Kyligence

Editor’s note: This Kyligence review is based on my evaluation of Kyligence Enterprise (v5.2, plus the Cloud 4.5.8 build on AWS Marketplace) and Kyligence Zen with Copilot in October 2026. I weighed pricing, setup, the OLAP engine, the metrics layer, security, product momentum, and rivals.

Quick verdict: I recommend Kyligence to enterprises running Apache Kylin or replacing SSAS cubes on a Hadoop or Spark lake, who need MDX and cell-level security. No public price, login-gated docs and a quiet English site since 2024 mean you need a written roadmap and support commitment before signing.

Kyligence, founded in 2016 by Apache Kylin’s creators, pre-computes indexes so BI queries on large data lakes return fast. In this review, I’ll take a closer look at Kyligence’s pricing, features, and product momentum, so you can see exactly whether it deserves a demo.

Key Takeaways:

• Enterprise adds multi-tenancy, Smart Pushdown, and row, column, and cell-level permissions to Kylin.
• Zen adds a metrics catalog, ZenML, Excel connectivity, and Copilot, with a no-card trial.
• There is no vendor price list. Zen on AWS bills $8.00 per undefined unit; Cloud needs an external license or private offer.
• The latest English blog post is dated February 20, 2024; AWS Cloud still lists Enterprise 4.5.8.
• Apache Kylin 5.0.5 shipped in July 2026 under Apache 2.0, a free, actively released alternative.

Pros and Cons

How Much Does Kyligence Cost?

Kyligence publishes no price list for Enterprise, Cloud, or Zen. Every product page routes to “Book Demo” or a free trial, so none of the three can be budgeted without a quote.

• Kyligence Zen (free trial, no credit card): For BI owners evaluating the metrics catalog and Copilot through self-serve registration at zen.kyligence.io. Kyligence publishes no paid rate for what follows the trial.
• Kyligence Zen on AWS Marketplace ($8.00 per unit): One Standard usage dimension covering the metrics catalog, Copilot, root cause analysis and the open API, with no tiers or add-ons. Subscriptions have no end date and cancel any time, AWS infrastructure is billed separately, and the listing never says what a unit measures.
• Kyligence Cloud on AWS (license bought outside AWS): For deploying Enterprise through a CloudFormation template. The listing advertises a 60-day free trial, while registering at account.kyligence.io supplies a 30-day trial license. No refunds, AWS infrastructure is separate, and a second listing is private offer only.
• Kyligence Enterprise (quote only): For on-premises or your own cloud, with SLA support. Kyligence publishes no SLA terms, so negotiate the license and the support obligations in the same contract.

Product	Public price	Trial	Billing
Kyligence Zen	No vendor rate	Free, no credit card	No published paid rate
Zen on AWS Marketplace	$8.00 per undefined unit	Not stated	Standard usage dimension; AWS infrastructure separate
Kyligence Cloud on AWS	External license or private offer	60-day listing trial; 30-day license	License plus AWS infrastructure; no refunds
Kyligence Enterprise	Quote only	Contact sales	Negotiated contract
Apache Kylin 5.0.5	$0 software license	Not applicable	Apache 2.0; community support

Imply Polaris starts at $100 per month with a $500 trial credit. ClickHouse Cloud’s Basic example starts at $66.52 per month with $300 in trial credits, priced for a service active 6 hours a day.

Cube Cloud Starter costs $40 per developer per month, with dedicated deployment at $0.60 per hour on top. Every one of those rivals lets me price a pilot before a sales call; Kyligence’s undefined AWS unit does not.

Is Kyligence Good Value for Money?

• Good fit: Enterprises replacing SSAS or running Kylin at scale, where MDX compatibility and an SLA justify a negotiated license.
• Less convincing fit: Teams with only a few terabytes or no Hadoop or Spark lake. ClickHouse or Imply cost less to evaluate and need no pre-built model.
• Budget priority: Start with Apache Kylin 5.0.5 at $0 in license fees if your team can operate it, and pay for Kyligence only when MDX, cell-level security or an SLA is a hard requirement.

