Entify

Entify is a record-linkage workspace that turns a matching process into a guided review rather than a specialist-only exercise. It is a good fit for data stewards working with modest datasets who want to inspect and trace proposed merges. Its strongest case is explainable local matching; its tight resource limits and localhost-only security assumption rule it out for larger or exposed deployments.

Overview

Built around Splink 4 and DuckDB, Entify helps users identify records that may refer to the same entity, review the evidence, and produce a deduplicated export. The MIT-licensed project can run locally with Docker or with Python 3.11+ and Node 20+. The Docker quickstart requires no accounts, API keys, or external services, and uploaded data stays on the machine—useful for teams that prefer a local workflow over shared hosted storage.

Key features

Entify profiles columns with row counts, distinct values, and completeness, then suggests likely column types and blocking rules. It explains those choices and lets users override them, which gives newcomers a workable starting point without locking experienced users into automatic decisions. Matching offers exact, Jaro-Winkler, Jaccard, and Levenshtein comparisons.

Reviewers can examine match evidence, scores, and threshold analyses before accepting a cluster. Merge exports keep one surviving record per entity while retaining source IDs, supporting traceability and undo; PDF audit reports provide a record of the review. CSV, TSV, Parquet, JSON, and Excel inputs cover a broad set of common data formats.

Semantic blocking is optional, but installing it downloads roughly 600 MB of Torch dependencies. That is a sizable addition for teams that need it; users relying on the standard comparisons can avoid it.

Pricing

Entify’s Open-source plan costs nothing. It includes the open-source tool, with uploads capped at 100 MB by default and only one resolution result held in memory at a time. There is no paid tier or seat allowance described. The project is free under the MIT license, but the caps make it a better fit for small jobs than recurring large-scale work.

Platforms

Entify supports API, self-hosted, and web use. Local deployment is possible through Docker or a Python 3.11+ and Node 20+ setup. The API has no authentication and assumes it is bound to localhost; do not expose it to untrusted callers as-is. Only one resolution result stays in memory, so a second concurrent run replaces the first. Uploads are read fully into memory, and workloads beyond roughly 100,000 rows call for a DuckDB file backend and job queue.

Who it's for

Choose Entify if you steward data or work in a team that needs record linkage without already knowing how to design blocking rules and match weights. Its guided setup, inspectable evidence, and traceable exports suit modest deduplication projects. Look elsewhere if your workflow needs larger uploads, concurrent runs, or an API exposed beyond localhost without additional security work.

Pros and cons

• Pros: Automatic setup explains suggested column types and blocking rules, helping non-specialists begin while retaining the ability to override decisions.
• Pros: Evidence review, threshold analysis, source-ID retention, and PDF audits make proposed and completed merges easier to examine and trace.
• Pros: The local Docker route needs no accounts, keys, or external services, and keeps uploaded data on the machine.
• Cons: The 100 MB default upload cap and memory-bound operation beyond roughly 100,000 rows limit its fit for larger datasets.
• Cons: A second concurrent run replaces the first result, restricting parallel work.
• Cons: The unauthenticated API is intended for localhost binding, not direct exposure to untrusted callers.
• Cons: Optional semantic blocking adds roughly 600 MB of Torch dependencies.

Alternatives

Tilores is worth considering if you want local entity resolution with a stated allowance of up to 100,000 records per deployment and no account or licence key. DupeCatcher is a free option across Android, iOS, and web, with all features included. AWS Entity Resolution may suit provider matching billed per 1,000 records processed, though provider licensing or subscriptions cost extra and availability is limited to the US East region.

Koalify offers a free sandbox or demo account, or portals below 10K records, and a Pro plan at $25 a month billed annually. Plauti Deduplicate has a free Salesforce edition; full-edition features require a trial or custom pricing. For a research-oriented project, pyJedAI is identified as a research project by the University of Athens AI-Team. Python Record Linkage Toolkit is a free package aimed at research use and small or medium files, while Splink is free MIT-licensed Python software for Linux and self-hosted use.

Verdict

Entify is a sensible free choice for teams that need guided, explainable record matching and want to keep data local. The ability to inspect evidence and trace merges is its clearest advantage. Choose another tool for larger datasets, parallel runs, or deployments that need an API safely exposed beyond localhost.