Delv
Data Analystby Hex★ 4.3

Hex Magic

Hex's AI assistant for data notebooks. Generates SQL, Python, and chart code; pairs perfectly with Hex's collaborative notebook UX.

Try it
B
Safety & Trust

Delv Safety Grade: B

Score 72/100 · assessed 2026-04-18

Maintainer85
Permissions65
Supply chain60
Transparency50
Incidents100

Hex Magic is a proprietary AI assistant embedded in Hex's commercial notebook platform. Hex Technologies is a well-funded, established data-collaboration vendor with enterprise customers and reasonable operational maturity. The agent generates and executes SQL and Python code within the notebook environment, which means it has database read/write access and can run arbitrary Python. The closed-source nature and lack of public repository limit transparency into its supply chain and security practices. No public incidents are known. The main risk is the breadth of permissions: it touches your data warehouse credentials, runs code in your environment, and operates within a proprietary SaaS boundary. For teams already committed to Hex's platform, the risk profile is acceptable, but you're trusting Hex's internal security posture without external verification. The tight integration with Hex's notebook UX is a green flag for usability but a red flag for vendor lock-in and auditability.

Green flags

  • Hex is a well-funded, established vendor with enterprise customers
  • Operates within Hex's existing security boundary (SOC 2, GDPR compliant)
  • No known security incidents or credential leaks
  • Iterative refinement loop catches errors without manual intervention
  • Scoped to notebook cells; doesn't access filesystem outside Hex environment

Red flags

  • Closed-source with no public repo or security audit trail
  • Executes arbitrary Python and SQL with full notebook environment access
  • Requires data warehouse credentials; scope depends on user's DB permissions
  • Proprietary SaaS; no self-hosted option for air-gapped environments
  • Vendor lock-in: agent only works within Hex's paid platform

Permissions requested

DB readDB writeShell executeOutbound networkAccess secretsExternal LLM call
Assessed by Delv Editorial using public metadata. Grades are advisory and update as the ecosystem changes. They do not replace your own review of permissions and code before granting an agent access to sensitive systems.

Pricing

PAID

Platforms

web

Review

Hex Magic sits inside Hex's collaborative notebook environment, which already puts it ahead of most data-analysis agents. You type a prompt in a cell, and it generates SQL queries, Python transformations, or chart configs without leaving the notebook. The autonomy here is narrow but practical: it writes code, runs it, checks the output, and refines if something breaks. I've watched it catch a column-name typo in a JOIN and fix it in the next iteration without me lifting a finger. Where it shines is the handoff between technical and non-technical teammates. A PM can ask for a cohort retention chart in plain English, Magic drafts the SQL and visualisation, and a data analyst can review or tweak the code inline. That's the workflow Hex built for, and Magic slots in cleanly. The generated code is readable, not a black box, which matters when you need to explain a metric to finance or debug a spike three months later. Failure modes are predictable. Complex joins across schemas it hasn't seen before produce plausible-looking SQL that returns empty sets. It doesn't know your business logic, so it'll happily calculate churn using the wrong date column if you're vague. The chart suggestions lean safe: bar charts and line graphs, rarely anything adventurous. If you need a Sankey diagram or a custom D3 visual, you're writing that yourself. Compared to GitHub Copilot in a Jupyter notebook, Magic has more context about your data schema and can execute code autonomously. Compared to something like Noteable's AI, Hex Magic benefits from tighter integration with Hex's version control and publishing tools. But it's locked into Hex's ecosystem. If your team lives in Databricks notebooks or Mode, this doesn't help you. The pricing is bundled into Hex's team plans, so you're paying for the whole platform, not just the AI. That's fine if you already need a collaborative notebook tool. If you're solo or your team is happy in Jupyter, the cost doesn't pencil.
Verdict

Pay for this if your team already collaborates in notebooks and you're tired of Slack threads asking for 'just a quick chart'. Skip it if you work solo, need deep customisation, or your data stack lives elsewhere.

Good at

  • Generates and iterates on SQL, Python, and charts autonomously within the notebook
  • Readable, reviewable code instead of opaque outputs
  • Tight integration with Hex's schema awareness and collaboration features
  • Handles the PM-to-analyst handoff better than most tools
  • Refines queries when it hits errors, saving manual debugging loops

Watch out

  • Locked into Hex's platform, no standalone option
  • Struggles with complex multi-schema joins or unfamiliar data models
  • Chart suggestions are conservative, rarely adventurous
  • Pricing bundled with Hex teams, expensive if you only want the AI
  • No awareness of business logic or domain-specific metrics

Use cases

  • Faster notebook authoring
  • Self-service analytics for PMs
  • Cross-team data collaboration
  • Auto-explaining a chart for slides