Reviewer’s Notes: Use the Zen trial for the metrics layer and the 60-day Cloud trial to size the workload. Then ask for the price in writing together with the current Enterprise version, the release cadence and the support SLA, because none of those is public.

My Experience With Kyligence

Kyligence has two separate entry points: a self-serve Zen trial for the metrics layer, and a CloudFormation template on AWS Marketplace for the Enterprise OLAP engine. A Zen trial says nothing about Enterprise index builds, BI concurrency or infrastructure costs, so I treated them as two evaluations.

Signing Up for Kyligence Zen

Zen’s registration starts at zen.kyligence.io and the trial needs no credit card. It is the right entry point for BI owners evaluating shared metric definitions, not the data-lake engine underneath.

The first step is uploading data or starting from sample datasets and metric templates. Kyligence claims templates save 80% of the setup time, but a template is a starting point, not proof that a company’s existing KPIs will transfer unchanged.

Next comes the Metrics Catalog, where each metric becomes a reusable business definition. ZenML, Kyligence’s YAML-based metrics language, defines the same metrics, dimensions and underlying relational datasets as code. I prefer that explicit step over letting chat define a KPI.

Copilot then searches and analyzes those metrics in natural-language chat, runs root cause analysis and builds dashboards. For an Excel-centered team, the Excel connector matters as much as Copilot, because the same catalog has to serve PivotTable-style analysis, not just conversational questions.

Starting a Kyligence Enterprise Trial on AWS

The AWS path begins with launching the CloudFormation template from AWS Marketplace, then registering at account.kyligence.io to download the 30-day trial license. That license is half the 60-day trial the listing advertises, so plan the evaluation around 30 days.

The Enterprise workflow runs in four steps:

1. Connect data: AWS sources include S3, Redshift, RDS, and Kafka.
2. Model the workload: Use the drag-and-drop modeler and AI modeling assistance to define dimensions and measures.
3. Build indexes: Pre-compute the OLAP structures that matching BI queries will hit.
4. Connect BI tools: Query the engine over SQL with JDBC/ODBC, MDX, or REST.

Cloud uses Spark or EMR for compute and S3 for storage, with pay-as-you-go compute. Budget the index-build stage separately from query serving, because a free software trial does not remove the AWS infrastructure bill.

Where the Docs Stop

Enterprise’s documentation index lists v5.2 as its newest book, alongside earlier v3.4, v4.5, and v4.6 books. Companion books cover Cloud and Manager v4.6, MDX v1.4, and Insight v1.3.

Every book and the release notes require a Kyligence account, and Zen’s documentation is also login-gated. That is a procurement disadvantage: limits, connector lists and version dates cannot be read before you register.

Apache Kylin 5 publishes its full documentation, including prerequisites: JDK 8, Hive, HDFS, YARN, ZooKeeper, and a MySQL or PostgreSQL metadata store.

Reviewer’s Notes: Create the Kyligence account before scheduling a technical evaluation, so the release notes can shape the test plan. Confirm which Enterprise version the CloudFormation template installs, because the AWS Marketplace listing still shows 4.5.8 from 2022.

The OLAP Engine: Pre-Computed Indexes, Smart Pushdown and BI Connectivity

Kyligence Enterprise combines pre-calculation, indexes, and cache in a hybrid OLAP engine, where Apache Kylin is pure MOLAP pre-calculation. Matching queries hit prepared indexes; Smart Pushdown sends queries that miss them to the underlying SQL engine.

That architecture suits repeatable, modeled BI workloads where the business already agrees on dimensions and measures. ClickHouse and Apache Druid answer ad hoc queries without a pre-built model, so they are stronger starting points for exploratory and event analytics.

