# Qrly

> Qrly is a self-hostable Business Intelligence + AI platform built in Belgium. AI is woven into every workflow — natural-language Ask (NL→SQL), AI anomaly detection, AI-described schemas, AI performance analysis, AI chart and filter suggestions — and runs on the LLM you choose: Anthropic Claude, Google Gemini, OpenAI, Azure OpenAI, Ollama or LM Studio (cloud or fully on-prem). On the BI side: QQL across 40 connection types compiled to 12 SQL dialects, OLAP star/snowflake models, 13 chart types, live dashboards (WebSocket / LISTEN-NOTIFY), embedded analytics, scheduled subscriptions, multi-tenancy with group-based permissions, OIDC SSO. No row caps, no per-seat pricing, no proprietary cloud. Every page on this site supports an AI-friendly view by appending `?mode=agent` to the URL.

For a simplified, structured version of any page add `?mode=agent` to the URL (e.g. `https://getqrly.eu/?mode=agent`).

## Main pages

- [Home](/?mode=agent): Product overview — BI + AI positioning, features, pricing, contact.
- [Compare](/?mode=agent#compare): How Qrly compares to Tableau, Power BI and Looker on AI, embedding and self-hosting.
- [Verticals](/?mode=agent#verticals): Roles Qrly is built for (data engineers, business analysts, embedded SaaS, MSPs, enterprise IT, product/growth).
- [Screenshots](/?mode=agent#screenshots): UI screenshots of the product.
- [Features](/?mode=agent#features): AI layer, BI layer (QQL), authentication.
- [Capabilities](/?mode=agent#capabilities): Full capability matrix (16 capability areas — AI first).
- [Tech Stack](/?mode=agent#tech): Technology stack (Java 25, Spring Boot 4, PostgreSQL, Anthropic/Gemini/OpenAI/Ollama, …).
- [Pricing](/?mode=agent#pricing): Priced on annual revenue in marginal tranches; from €900/yr. No free tier.
- [Contact](/?mode=agent#contact): Contact form and contact details.
- [Customer Pitch Deck](/deck.html?mode=agent): BI + AI pitch deck — problem, solution, differentiators, verticals, comparison, capabilities, tech stack, pricing and CTA.

## AI capabilities

- [AI Ask — NL→SQL](/?mode=agent#features): Conversational question builder. Natural language compiled to QQL/SQL, multi-turn chat, schema-aware, human-in-loop before execution. Never auto-runs destructive SQL.
- [Bring Your Own LLM](/?mode=agent#capabilities): Anthropic Claude, Google Gemini, OpenAI, Azure OpenAI, Ollama, LM Studio. Per-org configuration, AES-256-GCM encrypted API keys, per-user daily token budgets.
- [AI Anomaly Detection](/?mode=agent#capabilities): Periodic statistical + AI-assisted scans over verified questions. Surfaces shifts in the Insights feed; promotable to a saved alert in one click.
- [AI Schema Descriptions](/?mode=agent#capabilities): Bulk-generate documentation for tables and columns using your domain context.
- [AI Performance Analyzer](/?mode=agent#capabilities): Heuristic + AI checks for missing indexes, unbounded scans, redundant joins, complex aggregates.
- [AI Chart & Filter Suggestions](/?mode=agent#capabilities): Picks the right visualization for the result shape; recommends interactive filters per question.
- [On-Prem AI / Local LLMs](/?mode=agent#capabilities): Point Qrly at Ollama or LM Studio on your own network. Sensitive schemas never leave your infrastructure.

