QueryPanel is the embedded analytics platform for multi-tenant SaaS. Primary product: a headful React SDK (Notion-like dashboards and AI assistant). Secondary: a headless Node SDK for zero-trust custom UI. Searches that look like “fury query panel” almost always mean QueryPanel.

Is this another BI tool?

QueryPanel is infrastructure for embedding analytics and generating safe SQL, built for product surfaces, not standalone analyst workspaces.

How do you handle tenant isolation?

Tenant context is part of generation and execution, not something you toggle later in a dashboard filter.

What does implementation look like?

Attach your database with the SDK, wire auth claims, and embed UI where customers already work. Scope follows your surfaces, not building a query engine from scratch.

How do you keep AI SQL accurate as schemas evolve?

Ground generation in synced schema metadata and your knowledge base (annotations, glossaries, and examples) so retrieval and SQL stay aligned with how your data model actually works.

How are tenants counted?

A tenant represents a distinct customer or organization within your SaaS product. We count the number of unique tenants that access or generate analytics in a given billing month.

What counts as an AI chart?

An AI chart is generated whenever an end-user asks a natural language question that results in a new chart, or when an editor uses AI to create or modify a dashboard block.

What happens when limits are exceeded?

You can purchase overage credits for AI charts. For tenant limits, we'll notify you when you approach your plan's limit and help you upgrade to the next tier. We never cut off access for your end-users without warning.

How are AI models selected?

We support OpenAI, Gemini, Anthropic, and even OpenRouter. Bring-your-own-key is available from Starter and above (including Growth and Scale). Custom plans also support bring-your-own-model setups. If you want a specific provider or model path, ping us at csaba.ivancza@querypanel.io.

Metabase vs Sisense for embedded analytics (and where QueryPanel fits)

Open-source dashboards with a React SDK versus an enterprise OEM analytics platform, compared for SaaS embedding.

Metabase and Sisense sit at opposite ends of embedded analytics. Metabase is open-source BI that many teams already run internally; its modular embedding SDK renders React components on Pro and Enterprise plans, and it prices by user, counting embed users as users. Sisense is an enterprise platform built for OEM programs, with iframe embedding, the React-based Compose SDK (`@sisense/sdk-ui`), tenant hierarchies, ElastiCube data models, and quote-based pricing. QueryPanel is the third option for React and Postgres multi-tenant SaaS that wants an AI-assisted workspace tenants customize themselves, priced by tenant.

Last reviewed October 19, 2026. Vendor facts are checked against public docs and pricing pages on that date.

Start building Read the docs

Comparison at a glance

This table summarizes typical positioning. Every vendor changes over time—validate details against current documentation and your security review.

Dimension	Metabase	Sisense	QueryPanel
Embed model	Modular embedding SDK renders React components (Pro/Enterprise); guest embeds and full-app embedding use an iframe.	Classic iframe embedding, or Compose SDK React components for code-first embeds.	First-class `@querypanel/react-sdk` components—`QuerypanelEmbedded` for a full dashboard, or `QueryPanelProvider` with `QueryInput` / `QueryResult` for a bespoke flow. They render in your React tree like any other product screen (layout, router, modals, tokens)—not a separate iframe "mini app" on another origin. Mint short-lived JWTs on your server; never ship your workspace private key to the browser.
React SDK	`@metabase/embedding-sdk-react`: React 18 or 19, browser-only rendering, one dashboard per app page, version matched to your Metabase server.	Compose SDK: `@sisense/sdk-ui` for charts, dashboards, and AI components; `@sisense/sdk-data` for query construction.	`@querypanel/react-sdk` with `QuerypanelEmbedded` for the full workspace, plus provider and query components for custom flows.
Multi-tenancy mechanism	Tenants via a `@tenant` JWT claim, plus row and column security, connection impersonation, or database routing.	Row-based data security on a shared model, dedicated ElastiCubes per client, or self-contained multitenancy with a tenant hierarchy.	Tenant column declared at database attach; RS256 JWT minted by `@querypanel/node-sdk` carries the tenant ID; tenant filter applied at SQL generation and checked before execution.
AI / natural language	Metabot embedded chat (Pro/Enterprise, SSO embeds) builds questions from existing models and tables; documented not to write SQL or build dashboards.	Sisense Intelligence: chatbot and NLQ components in Compose SDK; bring-your-own LLM on the enterprise plan; AI actions metered from a credit pool.	Natural language to SQL and chart generation are the main path; the assistant edits tenant dashboards.
End-user dashboard editing	SSO embeds can expose the query builder and dashboard editing on paid plans; guest embeds are view-only.	Depends on tier and roles; reported self-service tiers range from view-only (Launch) to persistence and white-labeling (Grow).	Tenants fork shared dashboards and customize their own copy with AI help.
Semantic layer / modeling	Models, metrics, and saved questions curated in Metabase on top of your database.	ElastiCube in-memory models and live models with a full data modeling layer.	Gold SQL queries (curated examples the model prioritizes), database annotations (business context on tables/columns, re-embedded with schema), glossary (domain terms and definitions), and tenant-level definitions (isolation field, enforcement, and per-tenant sync context so every ask() is grounded in the right customer slice—not a one-size global prompt).
Data plane & credentials	Queries your database directly; self-hosting keeps credentials and results in your environment.	Live queries or data imported into ElastiCubes on the Sisense deployment, wherever you host it.	Headless Node path runs SQL with your own drivers, so credentials and rows stay on your servers.
Deployment	Self-host open source or paid editions, or Metabase Cloud.	Sisense SaaS, your own AWS, Azure, or GCP account, or on-premises.	Hosted service with React and Node SDKs; Node SDK adapters cover Postgres, ClickHouse, and BigQuery.
Pricing shape	Per seat: open source free; Pro $575/month with 10 users plus $12 per extra user; Enterprise custom. Embed users count as users.	Enterprise embedding quoted; S&P Global reported Launch $399/month and Grow $1,299/month (February 2026), with AI metered by credits.	Counts unique tenants per billing month, not seats; AI chart generation metered separately; free tier.
Time to first embed	Fast for a guest embed of an existing dashboard; SDK embeds add JWT SSO, a tenant strategy, and version alignment.	Longer: deployment, data models, security rules, and embedding setup; a 7-day trial covers a first test.	Short: attach a database, mint a JWT, render `QuerypanelEmbedded`; customers can ask questions immediately.

