Sundial vs. Hex
Everyone is shipping an analytics agent. The hard part was never the agent, it's making a business user trust the answer without a data scientist sitting next to them.
Hex has built one of the best notebooks in the market. If you write SQL and Python for a living, it's a joy. But most people asking questions of data and taking business decisions basis that data, don't want to write code, are often unable to audit a query by themselves, and have no way to tell a right answer from a confident-wrong one. For them, "we put an agent on the notebook" isn't the same as "we can rely on what it says."
Pick Hex if your primary users are data scientists who want a powerful, agentic workspace to build in.
Pick Sundial if you want business users to self-serve answers they can actually trust — while your data scientists and data engineers get all the knobs they'd want to run their own analysis.
How do you choose?
Both run on your warehouse. Both have an agent, a context layer, and a chat interface. In a demo they look alike. The difference is who each one is really for.
Hex is a notebook first. Its center of gravity is the technical user - take your analyst or data scientist who can read the code the agent writes, spot when it's wrong, and fix it. That's a real and valuable audience. But it means the people Hex serves best are the people who probably need it less than this other persona. The business user, the one who can't check the work, gets the same agent without the ability to verify it.
Sundial is built for the opposite starting point: the person who has the question but not the SQL, and who needs to know whether the answer is solid before they act on it. That's a different product, not a different skin.
The question isn't "does it have an agent?" It's "can someone who can't read the query trust what it says?"
What is Hex?
Hex is a warehouse-native notebook with a strong agentic layer on top — SQL, Python, R, data apps, and a genuinely capable analysis agent. For data scientists, it's excellent: fast, flexible, and built for people who want to go off-road and audit every step.
Two things follow from that design. First, it's a builder's tool — the intelligence is there, but the analytical judgment and the guardrails are yours to assemble. Second, and more importantly, it hands the actual tedious work back to the user. Context curation — the semantic models, the endorsements, the workspace rules that make answers trustworthy — is left almost entirely to your team to set up and maintain.
There's no day-one point of view; there's a workspace waiting for you to make it smart.
And for a business user, the thing that matters most is missing: verifiability. Hex can show its reasoning, but it doesn't tell a non-technical user how much to trust a given answer. Without that, self-serve quietly breaks — people either over-trust a wrong number or stop using the tool.
What is Sundial?
Sundial is an opinionated AI analyst built so business users can self-serve answers they can rely on. We've been insights-first for five years, pre-dating the LLM era, and spent that time encoding analytical frameworks — Playbooks — so anyone can analyze the way an expert would, without writing or reading a line of SQL.
Two things make that work.
Verifiability: every answer carries a confidence signal — guaranteed vs. directional — plus a transparent view of which data it drew on, so a non-technical user knows when to act and when to ask. That's the piece that makes self-serve real rather than aspirational.
And a smarter split of the work. We agree context is what makes an agent trustworthy — but we don't think the whole burden should land on your team. Your users hold context no tool can derive: what your business actually means by "active," which segments matter, the tribal knowledge in people's heads. That's the context worth their time. A large part of the rest — connecting sources, mapping semantics, standing up the baseline — can be automated, and Sundial does it, with a team of former data leaders plus agents setting it up with you in the first 30 days. You put in what's uniquely yours; we handle the plumbing.
Btw, Sundial has the open-ended exploration layer too. It can fire virtually any query against your warehouse; the semantic layer, Playbooks, and trust system vet the work, so you always start from something verified.
Feature comparison
| Capability | Sundial | Hex |
|---|---|---|
| Setup & time-to-value | ||
| Useful on day one, set up for you~30 days, by experts + agents | Available | Not offeredneeds someone to set up context first |
| Pre-built analytical frameworksRCA, retention, anomalies | Availableexpert Playbooks | Not offeredframeworks come from the model |
| The intelligence | ||
| Has a point of view on every questionPlaybooks + opinionated defaults | Available | Not offeredopen notebook |
| Proactive insightssurfaces what you didn't think to ask | Available | Available with caveatsrule/threshold alerts you configure |
| Context Engineorg, connectors, semantics | Availableowned & curated | Available |
| Real analysis, not just metric lookupswhy did retention drop? | Available | Available with caveatscapable, but no Playbooks to standardize it |
| Trust & rigor | ||
| Confidence signal on each answerguaranteed vs. directional | Available | Not offered |
| Observability + evals so rigor holds | Available | Available with caveatsvia the Review Agent |
| Self-serve & build surface | ||
| Dashboards + data apps on your warehouse | Availableopinionated, ready-made | AvailableDIY build |
| Spreadsheet UX + Python notebooks | Available with caveatsSQL / Python editability offered | AvailableSQL + Python + R |
| Foundation | ||
| Meet you where you workSlack, Teams, MCP | Availablenative chat | Available |
| Self-hostable / private cloud · BYO LLM | Available | Not offeredSaaS, model chosen by Hex |
In closing, anyone can put an agent on a notebook. Making a business user trust the answer is the hard part — and it's the part Sundial was built for. Teams like OpenAI and Gamma already run their analysis this way — see Sundial answer a real question on real-looking data, or try Sundial for free.