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
Hi Product Hunt.
I’m Pete from Databox. Today we’re shipping Artifacts inside Genie, our AI Analyst. An artifact is a report Genie generates from your connected data. It combines the charts you’d normally build by hand with the analysis of what actually drove the numbers, in one shareable document.
Reporting used to work like this... You’d build a bunch of dashboards. You’d compare performance to the previous period and to goal, try to understand how the work you did turned into the result you got. Then, you'd write out what happened and what you planned to do next.
Whether you hit the goal, missed it, or crushed it, the manual process was the same.
With Artifacts, none of that is necessary. It’s a one-time setup. You connect your data and define your metrics once. After that, you tell Genie what you did, and it compares performance to the previous period, checks whether the work correlated with what you achieved, digs into why through statistical analysis, and recommends what to do next. Then it builds the report.
A “what happened last month” question becomes a finished report in minutes.
A few very specific things it can do:
Turn the same report into a slide deck. One follow-up request, nothing rebuilt.
Restyle a report to your own brand or a client’s, mid-conversation.
If you're a professional services firm, build a cross-client report spanning every account, not one client at a time.
Ship as a public link, with no login for whoever you send it to, or a PDF. Turn the link off whenever you want.
You might be asking, why not just use Claude or ChatGPT? Because Databox already has your data connected, understands how to calculate your metrics, and knows how to run statistical analysis on them.
Three things the general chat tools don’t have or don't do consistently. Ask one of them for a report and you feed it the data yourself, every time, and hope it calculates and analyzes correctly. And when it gets something wrong, you’re stuck.
Databox artifacts are editable, so you add your own context and Genie refines it. Fine tuning, not redoing.
Less time spent reporting means more time doing the work and more time thinking about how to do it better.
If you’re not on Databox yet, I challenge you to automate your next report in three steps.
1. Start a trial,
2. connect your data,
3. and ask Genie to “create a report analyzing what happened last month and why.”
No coding, no spreadsheet-wrangling required.
PicWish
@pc4media how do cross client reports handle missing metrics? does genie partially render the artifact or flag the missing source before building the pdf?
@pc4media Half my week used to be someone rebuilding the same report because a stakeholder wanted one metric framed differently. My worry with an AI analyst is trust: when it flags a dip, can I click through to the exact query and rows behind it? That "where did this number come from" moment is what breaks these for execs.
This is the part of the job I've spent years fixing manually.
Reporting systems that actually get used are rare, most of them just create more work downstream. Being able to ask for a report and get back something client-ready, built on the live data, is the direction this space needed to go. Excited to see how agencies and marketing teams use this to stop rebuilding the same deck every month.
Databox
@lexi_morgan That's exactly the pattern we kept seeing too, reporting systems that generate more work than they save. The bar we held ourselves to was: if it doesn't end in something you'd actually hand to a client without touching it first, it doesn't count. Would love to hear how it holds up once you try it on a real client report.
Marketing analytics person here — the biggest benefit of this is not being the human BI layer anymore. You can ask for a report and get something narrative-ready back. And it's based on verified data (something that really matters but many marketers forget about). Super useful!
Databox
@carmenstrelec The "human BI layer" line is a good way to put it, that's the exact job we were trying to take off people's plates. And you're right that the data part matters more than it sounds, since a narrative-ready report is only useful if the numbers under it are the ones your team already trusts.
Databox
@etiennegarcia That's the split we kept coming back to as well, getting the answer is only step one, turning it into something you can actually hand off is the part that eats the time. Thanks for the kind words, glad it landed.
Databox
The hardest question we had to answer while building Artifacts wasn't technical, it was about control: when AI writes your report, what still belongs to you? Our answer: Genie builds the document from your live data (with a real design system behind it and a deterministic engine doing the math, so numbers are computed, never guessed), and you keep the last word, literally. Edit any text directly, and Genie knows about your edits, so its next update builds on your version instead of overwriting it. Would love your take on whether we drew that line in the right place. I'll be around all day for questions.
Databox
@grega_cej That control question came up in almost every internal review too. The edit mode is the answer we kept landing on: type mode for the words, prompt mode for anything structural like charts or layout. Small thing, but it means you're never stuck rewriting a whole report just to fix one sentence.
Databox
Every team I talk to has some version of the same Friday afternoon task: turn the week's numbers into something presentable and pass it around. Artifacts replaces that with a prompt. You ask Genie for the report, get back a finished document, and either share the link or download the PDF. You skip the whole assemble-and-format slog in the middle.
Databox
@thomas_bossee1 Exactly the moment we built this for. That Friday afternoon "turn the numbers into something presentable" task shouldn't take longer than the analysis itself. Ask Genie, get the doc, share the link. That's the whole workflow now.
Databox
The part worth explaining is how the artifact actually gets built. It's not assembled from a template or a databoard, Genie authors the layout, structure, and visuals fresh for each request, following a shared design spec so it comes out on brand by default. The data inside it is pulled live from your connected metrics at generation time, so what you get is a real document backed by your real sources, not a paraphrase of a chart you pasted in.
Databox
@jakobzmrzlikar Exactly this. And the part I'd add: the model doesn't do the math either. Every number in the artifact is computed by a deterministic query engine before the model writes a word around it. Funny how the least glamorous piece of the stack, essentially a calculator, made the biggest difference in how accurate the numbers are. Authored fresh, never improvised.
Databox
One thing I'd add from the marketing perspective: the most useful thing about Artifacts is that it closes the "I have an answer" to "I have something shareable" gap.
For example, ask the AI Analyst to build you a report or deck, and it does so using your real data. Then you can hand it off via public link or PDF and there's no login needed on the other end.
I know this is HUGE for anyone who's ever had to rebuild the same client report by hand every month, and it's a great way to add a supporting doc to your quick answer. Happy to answer questions if useful!
I tested this by asking Genie for a report, then asking it to turn that same report into a slide deck. Same data, same analysis, different format, one follow-up prompt. That's the kind of thing that used to mean rebuilding the whole thing in a different tool. Here it just carried over.
Databox
@tadej_kelc That follow-up prompt trick is honestly my favorite part too. No re-uploading data, no rebuilding charts in a new tool, just "turn this into slides" and it carries the whole analysis over. Glad it held up when you tried it yourself.
For teams migrating from existing BI tools, how seamless is the process of importing historical datasets and custom metrics, and does Databox offer any AI assisted mapping to speed up that transition?
Databox
@aymi_malik Good question, though worth separating two things here. Artifacts itself pulls from data you've already connected to Databox (130+ native integrations), so no import step for that part. For historical datasets and custom metrics, we also have an Ingestion API, so you can push data from any other tool, or even directly from an AI client like Claude, straight into Databox. That covers a lot of migration cases without needing a native integration to exist first. I don't want to overstate the mapping side though, there's no AI-assisted mapping for that step yet.
@zigapotocnik Thanks so much for the detailed breakdown, that's really helpful context! Great to know that Artifacts pulls directly from the 130+ native integrations without needing an import step, and the Ingestion API sounds like a smart, flexible fallback for historical data or tools without native support yet.
Appreciate the transparency on the AI-assisted mapping front too, it's a helpful roadmap item to keep in mind. Excited to see how this evolves. Congrats on the launch!
Databox
@aymi_malik Anytime. Thank you Muhammad!