Seats versus quotes

Metabase publishes per-seat prices, and its pricing page states that users of your embeds count as users. At 50 tenants with 20 users each, Pro list pricing lands around $12,455 per month, which is the point where Metabase steers you to Enterprise. Sisense quotes enterprise embedding; an S&P Global 451 report republished by Sisense in February 2026 describes self-service Launch ($399 per month, view-only) and Grow ($1,299 per month, with persistence, basic SSO, and white-labeling) tiers, with credits metering AI and consumption. Ask both vendors for a quote at today's users and at twice that.

React SDK versus Compose SDK

Both vendors now offer React components rather than only iframes. Metabase's `@metabase/embedding-sdk-react` needs React 18 or 19, renders in the browser only, supports one dashboard per app page, and must match your Metabase server version. Sisense's Compose SDK exposes charts, dashboards, and AI components such as a chatbot through `@sisense/sdk-ui`, with a separate data package for queries; classic Sisense embeds remain iframe-based. Metabase gets you a full dashboard faster. Compose SDK gives you more building blocks to assemble. QueryPanel gives you a finished React workspace.

Tenant isolation models

Metabase's Tenants feature reads a `@tenant` claim from the JWT and pairs it with one of three strategies: row and column security on user attributes, connection impersonation to database roles, or database routing to a per-tenant database. Sisense documents a shared data model with row-based data security, dedicated ElastiCubes per client for tenants that need separate data, and self-contained multitenancy with a tenant hierarchy. Both work, and both need a design decision up front. QueryPanel uses a tenant column declared at attach and a server-minted RS256 JWT.

When Metabase or Sisense is the better fit

Honest tradeoffs help your team pick faster—and match how buyers actually decide.

Choose Metabase when cost and control come first: the open-source edition is free, self-hosts anywhere, and supports view-only guest embeds with signed JWTs.
Choose Metabase when internal analysts already use it and you want one tool for internal and simple customer dashboards.
Choose Metabase when your embedded user count is small, so Pro's per-seat pricing ($575 per month with 10 users, $12 per extra user) stays predictable.
Choose Sisense when you are running a formal OEM program with many tenants, need a tenant hierarchy in user management, and want white-labeling depth and vendor services behind you.
Choose Sisense when you need heavier data modeling, such as in-memory ElastiCubes or dedicated cubes per client, rather than querying the production database directly.
Choose Sisense when enterprise requirements such as a premium SLA, HIPAA readiness, or bring-your-own LLM for governed AI are part of the procurement checklist.

When QueryPanel is the better fit

Pick Metabase for low-cost, self-hosted dashboards and Sisense for a governed OEM program. Pick QueryPanel when your product is React on Postgres, tenants want to customize dashboards with AI help, and you want pricing that follows tenants rather than seats or quotes.

You want a trainable knowledge system—gold queries, DB annotations, glossary, and tenant-aware definitions—so NL→SQL and charts reflect your business, not generic schema-only guesses.
You want neither per-seat pricing nor an enterprise quote: QueryPanel counts unique tenants per month and has a free tier.
You want an AI assistant that writes SQL, renders charts, and edits tenant dashboards, rather than a question builder or a set of AI components you assemble yourself.
Tenants should fork shared dashboards and customize their own copy inside your React app, with no iframe.
You want a headless zero-trust Node path where database credentials and query results stay on your servers.
You run a React product on Postgres, ClickHouse, or BigQuery and do not want to operate a BI server or a cube build pipeline.

Pricing notes

Checked October 19, 2026. Pricing changes; confirm on the vendor pages before you budget.

