San Jose, CA May 02, 2026 --(PR.com)-- Datavor.ai today announced the general availability of Datavor version 2.5, a significant leap forward for its AI-native database pipeline tool built for Anthropic’s Claude Desktop. Running natively on Mac — including Mac mini, MacBook, iMac, and Mac Pro — Datavor enables developers and data professionals to sync, transform, schedule, and monitor database pipelines entirely through natural language conversation. No SQL expertise required. No complex graphical interfaces. No subscription fees.
Version 2.5 is the most substantial release in Datavor’s history, introducing five new capability areas that collectively bring enterprise-grade data pipeline features to Mac users for the first time through conversational AI.
Datavor v2.5 introduces Change Data Capture (CDC) for both PostgreSQL and MySQL, enabling real-time replication of every database change — inserts, updates, and deletions — as they happen. PostgreSQL CDC uses logical WAL (Write-Ahead Log) replication; MySQL CDC uses binary log streaming in ROW format. CDC checkpoints are automatically persisted in Datavor’s Context Engine, ensuring streams resume cleanly after any restart with no missed events.
For Mac users running production-adjacent workflows, this means data can flow continuously between source and destination databases without polling delays or manual sync triggers — activated with a single natural-language instruction to Claude.
Dependency-Aware DAG Scheduling
Datavor’s scheduler now supports full dependency-aware pipeline orchestration via a Directed Acyclic Graph (DAG) model. Users can express complex pipeline relationships in plain English — “make the analytics rollup job wait for both the orders sync and the customers sync” — and Datavor enforces the dependency chain automatically.
Cycles are rejected at definition time. Failed parent jobs block their downstream children. All jobs include exponential backoff retry logic. A visual graph command renders the full dependency topology on demand. This capability, previously available only in dedicated pipeline orchestration platforms, now runs entirely on a Mac through Claude Desktop.
Reusable Transform Recipes
v2.5 introduces a Recipe Manager that allows any transform configuration — field renames, type casts, row filters, value mappings, computed columns — to be saved as a named, reusable recipe and applied to future syncs with a single command. Frequently-used transform patterns are also auto-captured from sync history, meaning Datavor learns common workflows without requiring any manual setup.
Proactive AI Suggestions from the Context Engine
Building on the persistent knowledge store introduced in v2.0, Datavor’s Context Engine now surfaces proactive, actionable suggestions — surfaced automatically and reviewed on demand. The engine detects three categories of insight:
- Schema changes — when a new column appears on a table that is actively syncing, Datavor suggests adding it to the destination (and never applies DDL changes automatically)
- Error patterns — recurring failures across sync runs are detected, counted, and surfaced with suggested resolutions; once a resolution is recorded, it is applied to future occurrences
- Optimisation opportunities — full-replace syncs that consistently take a long time are flagged with a recommendation to switch to incremental sync or CDC
Each suggestion carries an accept or dismiss lifecycle. Dismissed suggestions are never re-raised.
Per-Record Fault Tolerance
In all previous versions of Datavor, a single malformed row could abort an entire sync batch. Version 2.5 introduces per-record fault tolerance: each batch attempts a bulk write first, then falls back to row-by-row processing on failure, isolating any problematic records while committing the remainder of the batch. Every sync response now reports a ‘Rows synced / Rows failed’ breakdown, with full failure details available through the dashboard.
By the Numbers: Datavor v2.5
- 45 MCP tools available in Claude Desktop (up from 34 in v2.0)
- 5 database connectors: MySQL, PostgreSQL, SQL Server, SQLite, and Snowflake
- 642 automated tests passing across all connectors and pipeline modes
- 11 new tools in v2.5 alone: CDC streaming, recipe management, suggestion lifecycle, and DAG scheduling
- Zero cost: free for personal and commercial use
“Datavor started from a simple belief: that managing data pipelines should not require a team of data engineers or a five-figure annual software budget. With v2.5, Mac users can now stream real-time database changes, build sophisticated dependency chains, and let the tool proactively surface what needs attention — all through natural conversation with Claude. This is what AI-native tooling should feel like.” — Founder, Datavor
Availability
Datavor v2.5 is available now as a free download via npm and at datavorai.com. Installation on any Mac running Claude Desktop takes under two minutes.
Website: datavorai.com
Install: npm install -g datavor
License: Free for personal and commercial use
Platform: macOS (Mac mini, MacBook, iMac, Mac Pro); also Linux and Windows
Requires: Node.js 18+ and Claude Desktop (free)
Database support: MySQL, PostgreSQL, SQL Server, SQLite, Snowflake
Contact Information:
Datavor
June Rose
+44 07521212228
Contact via Email
www.datavorai.com
Read the full story here: https://www.pr.com/press-release/967418
Press Release Distributed by PR.com
