The record of firsts and onlys in entity resolution.
Firsts & Onlys
A sixth-generation engine that covers every need: from 1M-record workloads to a hundred billion records, across real-time transactional, dynamic (just-in-time), and Spark batch, for both agentic and traditional workflows.
Only Senzing® delivers…
Senzing was the first to deliver…
The Through-Line
Every release advances the same story: an engine that scales in both directions, reaches across the world’s data, and moves natively into agentic AI.
Arc 01 · Form factor
Sub-second cold start, a ~560 MB footprint, pure in-memory mode, and a composable SDK. Ideal for agentic workloads: scale up to billions of records on compute clusters, and scale down to laptops for very small data sets.
Arc 02 · Across languages
Native cross-script matching between CJK (Chinese, Japanese, Korean) and English, with no transliteration. Many more languages and scripts are supported through transliteration, alongside semantic search and regionally-trained address parsing.
Arc 03 · The agentic spine
Senzing was first to bring entity resolution to a large language model with Conversational Entity Resolution™. It followed with the first standalone MCP server for entity resolution, so agents can stand up resolution pipelines directly.
Always-On Foundations
These properties are built into the engine itself and hold across every advance on the timeline. Explore each in depth.
An embeddable library that runs inside your own stack, so no data ever flows to Senzing.
Resolves from built-in principles, with no per-dataset training or tuning.
Organizes data around entities and self-corrects as new records arrive.
Resolves in real time, one record at a time, as data streams in.
Surfaces the non-obvious connections between resolved entities.
Every match and non-match is transparent: the why, why-not, and how.
The same result no matter what order records are loaded.
Any single record can be removed and its effect cleanly, fully reversed.
Every resolved value traces back to the exact source record it came from.
Marquee moments from a fast-moving release cadence. For every point release in between, see the full release notes.
Fully serverless entity resolution with incremental adds and no full reloads. Later benchmarked at about 10 million records in about 3 hours for under $100.
Entity resolution brought to a large language model via plug-in, roughly six months after ChatGPT launched. The origin point of the agentic arc.
Automatic deterministic and non-deterministic encryption of sensitive columns, on the fly. Unique in the industry.
Run resolution and search against read replicas and standbys, spreading load and hardening availability.
Entity resolution on ARM servers such as AWS Graviton and on edge hardware, widening where it runs and lowering compute cost.
Free, Senzing-ready JSON snapshots (sanctions lists, corporate registries, offshore leaks, watchlists), so teams can prove value and explore third-party data in minutes, with no mapping. Explore CORDs →
Native CJK↔English matching for names and addresses with no transliteration, alongside Why Search, Record Preview, Match Key Details, Feature Traceability, and embedding-based semantic search.
A first commercial customer surpassed 10 billion input records; a public-sector deployment reached roughly 10× larger. Real-time resolution at that scale, from a single composable engine.
The free universal place identifier brought in-house to strengthen location matching, and kept open for the community. Learn about Placekey →
.NET joins Python and Java, widening the languages that can embed the engine directly.
Cold start cut to about half a second, down from several seconds, and the default for everyone. It makes the engine viable inside agentic workflows and serverless functions.
AWS Aurora PostgreSQL IAM roles and Azure SQL Entra ID auth, using short-lived identity tokens instead of stored credentials.
A standalone Model Context Protocol server: natural-language data mapping, SDK scaffold code, and about 30 example repositories. It never sees your data, and it cuts new-source onboarding from about 300 hours to under three.
End-to-end agentic pipelines across Spark batch, transactional SQL, and hybrid modes.
One-click agentic entity resolution in the Kiro IDE, the first entity-resolution power for Kiro.
86 million organizations, 101 million contacts, and 142 million locations in Senzing-ready JSON, with Placekey among 162 reference identifiers. Explore the dataset →
Senzing’s own regionally-trained address parsing, for unprecedented accuracy in address matching.
Dynamic, just-in-time entity resolution over smaller record sets, performed entirely in memory. No database.
The engine reduced to about 560 MB, roughly 7× smaller than earlier versions, making OEM embedding, serverless, and edge usage even more frictionless.
New capabilities land on a steady cadence. See every release in detail, or put the engine to work on your own data.
Stay up to date with our newsletter and email updates.
Please add your email address to opt-in to be subscribed to our email marketing list. You can unsubscribe at any time. For further information, please view our full Privacy Notice.
SENZING®, SMARTER ENTITY RESOLUTION®, SZ®, and ENTITY CENTRIC LEARNING® are registered trademarks of Senzing, Inc. and may not be used without prior written permission. POWERED BY SENZING™, SENZING INSIDE™, and CONNECT DATA POWER INTELLIGENCE™ are trademarks of Senzing, Inc. and may also not be used without prior written permission.
