Luzmo

Editor’s note: This Luzmo review is based on my hands-on evaluation of Luzmo as it is sold in October 2026 (the Embedded Everywhere plan and embed SDK 6.13). I weighed pricing, embedding and multi-tenancy, data connections and Warp, Luzmo IQ and AI agents, and hosting and security.

Quick verdict: I recommend Luzmo for SaaS teams that want customer-facing dashboards, self-service editing and AI answers without building a charting layer. The trade-off is a EUR 1,995 a month annual floor with no smaller tier, plus Warp’s default once-a-day sync.

Key Takeaways:

• One plan from EUR 1,995/month, billed annually, with 500 AI conversations and 100 million Warp synced rows a month.
• 19 native database connectors query live, so data stays in your database by default.
• Luzmo IQ, a hosted MCP server and an Agent API (June 2026) answer questions in-product and for agents.
• Warp syncs daily by default; faster intervals need a request to Luzmo.
• No published uptime SLA; a custom SLA is a paid enterprise add-on.

Among embedded analytics tools, Luzmo (formerly Cumul.io, renamed September 2023) is the most complete single-plan package I looked at: dashboards, an end-user editor, Luzmo IQ and a hosted MCP server.

In this review, I’ll take a closer look at Luzmo’s pricing, embedding, data connections and AI features, so you can see exactly whether its annual commitment fits your product.

Pros and Cons

How Much Does Luzmo Cost?

Luzmo sells one plan, Embedded Everywhere, from EUR 1,995 per month, billed annually. There is no smaller tier and no monthly billing, and the list price is in euros only.

• Embedded Everywhere (EUR 1,995/month): for SaaS teams that want white-label dashboards, end-user editing, Luzmo IQ and the MCP server under one contract. The platform fee stays fixed for the contract term, and usage is the variable part of the bill.
• AI conversations (500/month included): a conversation is a governed interaction between an end user and Luzmo AI in your product. Overage runs at a contract unit rate that Luzmo does not publish.
• Warp synced rows (100 million/month included): extra rows cost EUR/$0.25 per additional million. The allowance counts rows synced through Warp, not dashboard viewers.
• Monthly active users (contract-dependent): some contracts scale on MAUs instead. An MAU is an end user who uses the embedded analytics in a billing month, not an internal seat, and the rate is agreed at signature.
• Extra usage (measured monthly): extra volume is billed at the same unit rate, with no retroactive back-billing.
• Free trial (no credit card): the trial unlocks the complete product, so the editor, embedding and IQ can all be tested before you commit.
• Enterprise add-ons (quote only): private infrastructure, a custom SLA, stronger compliance controls, source-code escrow and enterprise implementation support. Luzmo sells these as the same product with tailored deployment, not extra features.

Item	Included	Extra usage
Platform fee	EUR 1,995/month, billed annually	Fixed for the contract term
AI conversations	500 a month	Contract unit rate, not published
Warp synced rows	100 million a month	EUR/$0.25 per extra million
Monthly active users	Contract-dependent	Rate agreed at signature
Enterprise add-ons: private infrastructure, custom SLA, compliance controls, source-code escrow	Not included	Quote

Is Luzmo Good Value for Money?

• Good value if you would otherwise build it: one plan replaces three in-house builds, a charting layer, a customer-facing editor and AI chat. For static reports alone, the fee is hard to justify.
• Cheaper entry elsewhere: Metabase Pro costs $575 a month with multi-tenant embedding, though its per-user fees count embed users. Power BI Embedded bills A-SKU capacity by the hour and can be paused.
• No public prices to compare: Embeddable, Qrvey, GoodData and Sisense all quote privately, so Luzmo’s visible price does not make it the most expensive option by default.
• Weak fit before revenue: the floor works out to EUR 23,940 a year before overage or add-ons. For a pre-revenue product, I recommend Metabase instead.

Reviewer’s Notes: Embedded Everywhere is my pick for funded SaaS teams that need the editor and AI alongside dashboards. Use the trial to build one tenant’s dashboard end to end, then get the AI conversation overage rate, MAU terms and any SLA in writing before you sign the annual contract.

My Experience With Luzmo

Signing Up and Picking a Region

I signed up for the free trial without a credit card. Luzmo asks for a hosting region at signup: EU in AWS Ireland or US in AWS Ohio.

That choice sets the endpoints for everything that follows: app.luzmo.com and api.luzmo.com for the EU, app.us.luzmo.com and api.us.luzmo.com for the US. On a US account, the embed component also needs apiHost and appServer pointed at the US addresses, so pick the region your customers’ data residency demands before you connect anything.

Connecting Data

I connected a PostgreSQL database through the native connector and imported the tables I needed as datasets. Nothing was copied: queries go live to the database whenever a dashboard opens, and Luzmo generates the SQL itself, pushing aggregations and filters down to PostgreSQL. For tighter control over that SQL, Luzmo accepts database views, SQL datasets or a plugin.

