Great Expectations (GX)

What it is

Python data-quality framework: Expectations are “expressive and extensible unit tests for your data”, and validation results generate documentation (GX Core docs). Tagline on GitHub: “Always know what to expect from your data.”

Maker, history

Open-source project created by Superconductive (the company that built GX Cloud; its blog now lives on greatexpectations.io). Superconductive announced a USD 21M raise on 2021-05-20 (GX blog; round details were not in the page text I could read).

Fivetran stewardship (not an acquisition announcement)

2026-05-13: Fivetran announced it will become steward of the Great Expectations open-source community and GX Core. The release says GX Core “will continue as an open source, community-driven project”, that Fivetran will support maintenance, ecosystem integrations and community engagement, and that it expects to hire engineering talent from the GX Core team. It does not say whether this is an acquisition of the company and does not address GX Cloud (Fivetran press release). The GitHub repo great-expectations/great_expectations now redirects to github.com/fivetran/great_expectations (Apache-2.0, 11,866 stars, pushed 2026-10-07; GitHub API, 2026-10-08). Fivetran is itself part of the merged Fivetran + dbt Labs company (see fivetran).

Editions and deployment

• GX Core: Apache 2.0, about 11.9k GitHub stars, pushed 2026-10-07 (GitHub API); docs and PyPI show version 1.24.0 (2026-10-07); Python 3.10-3.14. Runs inside your pipeline.
• GX Cloud: managed offering run with a customer-side GX Agent (so datasets stay in the customer environment). The current GX docs sitemap lists only 0.18-era Cloud pages and /docs/cloud/ returns 404 (2026-10-08), so product detail rests on GX blog posts below. Plans and pricing: see greatexpectations.io (no amounts recorded).

Core architecture

Python library; expectations run against pandas/Spark/SQL data sources; results and Data Docs. Concept names beyond “Expectations” not verified.

Role in an enterprise AI rollout

• Quality: schema, null, range and distribution tests on training sets, RAG source tables and feature data; CI gate.
• Lineage / access / PII / semantics: out of scope; use a catalog (data-catalogs-compared).

AI features as of October 2026

GX Core docs I fetched do not mention AI or MCP. For GX Cloud (GX blog):

• ExpectAI: AI recommendation engine that proposes Expectations; labelled beta in the 2025-05-01 post, which added an approval workflow (suggestions kept 48 hours, email when ready).
• 2026-02-24: bring-your-own OpenAI credentials for GX Cloud Agent deployments so ExpectAI runs inside the customer environment, with the post saying GX Cloud cannot access the underlying datasets (“now available”; demo request).
• 2026-04-21: GX blog positions data quality as a pillar of an “AI context layer”.
• MCP: no MCP page or post appears among the 893 URLs of the greatexpectations.io sitemap or the docs sitemap (2026-10-08); treat as not existing/unannounced.

Integrations

Orchestrators and warehouses via Python; catalogs can ingest results (for example openmetadata and datahub have quality integrations; not verified here).

Strengths and weaknesses (opinion)

Largest open-source mindshare for code-first data tests; more setup than contract-first tools such as soda-data-quality, no ML anomaly detection like monte-carlo-data.

Self-learning

• Docs: https://docs.greatexpectations.io/ (free); repo https://github.com/great-expectations/great_expectations. No certifications verified.

Sources

Second pass (2026-10-08): https://www.fivetran.com/press/fivetran-to-become-steward-of-the-great-expectations-open-source-community-and-gx-core-project ; https://www.greatexpectations.io/blog/gx-expectAI/ ; https://www.greatexpectations.io/blog/secure-ai-powered-data-quality-expectai-for-the-agent/ ; https://www.greatexpectations.io/blog/trustworthy-ai-agents-the-data-quality-pillar-of-the-ai-context-layer/ ; https://www.greatexpectations.io/blog/superconductive-announces-usd21m-raise-for-great-expectations/ ; https://www.greatexpectations.io/sitemap-0.xml ; https://api.github.com/repos/fivetran/great_expectations.
Fetched 2026-10-07: https://docs.greatexpectations.io/docs/home/ ; https://api.github.com/repos/great-expectations/great_expectations (Apache-2.0, 11,865 stars) ; https://github.com/fivetran/great_expectations ; https://pypi.org/project/great-expectations/ (1.24.0).

Open items

• GX Cloud current docs (404 on 2026-10-07 and 2026-10-08), GA status of ExpectAI after the 2025 beta label.
• Whether GX Cloud / the company is now owned by Fivetran: the 2026-05-13 release only says stewardship of the community and GX Core; deal terms not published on pages opened.
• Founders and the investors in the 2021 USD 21M round (not in the page text read). agy returned nothing in the second pass.