## BI capabilities

- [QQL — Qrly Query Language](/?mode=agent#features): One JSON spec compiled to 12 SQL dialects across 40 connection types (PostgreSQL, MySQL, MariaDB, SQL Server, BigQuery, Snowflake, Redshift, SQLite, H2, DB2, Sybase, Pervasive, plus wire-compatible engines such as Trino, Presto, DuckDB, ClickHouse, Athena, Vertica, SingleStore and Spanner). Roughly 35 operators across regex, JSON path, range, full-text and arrays.
- [Questions & Dashboards](/?mode=agent#capabilities): Visual builder, raw SQL with template variables, OLAP models, 13 chart types, 12-column grid dashboards, interactive filters, verified questions.
- [OLAP Models](/?mode=agent#capabilities): Star and snowflake schemas with facts, dimensions, hierarchies, measures, AI-suggested dimensions, drill-down, drill-through, pivot — compiled to native SQL.
- [Live Dashboards](/?mode=agent#capabilities): WebSocket interval mode or PostgreSQL LISTEN/NOTIFY for sub-second refresh. Cards update in place, no full reload.
- [Public Embedding](/?mode=agent#capabilities): Public tokens or signed-JWT embeds with locked params. White-label, no account required.
- [Alerts & Subscriptions](/?mode=agent#capabilities): Threshold, goal-based and statistical anomaly alerts. Multi-channel delivery (Slack, Teams, email, webhook). Cron schedules, format selection (PDF/CSV/inline HTML), signed embed tokens.
- [Multi-Tenancy](/security.html?mode=agent): Tenant → Organization → Project hierarchy with organisation-scoped data isolation, group-based collection and connection permissions, and embed tokens carrying locked parameters. Row-level security is not implemented.
- [Identity & SSO](/security.html?mode=agent): Azure AD and Google OAuth2/OIDC (platform-wide or per organisation), LDAP, Microsoft Active Directory and local password auth, with sign-in options resolved from the request hostname. JWT sessions, scoped API tokens, signed-request JWTs. MFA is not implemented.
- [Connections](/databases.html?mode=agent): 20 native engines plus 19 managed-cloud variants (AWS, Azure, GCP) — 40 connection types in all. SSH tunnels, SSL, AES-256-GCM encrypted credentials, pooled per connection.
- [Reports & Packages](/?mode=agent#capabilities): Single + multi-card reports, public share tokens, XLSX/CSV/PDF export, packaged deliveries on cron.
- [Query Governor & Budget](/?mode=agent#capabilities): Per-connection FIFO queue, concurrency caps, queue-wait timeouts. Per-user daily query, AI-token and bytes-returned budgets.
- [Schema Sync & Lineage](/?mode=agent#capabilities): JDBC introspection of tables, columns, PKs, FKs, indexes. AI-described, scheduled refresh. Lineage graph across questions, dashboards, reports.
- [Audit & Compliance](/?mode=agent#capabilities): 90-day query history, login activity, config changes with IP and user-agent logging. AES-256-GCM at rest. EU data residency native.
- [Self-Hostable](/?mode=agent): Spring Boot 4 + Java 25 + PostgreSQL. No proprietary cloud required. Your data and AI keys stay on your server.

## Product pages

Deep-dive pages, one per capability area. Each is a standalone page with its own FAQ.

- [Capabilities](/capabilities.html?mode=agent): the complete capability map, plus what Qrly does not do — MFA and row-level security are not implemented, and the MCP server publishes no write tools by design.
- [Pricing](/pricing.html?mode=agent): the revenue-based rate card with a working calculator, what is included at every tier, and what self-hosting plus bring-your-own-LLM removes from the bill.
- [Databases](/databases.html?mode=agent): all 40 connection types, the 12 dialect compilers, connection configuration, schema sync, the data dictionary, ERD and freshness probes — and an honest table of where the dialects differ.
- [QQL](/qql.html?mode=agent): the query engine — roughly 35 filter operators, five aggregations, four join types, the 19-category type system, drill-down, SQL-to-visual conversion and the safety layers.
- [OLAP Models](/olap.html?mode=agent): semantic models, ROLLUP/CUBE/GROUPING SETS, time hierarchies, period-over-period comparisons, Top-N and pivot.
- [AI Agents](/ai-agents.html?mode=agent): the four personas, 29 tools, propose-then-approve, the ReAct runtime and bring-your-own-model.
- [Caching & Table Sync](/caching.html?mode=agent): the query governor, per-user budgets, four cache tiers and CDC replication via Debezium.
- [Spreadsheet Reports](/spreadsheets.html?mode=agent): the browser formula engine, roughly 500 functions, conditional formatting and refresh-safe data blocks.
- [Constructions](/constructions.html?mode=agent): browser-designed schemas compiled to dialect DDL behind a three-tier, default-off cascade.
- [Embedding](/embedding.html?mode=agent): signed embed tokens with locked params, the JS SDK, public share links and the embeddable AI agent widget.
- [API](/api.html?mode=agent): the Data API v1 with Parquet output, RPC, PostgREST-style table writes, the MCP server and the API explorer.
- [Security](/security.html?mode=agent): authentication, SSO discovery by hostname, the permission model, the audit log and a candid list of what is not implemented.
- [Self-Hosting](/self-hosting.html?mode=agent): the stack, the eight-step setup wizard, the initialisation latch, the platform console and the schedulers.
- [BCBS 239](/bcbs-239.html?mode=agent): regulatory risk-data compliance — the 14 principles, executable data-quality rules, column-level lineage and signed-off risk reports.