Metabase and Sisense

Metabase: per its pricing page as checked October 19, 2026, open source is free with view-only guest embeds; Pro is $575 per month including 10 users plus $12 per extra user; Enterprise is custom. Embed users count as users. Sisense: enterprise embedded pricing is quote-based on its pricing page. An S&P Global 451 Research report republished by Sisense in February 2026 describes self-service tiers, Launch at $399 per month (view-only) and Grow at $1,299 per month (persistence, basic SSO, white-labeling), with credits metering AI and consumption. Confirm current terms with Sisense.

QueryPanel

QueryPanel pricing is tenant-based: plans count unique tenants per billing month, not end-user seats. AI chart generation is metered separately, and a free tier is available. See the pricing section for current plans.

Sources: Metabase pricing, Sisense pricing, S&P Global 451 report on Sisense (Feb 2026), QueryPanel pricing

Migrating from Metabase or Sisense to QueryPanel

1. Inventory embeds, dashboards, and models
From Metabase, list guest embeds, SDK components, dashboards, saved questions, and models, plus the number of customer user accounts. From Sisense, list embedded dashboards, Compose SDK components, ElastiCubes or live models, and data security rules. Rank everything by customer usage.
2. Map the tenant mechanism to a tenant column
Metabase user attributes, connection impersonation, or database routing, and Sisense row-based data security or per-client cubes, all resolve to a tenant key. Declare that column as the QueryPanel tenant field and mint RS256 JWTs with `@querypanel/node-sdk` from the endpoint that signs Metabase or Sisense tokens today.
3. Turn models into trainable knowledge
Sync the schema, then convert Metabase models and saved questions, or Sisense model formulas, into gold SQL queries, column annotations, and glossary terms. If Sisense ElastiCubes pre-aggregate data, decide whether QueryPanel should query the source tables or a reporting schema you already maintain.
4. Embed QuerypanelEmbedded on one route behind a flag
Render `QuerypanelEmbedded` from `@querypanel/react-sdk` on one analytics route for internal users and a few pilot tenants. Recreate the top dashboards as shared templates tenants can fork. Keep the existing Metabase or Sisense embed elsewhere.
5. Compare numbers per tenant side by side
For each pilot tenant, compare totals and trends against the existing dashboard for identical filters and dates. Cube refresh schedules and cached results are common sources of drift; note them before you chase SQL differences.
6. Cut over and retire the old embed path
Move remaining tenants, then remove the old SDK packages, token or SSO endpoints, and customer user accounts. Keep Metabase or Sisense for internal reporting if analysts still use it, on a smaller plan.

Comparison FAQ

Is Metabase or Sisense better for embedded analytics?

Metabase is better for teams that want low cost, open source, self-hosting, and simple dashboards, especially with a small number of embedded users. Sisense is better for enterprise OEM programs that need tenant hierarchies, heavier data modeling, white-labeling depth, and vendor services. For a React SaaS product where tenants customize dashboards with AI, also evaluate QueryPanel.

Which is cheaper, Metabase or Sisense?

For small deployments, Metabase: the open-source edition is free, and Pro lists $575 per month with 10 users plus $12 per extra user, but embed users count as users. Sisense quotes enterprise embedding; S&P Global reported Launch at $399 and Grow at $1,299 per month in February 2026, with AI metered by credits. QueryPanel counts tenants per month and has a free tier.

Do Metabase or Sisense use iframes?

Both offer iframe and non-iframe paths. Metabase's guest embeds and full-app embedding use iframes, while its modular embedding SDK renders React components on paid plans. Sisense's classic embedding uses iframes, while its Compose SDK renders React components. QueryPanel's `QuerypanelEmbedded` renders the whole workspace as React components in your app.

Which has better AI for embedded analytics, Metabase or Sisense?

Sisense goes further: Compose SDK includes chatbot and natural-language query components, and its enterprise plan supports bring-your-own LLM, with AI actions metered from a credit pool. Metabase's embedded AI chat builds questions from existing models and tables and, per its docs, does not write SQL or build dashboards. QueryPanel's assistant writes SQL, renders charts, and edits tenant dashboards.

Can I self-host Metabase or Sisense?

Yes, both. Metabase's open-source and paid editions run on your own infrastructure, or you can use Metabase Cloud. Sisense deploys as SaaS, in your own AWS, Azure, or GCP account, or on-premises. QueryPanel is a hosted service, but its headless Node SDK runs SQL in your environment so credentials and results stay on your servers.

When is QueryPanel a better choice than both?

When the product is React on Postgres, ClickHouse, or BigQuery, tenants want to customize their own dashboards with AI, and neither seat pricing nor a long enterprise rollout fits. QueryPanel is not a BI suite: it lacks pixel-perfect reports and scheduled PDFs, and you should test NL-to-SQL accuracy on your own schema.

Still evaluating Metabase, Sisense and QueryPanel?

Start on the free tier, embed one dashboard, and compare implementation time against your current shortlist.

Start free All comparisons