Artifacts by Databox - Ask your AI Analyst and get back a ready-to-share report
by•
Turn any conversation with your AI Analyst into a polished report, slide deck, or interactive document built from your live data. Generate it from a single prompt, then share it via public link or download as a PDF.
Replies
Databox
One implementation detail I'm particularly happy with is how Artifacts evolve through a conversation.
When Genie generates a report, you don't have to start over if you want something different.
Need another section? Just ask.
Want a different chart? Ask.
Need to adjust the branding for a client? Ask.
Those kinds of changes happen directly on the existing artifact, so you're refining what you've already built instead of starting from scratch.
For bigger transformations, like turning a report into a slide deck, Genie creates a new artifact while keeping the original report intact. Both versions are available to share, and because they're built from the same connected data and conversation, you don't lose the context that got you there.
We didn't want AI to feel like a series of disconnected prompts. We wanted it to feel like working on a living document that evolves with you.
congrats on the launch! is there a limit to how large the data sets can be/upfront usage? for example if I'm migrating from another tool could I upload my larve pre-existing data sets?
Databox
@ethan_cheng Thanks Ethan. For data already connected to Databox, there's almost no dataset size limit that would block this, it pulls from your existing metrics. For migrating from another tool, we have an Ingestion API where you can push historical data in from other tools or even AI clients like Claude, though I don't have a specific size ceiling for that path in front of me. If you're planning a larger migration, happy to have someone follow up with specifics.
Databox
@dhirajwohra Thanks Dhiraj. That part's really down to what you ask Genie for rather than a separate toggle, since Genie is doing the analysis before it builds the artifact. So asking something like "flag any outliers in this month's data" as part of the prompt should get you that layer of analysis baked into the report, not just the raw numbers charted out.
Databox
Worth flagging for anyone who tried an earlier version of this: you can now edit text and titles directly inside the artifact panel, no prompt needed for small fixes. And every artifact you make is saved to its own gallery with search and favorites, so finding last month's report doesn't mean digging through chat history anymore.
Don't get me wrong i like the idea, but what's the biggest difference between Databox and Cloude with proper skills?
Databox
@kamil_infeld Fair question. With Claude and skills, you're still the one pulling the data together first, exporting from wherever it lives and feeding it in. The core difference here is Genie already has your data connected, 130+ integrations, so there's no export or paste step at all. You ask, and it's already working from your actual numbers.
@zigapotocnik Make sense, and actually add value
Being able to ask for the same analysis restyled for a different client's branding without rebuilding it is the part I would use most. Pulling from live data at generation time also avoids the usual stale screenshot problem. When a report goes out as a public link, does it refresh against live data on every view or is it frozen at the moment you generated it?
Databox
@adamkamaneh Good question, and the honest answer: it's frozen at the moment you generate it. A public link shows a snapshot, so it won't refresh on its own if someone views it later or the underlying numbers change. If you want fresh numbers, you'd re-trigger the report and share the new link. Live-refreshing links are on our roadmap, just not there yet.
This looks useful for teams that need to share reports quickly. Can you also trace each number back to the original data and filters when someone asks about it?
Databox
@liana_preston Good question, and I want to be careful not to overstate this one. Genie pulls real numbers from your connected sources into the artifact, so what you see matches your dashboards, but there isn't a built-in "trace this number back to the filters/query" feature inside the artifact itself today. If you need to verify a specific figure, you'd go check the underlying metric or databoard directly. Let me know if that's the kind of drill-down you're picturing, that'd be useful feedback for the roadmap.
the edit-mode-vs-prompt-mode split is a smart way to avoid the "AI overwrote my wording" problem. question on that: once I've hand-edited a few sections and then ask Genie for a structural change later, is there version history I can roll back to if the new prompt mangles something near my edits, or is the artifact a single mutable document with no undo path back to an earlier state
Databox
@galdayan Good question, and I want to answer this carefully since it matters. Genie is made aware of your edits before it makes a structural change, so it builds on your version rather than blindly overwriting it. What I'm not certain about is whether there's a full version history/rollback you can access if something still gets mangled near your edit, versus the artifact being a single mutable document. Don't want to guess on that one, let me confirm and get back to you rather than answer wrong on something this specific.
Turning an analyst request into a shareable report could remove a lot of reporting friction. Can readers trace every conclusion back to the underlying metric, source system, and transformation used by the AI analyst?
Databox
@russlan_ramdowar Good question, and want to be precise here since it's an important distinction. Genie pulls real numbers from your connected sources into the report, so what you see matches your dashboards, not a hallucinated figure. That said, there isn't a built-in trace-back panel inside the artifact itself showing the exact transformation or source system per number today, you'd check that in the underlying metric or databoard directly. Good feedback for the roadmap though, appreciate you raising it.
honestly this looks pretty solid, especially for teams that don't want to deal with a clunky BI setup. one thing i'd love to see is a built-in anomaly alert system that flags weird spikes or drops in your custom metrics straight to slack, basically instead of having to dig through dashboards yourself. would save a ton of time.
Databox
@jimmorrison Thanks Jim. Good news on part of this, Genie already does anomaly detection, so you can ask for a report that flags spikes or dips in your metrics today. The proactive piece, alerts pushed to Slack automatically without you asking, isn't there yet as far as I know, that's more of a "check when you ask" flow right now rather than a standing alert system. Good idea to flag for the roadmap though.