Query caching is a separate switch, per dataset or per connection. Cached results are held for a set period and only fill from user-triggered queries, unlike Warp’s scheduled copy. Warp itself sits in the dataset’s three-dot menu, and only the dataset owner can activate it.

Building the First Dashboard

I built the dashboard in Luzmo Studio’s drag-and-drop editor. The chart and filter set covers a typical customer dashboard:

• Charts: area, bar, box plot, donut, heat map, maps, pivot table and word cloud.
• Filters: date, dropdown and slider components.
• Formulas: nested aggregations, generally available on all connectors since July 2026, plus a SPLIT_PART operator.

For multi-tenancy, I put the tenant filter on the dataset as a parameter, so each embed token fills in a different customer.

Embedding It in an App

Embedding starts with an API key and token, created by an Organization Owner in profile settings and kept server-side only. My backend called the create-authorization endpoint (type embed) with username, name, email and access, added parameter_overrides for the tenant value, and returned the authorization ID and token to the frontend.

Tokens expire after 24 hours by default, up to a maximum of one year. The role defaults to viewer, with designer and owner for editing, and an optional inactivity_interval (minimum 2 minutes) limits idle sessions.

On the frontend, one package per framework renders the luzmo-embed-dashboard component with authKey, authToken and dashboardId:

• React: @luzmo/react-embed
• Angular: @luzmo/ngx-embed
• Vue: @luzmo/vue-embed
• Web components: @luzmo/embed

Reviewer’s Notes: Put row-level tenant filters at dataset level, never only on the dashboard. A user with use rights can duplicate a dashboard and drop a dashboard-level filter, which would expose other tenants’ rows.

Embedding, Multi-Tenancy and the Flex SDK

Luzmo supports three multi-tenancy patterns, so you do not have to force every customer into one shared database:

• Shared dataset: parameter_overrides filters rows per tenant.
• Database per tenant: account_overrides swaps connection credentials per token.
• Custom fields per tenant: covers columns that differ between customers.

I rate the connection override highest for B2B SaaS products whose larger customers demand their own database: the dashboard stays the same and only the token changes. Metabase Pro also offers multi-tenant embedding, but per-user fees there count every embed user, while Luzmo’s base plan has no fixed MAU quota.

• Editor modes: view, editLimited and editFull; editing needs a designer or owner role and at least use rights.
• White-label theming: dashboards carry your product’s styling instead of Luzmo branding.
• Localization: multi-lingual, multi-currency and multi-timezone support.
• Governance: a semantic layer and an access control layer.
• Feature overrides: the embed token accepts optional feature_overrides, set per token.

The Flex SDK is the code-first route: individual chart components for React, React Native (@luzmo/react-native-embed), Angular, Vue and web components, for teams that want a chart beside their own controls rather than a whole dashboard. The chart set matches Studio’s, from area and bar charts to pivot tables and word clouds. Sisense answers with its Compose SDK and iframe embedding, while Embeddable keeps models in your repository on a Cube-based stack.

Capability	Luzmo support	Constraint
Embedded editing	editLimited or editFull	designer or owner role plus use rights
Core API	10 calls per second per token	Bursts to 200, then returns 429
Scheduled exports	CSV, PNG, PDF by email	Single charts
Download permissions	CSV/XLSX exports	Can be restricted

The Core API’s 10 calls per second per token is the limit to design around: a page that fires many requests on one token will hit 429 once the 200-call burst is spent.

Recurring email exports now run on single charts, and CSV/XLSX downloads can be restricted, which matters when a customer who can view a chart should not be able to extract the rows behind it. PDF and PNG exports also reflect the viewer’s picker settings.

Data Connections and Warp

Luzmo ships 19 native database connectors, 2 web-service connectors (Google Analytics and Google Drive), plugins, API push and local upload:

• Warehouses: Snowflake, BigQuery, Redshift, Databricks and Azure Synapse.
• Analytical engines: ClickHouse, Trino, Presto and DuckDB (added May 2026, needs the httpserver extension).
• Operational databases: PostgreSQL, MySQL, MariaDB, SQL Server, Oracle, MongoDB and SAP HANA.

Database and plugin connections query live, so data stays in your database unless you switch on caching or Warp. API-pushed data, uploads and web-service data are copied into Luzmo’s OLAP store instead.

For a source with no native connector, the Plugin API takes a REST adapter with /authorize, /exchange, /datasets and /query endpoints. That adapter is code your team builds and maintains.

How Warp Works

Warp copies a dataset into Luzmo’s ClickHouse-backed store and runs accelerated queries against that copy; Luzmo’s own benchmark claims up to 25x faster dashboards on a non-optimized PostgreSQL database.

