Entify

Entify is a free, open-source workspace for linking records and deduplicating datasets. It uses Splink 4 and DuckDB to compare records, then presents explainable clusters for review. Automatic setup identifies likely column types, suggests blocking rules, and explains its choices; users can override them. It reads CSV, TSV, Parquet, JSON, and Excel files, and reports row counts, distinct values, and completeness by column. Matching options include exact, Jaro-Winkler, Jaccard, and Levenshtein comparisons. Users can review match evidence and threshold analyses, export a merged dataset with source IDs retained, and generate a PDF audit report. It runs locally with Docker or with Python 3.11+ and Node 20+. Demo mode works without sign-in, and the Docker quickstart needs no accounts, API keys, or external services; uploaded data stays on the machine. Optional semantic blocking adds roughly 600 MB of Torch dependencies. Entify is available as web software, an API, or self-hosted, under an MIT license.

Entify combines record matching with review and export tools, including traceable merges and PDF audit reports. Its local operation is useful, but the upload, scale, and API security limits matter when planning deployment.

Entify is aimed at data stewards and teams that need record linkage without specialist expertise. It may suit users who want a local workflow with reviewable matches and audit records.

Overview

Entify is a free, open-source workspace for resolving entities, connecting records and removing duplicates from untidy datasets. It fits data stewards and teams that need guided record linkage without specialist expertise, particularly for local, moderate-sized jobs. Its transparent review and reversible exports are valuable, while its resource ceilings and insecure-by-default API rule out casual public deployment.

Key features

Entify profiles every column with row counts, distinct values and completeness, then auto-configures likely types and blocking rules. It explains those choices and lets users override them, giving newcomers a workable starting point without hiding the controls that experienced practitioners need. The matching engine combines Splink 4 with DuckDB and supports exact, Jaro-Winkler, Jaccard and Levenshtein comparisons.

Review is a genuine part of the workflow rather than an afterthought. Match evidence, score and threshold analyses, and charts help reviewers challenge questionable clusters before accepting them. Merge export leaves one surviving record per entity while preserving source IDs, so a decision can be traced or undone. PDF audit reports provide a portable account of the resolution process.

CSV, TSV, Parquet, JSON and Excel input covers the formats most teams encounter. Semantic blocking is available as an optional enhancement, but installing it downloads about 600 MB of Torch dependencies. That is a meaningful storage and setup cost if semantic methods are essential, and needless overhead for users satisfied with the standard comparisons.

Pricing

The Open-source plan costs free. It is MIT-licensed and includes the full Entify workspace, with no paid tier described. The trade-off is operational rather than financial: uploads are capped at 100 MB by default, and only one resolution result remains in memory at a time, so a later run replaces an earlier result. There is no separate seat allowance or trial term to manage.

Platforms

Entify is offered as an API, a self-hosted application and a web interface. Docker can run it locally without accounts, API keys or external services, keeping uploaded data on the machine. It can also be assembled with Python 3.11 or later and Node 20 or later; the stated stack includes FastAPI, Splink 4, DuckDB and Next.js.

Optional sign-in and shared storage use Clerk authentication and Supabase persistence, while demo mode works without either service. The backend API has no authentication and assumes localhost binding. Exposing it to untrusted callers would therefore require additional protection rather than being a safe default.

Who it's for

Choose Entify when a data stewardship team wants explainable fuzzy matching, a review queue and auditable, traceable merges without buying a hosted service. Its automatic setup is especially helpful when users understand their data but not blocking rules or match weights. Local Docker operation also suits sensitive datasets that should remain on one machine.

It is a poor fit for large production pipelines as supplied. Uploads are read fully into memory with a 100 MB default limit, and the project describes workloads beyond roughly 100,000 rows as needing a DuckDB file backend and job queue. Concurrent work is constrained because only one resolution result is held in memory.

Pros and cons

Pros

• Explainable configuration: Suggested types and blocking rules can be inspected and overridden, reducing the barrier for non-specialists.
• Auditable merges: Evidence views, threshold analysis, source IDs and PDF reports make decisions reviewable and reversible.
• Broad file support: CSV, TSV, Parquet, JSON and Excel cover common exchange workflows.
• Private local mode: Docker needs no account or external service, and data stays on the machine.
• No license cost: The MIT-licensed project is free to run and modify.

Cons

• Small-workload ceiling: The 100 MB cap and memory-bound design restrict larger datasets.
• Single-result concurrency: A second resolution run replaces the first in memory.
• Unsafe API exposure: The backend lacks authentication and is intended for localhost, not untrusted callers.
• Semantic setup overhead: Optional semantic blocking adds roughly 600 MB of Torch dependencies.

Alternatives

DupeCatcher is a free Android, iOS and web option when you want all features included without payment and do not need Entify's self-hosted workflow.

Workbooks is the better choice for a broader CRM platform, with a 30-day free trial and a CRM plan priced at $47 a month on a purchased 1-, 2- or 3-year subscription.

Plauti Deduplicate suits Salesforce users who want a free AppExchange edition, with full-edition features requiring a trial or custom pricing.

Python Record Linkage Toolkit is a free Linux and self-hosted alternative aimed at research use with small or medium-sized files.

Splink is the direct free, MIT-licensed Python-package alternative for teams comfortable building their own interface and workflow.

Tilores offers local mode for up to 100,000 records per deployment without an account or licence key.

AWS Entity Resolution is the paid API and web option for provider-based matching, billed per 1,000 records and requiring a provider licence or subscription.

Koalify fits hosted sandbox work, with a free tier for demo accounts or portals below 10K records and a Pro plan at $25 a month billed annually.

Verdict

Entify is a strong recommendation for data stewards and small teams that need free, explainable record linkage and traceable local exports. Choose it for guided matching and defensible merge review; look elsewhere for authenticated public APIs, concurrent jobs or datasets that exceed its memory and upload limits.