Hex

What it does

Hex is a collaborative data workspace built around notebooks that combine SQL, Python, R, and no-code cells, with an AI agent layered throughout to help generate queries, code, and charts. It’s aimed at data analysts and data scientists who already work in a technical stack and want AI to speed up that existing workflow rather than replace it.

What sets Hex apart from a plain notebook tool is its collaboration and publishing model: notebooks can be shared, versioned, and commented on by a team, then published directly as interactive apps or dashboards for non-technical stakeholders to use. That combination — technical depth for the people building the analysis, plus a clean consumption layer for everyone else — is Hex’s core differentiation from both conversational tools like Julius AI and enterprise BI platforms like ThoughtSpot.

Ideal users

• Data analysts and data scientists who already work in SQL and Python and want AI assistance layered into that workflow
• Data teams that need to collaborate on notebooks and publish results as shareable internal apps or dashboards
• Organizations that want a single workspace connecting multiple data sources for both ad hoc analysis and reusable reporting

Who should avoid it

• You or your team don't write SQL or Python — a conversational tool like Julius AI or a no-code dashboard like Polymer Search is a better fit
• You need a governed, natural-language search interface for non-technical business users at scale — ThoughtSpot is built for that
• You need standing predictive models rather than exploratory analysis and reporting — better served by DataRobot or Pecan AI

Key features

• Notebook interface supporting SQL, Python, R, and no-code cells in one workspace
• Notebook Agent that can generate SQL, Python, charts, and full analyses from natural-language prompts
• Threads agent and semantic model agent for AI-assisted collaboration on Team/Enterprise tiers
• Publish notebooks as interactive apps and dashboards for non-technical stakeholders
• Connects to a wide range of databases and warehouses; scalable pay-as-you-go compute for larger workloads

Pricing

Free — $36/editor/month (Professional plan)

Confirmed on hex.tech/pricing: Community (free, up to 5 notebooks, small compute, Notebook Agent trial), Professional ($36/editor/month, unlimited notebooks, up to 5 published apps, medium compute), Team ($75/editor/month, unlimited published apps, Threads and semantic model agents, medium compute included), and Enterprise (custom, SSO, HIPAA add-on, single-tenant option). Pay-as-you-go compute beyond the included tier is billed per minute, ranging from $0.32/hour (Large) up to $4.06/hour (A10G GPU).

Typical workflows

• A data analyst starts a notebook by prompting the Notebook Agent to pull and chart weekly active users from the warehouse, then refines the underlying SQL by hand before publishing it as a live app for the product team.
• A data science team collaborates on a shared notebook investigating a metric anomaly, using version history and comments to track hypotheses before publishing the final findings as an internal report.

Integrations

• Snowflake, BigQuery, Redshift, Databricks, and other major warehouses
• dbt
• Slack (for scheduled alerts on Team/Enterprise tiers)

Privacy & security notes

Hex's LLM providers (OpenAI and Anthropic) operate under zero-data-retention agreements by default; Workspace Admins can opt in to allow retention for specific models that require it for safety monitoring, per Hex's trust documentation. Separately, whether Hex itself uses AI-session data to improve its own features depends on plan terms, and this can be turned off in Settings > AI & agents. Beyond the Enterprise single-tenant add-on, Hex has not published additional data-residency options.