• Full: reloads the whole dataset, at most once a day.
• Delta: syncs new, updated and deleted rows; needs a unique ID column (composite up to 3 columns) plus created_at, updated_at or deleted_at as applicable.
• Full-delta: delta runs plus a full sync every N runs, with N at least 2.

The default and minimum interval is one day. Intervals in hours, minutes or seconds have to be enabled by Luzmo, so settle freshness before you sign. Warp works on database and plugin datasets, not on local files or web-service data.

Failed syncs retry up to 8 times, then fall back to the last good sync, or to the source if none has succeeded. The owner gets an email once data is out of sync for 1.5x the schedule. Since May 2026, safe schema changes such as new columns recover with a full reload; primary key and tracking-column changes still need manual review.

Luzmo IQ and AI Agents

Luzmo IQ answers data questions inside your product through a chat component (full page or popover) and an answer component that returns charts, text or both. Both ship for web components, Angular, React, React Native and Vue, and theme through CSS variables.

Since July 2026, IQ can be switched on for any embedded dashboard from Studio’s dashboard settings, with no re-embedding, and its answers respect the filters active on that dashboard. AI Context, from the same release, attaches business definitions and caveats to datasets, columns and dashboards through a Markdown editor or the REST API. Fill it in before you judge answer quality on any metric with a company-specific meaning.

Embedded MCP Server and Agent API

The June 2026 release added a hosted Embedded MCP Server for agents such as Claude or ChatGPT:

• IQ answers: for customers on IQ.
• Chart generation: for all customers.
• Dataset and column search: for all customers.
• Access control: the embed token limits what an agent can see.

The same release unified the Agent API at /AIPrompt. /IQMessage still works, but new capabilities ship only on /AIPrompt. For AI coding tools, @luzmo/agent-skills installs 10 skills and 4 subagents, and Luzmo Analyst (May 2026) needs your customer success manager to enable it in your own app.

GoodData’s Enterprise plan also has an MCP server, so agent access alone does not set Luzmo apart. Luzmo’s edge is getting it in the base plan, next to dashboards and IQ.

AI Data Handling and Cost

IQ and AI summaries send summarized, aggregated data to OpenAI as a sub-processor, and that data may include personal information. AI can be disabled in organization settings.

The plan includes 500 AI conversations a month, and the overage rate is set in your contract, so price it before you roll IQ out to every customer.

Hosting, Security and Reliability

Luzmo runs multi-tenant hosting in EU (AWS Ireland) or US (AWS Ohio), chosen at signup. A dedicated cloud in any public AWS, Azure or GCP region is a quote-based add-on.

• SOC 2 Type 2: reports for 2025 and 2026, on request.
• Penetration test: a 2025 third-party report, on request.
• Privacy regimes: GDPR, UK GDPR, CCPA/CPRA, FADP and PIPEDA, among others. Luzmo also claims NIS2, EU Data Act and EU AI Act compliance, without certificates.
• Encryption: at rest and in transit; database connections use TLS 1.2 or higher.
• HIPAA: Luzmo offers a BAA for HIPAA workloads.
• SSO: OIDC for the Luzmo organization login, one configuration per organization.
• ISO 27001: not listed.

The OIDC setup covers your team’s Luzmo login only; your customers get in through embed tokens. If your procurement checklist requires ISO 27001, raise it before the trial, because Luzmo does not list it. Warp copies are stored in your chosen EU or US tenancy and removed when you deactivate Warp.

Luzmo publishes no standard uptime SLA, and a custom SLA is an enterprise add-on. EU API uptime stood at 99.998 percent over the 90 days to October 22, 2026, but history is not a contract: if your product promises availability to its own customers, get the SLA in writing.

How Does Luzmo Compare to Competitors?

Luzmo wins on AI depth and an all-in single plan; Metabase wins on entry price, Qrvey on self-hosting, Embeddable on code-first models and Power BI Embedded on existing Microsoft licensing.

Tool	Starting price	Model	Best at
Luzmo	EUR 1,995/month, billed annually	Hosted embedded analytics, EU or US, dedicated cloud option	SaaS dashboards plus AI answers in one plan
Metabase	Open source free; Pro $575/month	Open source BI with embedding	Low-cost start, self-hosting
Power BI Embedded	Hourly A-SKU capacity, pausable	Azure capacity	Microsoft shops
Sisense	No public price	Enterprise analytics with Compose SDK	Large enterprises needing a 99.99% premium SLA
GoodData	No public price; per workspace, unlimited users	Analytics platform with embedding	Many tenants on a per-workspace bill
Qrvey	No public price; flat rate	Deploys into your own cloud on Kubernetes	Self-hosted, unlimited tenants
Embeddable	No public price; flat subscription	Code-first, models in your repo	Engineering-led teams