Enterprise’s engine capabilities include:

• Smart Pushdown and raw-data drill-down: Queries that need detail beyond the prepared indexes still return.
• Intelligent auto modeling: AI modeling assistance helps define the analytical model.
• Read/write separation and multi-tenancy: Separate workloads share one deployment.
• Active high availability: Job and query nodes run in active HA by default.
• Spark scale-out: Processing distributes across Apache Spark.
• Warehouse connectivity: Hive, plus Impala, Oracle, and Greenplum connectors.
• Azure integration: Kylin and ClickHouse engines over Blob Storage and Data Lake Storage.

MDX connectivity carries the most weight for SSAS replacements that must keep Excel PivotTables. SQL connectivity matters more when Tableau or Power BI already owns the reporting.

Interface	BI connection	Buying relevance
SQL through JDBC/ODBC	Tableau, Power BI, other SQL clients	Existing SQL reporting
MDX	Excel and multidimensional BI	SSAS modernization
REST	Programmatic integrations	Application access

Kyligence also lists MicroStrategy, Cognos, Zeppelin, and Superset as supported BI tools. Validate your exact client versions and authentication path in the trial rather than relying on a compatibility list.

Kyligence claims sub-second SQL on petabyte-scale datasets and sub-second results for 1,000+ concurrent users. Those are Kyligence’s own figures, and it publishes no benchmark behind them.

Apache Kylin users report operational friction in the project’s issue tracker:

• Index-build overhead: Builds are Spark jobs that need sizing and scheduling; one open report describes a 71,734-row model with 97 indexes taking about 90 minutes.
• Pushdown edge cases: Reports cover case sensitivity and projects with differing pushdown configurations.
• Upgrade work: Kylin 4 metadata cannot move directly to Kylin 5; models and related components must be rebuilt.

These are trial checks, not proven Enterprise defects. Kylin 5 also added model and index recommendations, so automation alone is not a reason to buy the commercial product.

Best for: Governed dashboards, established models, and Excel-heavy SSAS modernization. Skip if: Constantly changing event queries matter more than reusing modeled aggregates; start with ClickHouse or Druid.

Kyligence Zen and Copilot: The Metrics Layer

Kyligence Zen reached general availability on April 10, 2023, with a Metrics Catalog, ZenML, the Kyligence Connector for Excel, Auto Root Cause Analysis, and dozens of metric templates. Its strongest case is defining a business metric once and reusing it across chat, charts and Excel.

The catalog supports unlimited derived and composite metrics, and ZenML expresses the same definitions in YAML. That semantic layer matters more than the chat window, because Copilot can only analyze the metrics someone defined well.

The Excel connector works with all Excel versions, including Office 365 on Windows and macOS. For spreadsheet-heavy teams, that is a more relevant test than a chat-only demo.

• Goals tracking: Follow business targets next to the metrics that measure them.
• Built-in visualization: Drag-and-drop charts without a separate BI tool.
• Open Metrics APIs: Embed Copilot or other metric applications in your own applications.
• Industry examples: Kyligence lists fleet maintenance, retail product-launch forecasting, healthcare KPI consolidation and restaurant menu management.

Copilot’s Model and Limits

Copilot launched in mid-2023 on Azure OpenAI, and its launch material names GPT-3.5. Current product pages say only that Azure OpenAI powers the platform, with no newer model named.

Kyligence offers Copilot for SaaS and private clouds. Its stated functions are metric search, analysis, dashboard creation, and root cause analysis through chat.

Kyligence’s own figures promise 10-second metric analysis and 20-second root-cause answers. Those are marketing figures, and Zen has no public release notes, connector list or documented limits to check them against.

Cube or dbt come first when public pricing and readable documentation are prerequisites. Cube Cloud has a free plan and Starter at $40 per developer per month, while dbt Starter includes its Semantic Layer at $100 per user per month.

dbt Starter includes 5,000 queried metrics per month; its free Developer plan excludes the Semantic Layer. AtScale is quote-only, like Kyligence Enterprise.

My shortlist: Zen for Excel plus metrics chat, Cube for API-first metrics, and dbt for teams already modeling there.

