficient in-database data processing, reducing reliance on external computation. It also supports cross-shard read operations, which were previously unavailable with Lua scripts.
 
 ## Improvements
@@ -48,7 +46,7 @@ TLS-based replication now supports Server Name Indication (SNI) to improve compa
 
 ## Changes
 
-### Backward-incompatible changes
+### Breaking changes
 Redis 7.2 introduces several backward-incompatible changes. Lua scripts no longer support the print() function, blocking of `PFCOUNT` and `PUBLISH` in read-only scripts, and time sampling freezing during command execution. Error handling updates include case changes in error responses, new behavior for `ZPOPMIN/ZPOPMAX` with count 0, and adjustments to `XCLAIM/XAUTOCLAIM`. ACL changes affect command categorization and key access permissions, while command introspection now includes per-subcommand stats. Redis now allows certain CONFIG commands during loading and tracks stats only when commands execute.
 
 For more details, see [Redis 7.2 Breaking Changes](https://redis.io/docs/latest/embeds/r7.2-breaking-changes/).

content/develop/redis-versions/7-4

@@ -20,7 +20,7 @@ Redis 7.4 introduces several new features and improvements aimed at enhancing me
 Redis 7.4 adds the ability to set expiration times for individual hash fields or adjust their remaining TTL. This feature, long-requested by users, improves memory efficiency and performance, especially in caching and session storage scenarios.
 
 ### New memory-efficient data types for AI workloads
-With the growing demand for AI applications, Redis 7.4 introduces `bfloat16` and `float16` data types. These new types reduce memory usage by up to 47% and lower latency by as much as 59% under load, making them ideal for vector databases and Retrieval Augmented Generation (RAG) applications.
+With the growing demand for AI applications, Redis 7.4 introduces `bfloat16` and `float16` data types. These new types reduce memory usage by up to 47% and lower latency by as much as 59% under load, making them ideal for storing and processing vector embeddings in AI-powered applications, including vector databases and Retrieval Augmented Generation (RAG) systems.
 
 ### Time series optimization with insertion filters
 Redis 7.4 introduces insertion filters for time series data, allowing sensors to ignore new measurements when the differences in time or value are minimal. This feature helps reduce the size of time series data and boosts efficiency.