• Luzmo vs Metabase: Metabase is free as open source and $575 a month on Pro with multi-tenant embedding, but per-user fees count embed users and Enterprise starts at $20,000 a year. Pick it for internal BI or a price-sensitive start.
• Luzmo vs Power BI Embedded: hourly A-SKU capacity that you can pause suits Microsoft shops with Azure already in place, but publishing users need Pro licenses. Luzmo has no separate licensing layer for the people who build dashboards.
• Luzmo vs Sisense: Sisense offers a 99.99% premium SLA on Enterprise plus its Compose SDK, where Luzmo publishes no SLA at all. Pick Sisense when availability terms decide the deal.
• Luzmo vs GoodData: GoodData prices per workspace with unlimited users and data, and its Enterprise plan adds an MCP server, SAML/OIDC and an SLA. Its annual plans cannot be downgraded mid-term, and it wins when you serve many tenants on a per-workspace bill.
• Luzmo vs Qrvey: Qrvey deploys into your own cloud on Kubernetes at a flat rate, with unlimited tenants and users and a perpetual option. Pick it when hosted infrastructure is ruled out.
• Luzmo vs Embeddable: Embeddable charges a flat subscription with unlimited usage and keeps models in your repo. Luzmo wins when the embedded editor and IQ are must-haves and you would rather not write chart code.

Explo, once a common Luzmo alternative, was acquired by Omni on October 22, 2025, with its founders supporting about 12 months of transition, so I left it off the table. If your shortlist stretches to heavier BI suites that also embed, Looker and Tableau are the usual names.

How I Reviewed Luzmo

I worked through Luzmo’s trial path end to end: picking a region, connecting a PostgreSQL database, building a dashboard in Studio and embedding it with token-based tenant filters. I also went through the Flex SDK and IQ components, Warp’s sync modes against live queries, the plan and usage terms, and the hosting and security options.

I compared cost and capabilities with Metabase, Power BI Embedded, Sisense, GoodData, Qrvey and Embeddable. My verdict weighs customer-facing scope, tenant isolation, data freshness and first-year cost, including the EUR 1,995 monthly floor and Warp overage at EUR/$0.25 per extra million rows; it is not a speed benchmark.

Prices current as of October 2026.

Should You Use Luzmo for Embedded Analytics?

Luzmo is the most complete single-plan embedded analytics and AI package I looked at: dashboards, an end-user editor, IQ and a hosted MCP server under one contract. The limits are the EUR 1,995 a month annual floor, Warp’s daily default sync and no published SLA. Live queries by default and token-based tenant filters make it a safe fit for products with strict data residency, as long as one of the EU or US regions works for you.

• Funded B2B SaaS teams that want customer dashboards, end-user editing and AI answers live this quarter: buy it, and negotiate the AI overage rate, MAU terms and SLA before signing.
• Early-stage products or internal BI: start with Metabase instead, where Pro costs $575 a month.
• Teams that must self-host: choose Qrvey, which deploys into your own cloud on Kubernetes.
• Microsoft-standardized companies: choose Power BI Embedded and its pausable hourly capacity.

Start the Luzmo free trial and embed one tenant’s dashboard before you talk to sales.

FAQs

How much does Luzmo cost?

Luzmo costs from EUR 1,995 per month, billed annually, on its single Embedded Everywhere plan. That includes 500 AI conversations and 100 million Warp synced rows a month. Extra Warp rows cost EUR/$0.25 per million, while the AI conversation overage rate is set in your contract. See How Much Does Luzmo Cost? above for the full breakdown.

Is Luzmo the same as Cumul.io?

Yes. Cumul.io was renamed Luzmo in September 2023, and the product carried on under the new name. Today it sells one Embedded Everywhere plan from EUR 1,995 a month, billed annually, with the same embedded dashboards plus Luzmo IQ.

Does Luzmo query my database live?

Yes. Database and plugin connections query live by default, and data stays in your database unless you enable query caching or Warp. With Warp on, queries hit a synced copy in Luzmo’s store instead. API-pushed data, uploads and Google Analytics or Google Drive data are always stored in Luzmo’s OLAP store.

Does Luzmo have a free trial?

Yes. Luzmo offers a free trial of the complete product with no credit card. Use it to connect real data, embed one tenant-filtered dashboard, test the editor modes and switch on IQ before you talk to sales, because the paid plan is billed annually from EUR 1,995 a month.

Where does Luzmo host data?

Luzmo hosts data in the EU (AWS Ireland) or the US (AWS Ohio), and you choose the region at signup. A dedicated cloud in any public AWS, Azure or GCP region is available as a quote-based add-on. Database connections query live, so data stays in your database unless you enable caching or Warp. Warp copies stay in the tenancy you chose and are removed when you deactivate Warp.