Security, Compliance and Product Momentum

Kyligence Enterprise supports row, column, and cell-level permissions, alongside end-to-end encryption and backup and recovery. Cell-level control is a real shortlist criterion when one shared model serves users with different access rights; Apache Kylin offers cube-level control only.

• Identity integration: LDAP and Kerberos connect Enterprise to existing identity systems.
• Azure permissions: Azure Active Directory permission management on Azure.
• SOC 2: Type I dates to August 2021 and Type II to August 2022.
• ISO certification: ISO 27001 and ISO 9001 date to March 2020.

Those certification dates are four to six years old, so request current compliance evidence for the deployment you are buying.

Is Kyligence Still Actively Developed?

Kyligence’s public signals for the commercial product do not show a current release cadence:

• English updates: The newest blog post is February 20, 2024; the newest newsroom item is November 27, 2023.
• AWS build: The Cloud listing still offers Enterprise 4.5.8, a 2022 build.
• Documentation version: Enterprise v5.2 appears in the docs index without a public release date.
• Funding milestone: The last listed round is a $70 million Series D from April 2021.
• Company location: Kyligence names San Jose and Shanghai as its dual headquarters.
• Site maintenance: The site-wide banner still carries the typo “fot.”

The AWS build and gated release history carry more buying weight than the banner typo, because they decide which version you would actually deploy.

Apache Kylin itself shipped 5.0.5 on July 13, 2026, has 3,782 GitHub stars, and had commits as recent as October 19, 2026. The open-source engine underneath is clearly alive; that activity says nothing about Kyligence Enterprise’s release schedule.

Before approving a contract, ask Kyligence for current release notes, the support SLA, and the roadmap in writing. Make the supported Enterprise version and the upgrade path contract terms, not a general promise of continued support.

How Does Kyligence Compare to Competitors?

Kyligence’s niche is modeled, governed BI with MDX on data lakes; real-time OLAP databases win on openness and published pricing, and dedicated semantic layers compete with Zen.

Tool	Starting price	Model	Best at
Kyligence	Quote only; Zen AWS $8.00 per undefined unit; trials available	Kylin-based OLAP plus metrics platform	MDX and cell-level security on lakes
Apache Kylin 5.0.5	$0, Apache 2.0	Open-source OLAP	Pre-computed analytics with community support
ClickHouse Cloud	Basic example from $66.52/month; $300 trial credits	Real-time column store	Ad hoc analytics at scale
Imply Polaris	Starter from $100/month; $500 trial credit	Managed Apache Druid	Streaming ingestion and interactive queries
StarRocks	Free open source; PhoenixAI managed pricing by quote	MPP OLAP	Lakehouse queries
Cube Cloud	Free; Starter $40/developer/month plus deployment costs	Semantic layer	API-first metrics for applications
AtScale	Quote only	Semantic layer	Excel and Power BI semantic models

• Kyligence vs Apache Kylin: Pick Kylin 5.0.5 when your team can run Hadoop and accept community support. Its model and index recommendations and internal tables make it more capable than Kyligence’s Kylin 4-era comparison implies. Choose Enterprise for MDX, cell-level permissions and an SLA, not for automation features Kylin now has.
• Kyligence vs ClickHouse: ClickHouse wins for changing analytical questions that do not fit a stable model. Its published billing examples make budgeting easier, though the $66.52 example assumes 6 active hours a day, not always-on production. Kyligence fits better when preserving Excel and modeled BI workflows drives the purchase.
• Kyligence vs Apache Druid and Imply: Imply Polaris is the pick for streaming and event analytics, with a $100 Starter plan and a $500 trial credit as a defined pilot route. Kyligence’s stronger argument is governed BI over an existing lake, where the model already exists and Excel users need MDX.
• Kyligence vs StarRocks: StarRocks suits MPP OLAP and lakehouse querying rather than SSAS cube replacement. Its managed service, formerly CelerData Cloud, is now sold as PhoenixAI, with a 30-day trial but no public starting price, so it shares Kyligence’s quote-first buying.
• Kyligence vs Cube and AtScale: Cube wins on API-first metric delivery and a free entry plan; dedicated deployment adds $0.60 per hour on Starter beyond developer seats. AtScale fits Excel and Power BI semantic models, but its quote-only pricing leaves the same budgeting gap as Kyligence, so it only wins on semantic-model depth, not cost clarity.

