Bilbo - Metabase AI Assistant : NL Queries, Visualize Data

What Is Bilbo — The Metabase AI Assistant?

Bilbo is a purpose-built artificial intelligence layer for Metabase—designed not as a standalone dashboard tool, but as an intelligent co-pilot embedded directly into your existing analytics workflow. It interprets everyday language—no jargon, no syntax—and converts user intent into precise database interactions: from ad hoc exploration and metric retrieval to dynamic chart generation. Unlike generic AI chatbots, Bilbo operates *within* your Metabase environment, respecting its permissions model, schema definitions, and cached metadata. The result? Analysts, marketers, product managers, and executives alike can ask questions like “Which region had the highest churn last quarter?” and receive both a rigorously executed query and an insightful visual—without writing code, waiting for engineering, or deciphering table joins.

Getting Started with Bilbo — Simple, Secure, Instant

Adopting Bilbo takes under two minutes. Install the lightweight Chrome extension, authenticate with your Metabase instance (cloud or on-prem), and begin typing—no configuration, no schema mapping, no training required. Because Bilbo connects natively to Metabase’s API and metadata layer, it instantly recognizes your tables, fields, calculated columns, and even saved questions. Ask “Compare Q3 2023 vs Q3 2024 revenue by product category”—and Bilbo identifies time filters, aggregates, dimensions, and relationships autonomously. Results appear in-context: clean tables, interactive charts, and export-ready visuals—all generated in real time, right inside your browser tab.

Go beyond one-off queries with Bilbo’s collaborative intelligence layer. Define team-wide context—such as standard fiscal calendar definitions, KPI naming conventions, or preferred date ranges—and pin them to shared workspaces. These contexts persist across sessions and users, turning tribal knowledge into reusable, versioned logic. Whether you’re scaling analytics across departments or onboarding new hires, Bilbo ensures everyone speaks the same data language—consistently, accurately, and securely.

Core Capabilities That Set Bilbo Apart

• Intent-Aware Natural Language Engine: Bilbo doesn’t just parse keywords—it models user intent using fine-tuned language understanding trained specifically on analytical dialogue. It distinguishes between “top customers by spend” (ranking) and “customers who spent over $5K” (filtering), and handles ambiguity gracefully—e.g., interpreting “last month” relative to your organization’s fiscal period, not just calendar logic.
• Schema-Native Query Generation: Rather than guessing at joins or aggregations, Bilbo reads Metabase’s internal schema representation—including foreign key hints, field descriptions, and semantic type annotations—to construct syntactically sound, performant SQL. It avoids common pitfalls like double-counting in many-to-many relationships or misapplying window functions—because it reasons about your data architecture, not just surface-level column names.
• Visual Intent Translation: Describe how you want to see the answer—not just what you want to know. Phrases like “show this as a stacked bar chart,” “plot weekly retention as a line with confidence bands,” or “compare metrics side-by-side in a table” trigger automatic visualization selection, axis labeling, color mapping, and responsive formatting—no manual chart builder needed.
• Context-Aware Collaboration: Share more than results—share *meaning*. Save named contexts that bundle definitions (e.g., “Active User = logged in ≥3x/week”), default filters (“exclude test accounts”), and visualization presets. Team members inherit these rules, ensuring reports align with business logic—not individual interpretation.
• Zero-Latency Browser Integration: Bilbo lives where your work happens—in the Metabase UI, via the Chrome extension. No tab-switching, no copy-pasting, no exporting raw JSON. Your query, your result, your chart—all rendered inline, preserving Metabase’s native look, feel, and access controls.

These capabilities converge to eliminate three persistent bottlenecks: the lag between question and insight, the friction between technical and non-technical roles, and the inconsistency born from decentralized reporting practices. With Bilbo, query turnaround drops from hours to seconds—and data literacy rises across the org, not just within the BI team.

Why Teams Choose Bilbo Over Generic AI Analytics Tools

Many tools promise natural language querying—but few deliver accuracy *and* trust *within* your existing stack. Bilbo succeeds because it’s built *for* Metabase, not bolted onto it. It inherits your authentication, respects row-level security, honors saved questions and dashboards, and surfaces only data your role is permitted to view. There’s no shadow database, no data replication, no external LLM hallucination risk—just deterministic, auditable, schema-grounded execution.

This architectural fidelity enables real-world impact: Sales ops teams generate territory health reports before morning standups. Customer success leads spot emerging support trends during live calls. Finance validates accruals mid-month without pulling engineering into a sprint. And leadership explores cross-functional metrics—like “LTV:CAC by acquisition channel and cohort”—independently, confidently, and correctly. Bilbo doesn’t replace analysts—it amplifies them, freeing capacity for strategic modeling instead of repetitive query writing.

Real-World Applications Across Functions

Frequently Asked Questions

Where does my data go when I use Bilbo?

Nowhere. Bilbo never stores, logs, or transmits your raw database contents. All query processing occurs locally in-browser or via secure, ephemeral API calls to your Metabase instance. Only the natural language prompt and resulting SQL (which Metabase itself executes) traverse the network—and both are subject to your existing TLS encryption and RBAC policies. Your data remains governed entirely by your Metabase deployment’s security posture.

Is Bilbo compatible with self-hosted Metabase environments?

Yes—fully and natively. Bilbo supports Metabase v0.45+ on any infrastructure: AWS EC2, Kubernetes clusters, Azure VMs, or private data centers. Since it authenticates via Metabase’s official session or API token system and communicates exclusively through its documented REST endpoints, no firewall exceptions, reverse proxies, or custom middleware are required. Compliance-sensitive organizations retain full control over data residency, audit trails, and infrastructure hardening.

What visualization formats does Bilbo support?

Bilbo leverages Metabase’s full native visualization suite—including bar, column, line, area, pie, donut, scatter, heatmap, and pivot tables—as well as custom visualizations built with Metabase’s SDK. You can request specific types (“show as a funnel chart”), refine appearance (“add trendline,” “sort descending,” “hide legend”), or let Bilbo auto-select the most statistically appropriate format based on data cardinality, measure types, and dimensionality—ensuring clarity without compromise.

How does shared context improve team-wide analytics?

Shared context transforms Bilbo from a personal productivity tool into an organizational knowledge engine. When your finance team defines “Net Revenue Retention” with precise filters (e.g., excluding one-time professional services), that definition becomes discoverable and reusable across sales, marketing, and leadership queries. Contexts can be versioned, commented on, and permissioned—so only relevant stakeholders see sensitive metrics. Over time, this builds a living glossary of business logic, reducing misalignment, accelerating audits, and making analytics truly scalable.