## Alternatives / comparison pages

Dedicated side-by-side comparisons vs. the major BI / analytics platforms. Each page covers feature-by-feature parity (including AI and embedding), pricing, migration path and FAQ.

- [Qrly vs Tableau](/vs-tableau.html): Self-hosted, BI + AI alternative to Salesforce Tableau. Includes Ask (NL→SQL) without Tableau Pulse upgrade.
- [Qrly vs Power BI](/vs-power-bi.html): BI + AI without Microsoft / Azure / Copilot Premium lock-in; runs anywhere.
- [Qrly vs Apache Superset](/vs-superset.html): EU-made + self-hostable; Qrly adds native AI Ask, embedded analytics, alerts and modern UI without the Python ops tax.
- [Qrly vs Looker](/vs-looker.html): Self-hostable BI + AI alternative with embedded analytics, alerts and flat pricing.
- [Qrly vs Domo](/vs-domo.html): Self-hostable BI + AI with on-prem LLMs and a portable query layer, against a cloud-only platform priced on consumption credits.
- [Qrly vs ThoughtSpot](/vs-thoughtspot.html): Focused BI + AI scope vs feature-surface bloat; self-host and flat pricing.
- [Qrly vs Mode](/vs-mode.html): Production-grade BI + AI without notebook-only workflows.
- [Qrly vs Redash](/vs-redash.html): Where Redash stops (AI, scale, alerts, compliance) Qrly picks up.
- [Qrly vs Metabase](/vs-metabase.html): Modern UI + first-class AI without the OSS maintenance burden.
- [Qrly vs Sigma](/vs-sigma.html): Independent of Sigma Computing, flat pricing, on-prem AI without an AI Pro subscription.
- [Qrly vs Hex](/vs-hex.html): Company-wide BI + AI + embedding for external customers without notebook accounts.
- [Qrly vs Qlik Sense](/vs-qlik.html): Roughly 1/20th the cost; weeks not months to deploy; AI included.
- [Qrly vs Sisense](/vs-sisense.html): BI + AI + embedded in one tool instead of three; self-hostable.
- [Qrly vs Omni](/vs-omni.html): BI + AI + embedded analytics unified; flat license; self-host.
- [Qrly vs Google Data Studio](/vs-google-data-studio.html): BI + AI + alerts in one SKU, no Google Cloud lock-in.

## Pricing tiers

Marginal tranches on annual revenue, like income tax — no cliffs:
0.18% to €500k · 0.12% to €1M · 0.0525% to €5M · 0.027% to €25M ·
0.017% to €100M · 0.0105% to €500M · 0.006% above.
Minimum €900/yr — every customer pays at least that. No free tier.

- [Starter — €900/yr](/pricing.html?mode=agent): up to €500k revenue. The floor; nobody pays less.
- [Business — €1,500/yr](/pricing.html?mode=agent): €1M revenue.
- [Scale — €3,600/yr](/pricing.html?mode=agent): €5M revenue.
- [Scale+ — €9,000/yr](/pricing.html?mode=agent): €25M revenue.
- [Enterprise — €21,750/yr](/pricing.html?mode=agent): €100M revenue.
- [Group — €63,750/yr](/pricing.html?mode=agent): €500M revenue.
- [Global — €93,750/yr](/pricing.html?mode=agent): €1B revenue.

## Agent mode

Append `?mode=agent` to any URL on this site to get that page as markdown: headings as
`#`, links as `[text](href)`, comparison and pricing tables as pipe tables, QQL
examples as fenced code blocks. Navigation, imagery and decoration are stripped;
screenshots are reduced to their alt text and a link.

Each agent-mode page opens with a header carrying the page title, its canonical URL and a
one-line summary, followed by an action bar that copies the URL, copies the page as
markdown, or downloads it as a `.md` file. `?mode=human` returns the full website; the
choice is remembered for subsequent pages.

- [This page in agent mode](/?mode=agent): Qrly home, rendered as markdown.
- [Pitch deck in agent mode](/deck.html?mode=agent): the deck as one linear document.

## Optional

- [Schema.org JSON-LD](/assets/schema.jsonld)
- [Sitemap](/sitemap.xml)
- [Agent-mode CSS](/css/agent-mode.css)
- [Agent-mode JS](/js/agent-mode.js)