For a team already invested in dbt, add the dbt Semantic Layer to the shortlist before Zen. The $100-per-user Starter plan makes the comparison concrete, though the free Developer plan excludes semantic querying.

How I Reviewed Kyligence

I evaluated Enterprise and Zen separately. For Enterprise, I covered the Kyligence Cloud trial path on AWS Marketplace, modeling, index builds, Smart Pushdown, BI connectivity and security. For Zen, I covered the trial signup and the catalog-to-Copilot flow, including ZenML and the Excel connector.

I compared every purchasing route Kyligence offers, from Zen’s undefined AWS unit to Cloud’s external license and private offer, and weighed the public release record against Apache Kylin’s. I compared Kyligence with Apache Kylin, ClickHouse, Imply, StarRocks, Cube, dbt and AtScale. I did not measure query latency, concurrency or Copilot answer quality, so no verdict here rests on Kyligence’s speed claims.

Prices current as of October 2026.

Should You Buy Kyligence for OLAP on Your Data Lake?

Kyligence combines modeled OLAP, MDX connectivity, and granular security for enterprise BI, and Zen adds reusable metrics with Copilot. Opaque pricing and dated public release signals keep it from an unconditional recommendation.

Shortlist Kyligence for SSAS replacements on large data lakes where Excel users matter. Existing Apache Kylin teams that need an SLA also have a clear reason to talk to sales, provided the contract names a current supported release and a roadmap.

• Managed OLAP with clear pricing: Start with ClickHouse Cloud or Imply Polaris.
• A metrics layer with public pricing: Compare Cube and the dbt Semantic Layer before committing to Zen.
• Self-managed Hadoop analytics: Choose Apache Kylin 5.0.5 for a $0 software license when community support is enough.

Do not buy Enterprise for its advertised query speed alone, or Zen for Copilot alone. Start a Zen trial or the 60-day Cloud trial through Kyligence, and ask for the current release notes on day one.

FAQs

How much does Kyligence cost?

Kyligence has no public price list for Enterprise, Cloud, or Zen. Zen on AWS Marketplace costs $8.00 per unit, but the listing does not define the unit. Cloud on AWS needs a license bought outside AWS or a private offer, and Enterprise is quote-only. Ask for a workload-based quote that separates software from infrastructure.

What is the difference between Kyligence and Apache Kylin?

Kyligence Enterprise is the commercial product built on Apache Kylin by Kylin’s creators. It adds hybrid OLAP, multi-tenancy, read/write separation, Smart Pushdown, cell-level permissions, and SLA support. Apache Kylin 5.0.5 is free under Apache 2.0 and has added its own model and index recommendations, so compare current versions rather than Kyligence’s older feature matrix.

Does Kyligence have a free trial?

Yes. Kyligence Zen offers a free trial without a credit card. Kyligence Cloud on AWS advertises a 60-day trial, while registering at account.kyligence.io supplies a 30-day trial license, so plan around the shorter period. AWS infrastructure charges apply during the trial.

Which LLM does Kyligence Copilot use?

Kyligence Copilot runs on Azure OpenAI, and its 2023 launch material names GPT-3.5. Current product pages name no newer model, only that Azure OpenAI powers the platform. Confirm the exact model, and whether you are on the SaaS or private-cloud deployment, before you judge Copilot against your requirements.

Is Kyligence still being developed?

Kyligence’s docs index lists Enterprise v5.2, but with no public release date. The English blog last posted in February 2024, and the AWS Cloud listing still ships a 2022 build. Ask Kyligence for current release notes before signing. Apache Kylin, the open-source engine underneath, shipped 5.0.5 in July 2026.