Routines by Databox - An AI Analyst that runs analysis and reports on a schedule

Hey Product Hunt! 👋 I'm Pete from Databox. Excited to finally share Routines with you.

This is the next capability launch in our agentic analytics platform. The first one, our AI Analyst Genie, made analysis instant. Ask a question about your live data, and you get a finished answer in minutes. The second, Artifacts, turned those answers into polished, shareable reports. But both still begin the same way, with a person opening a chat and asking.

So we built Routines.

Routines let you put recurring analysis on a schedule. Give our AI Analyst a prompt and a cadence, and it runs the analysis and delivers the finished report on its own, every time it's due:

• Your recurring analysis runs itself: every run happens on schedule, without anyone remembering to start it. (It can also be triggered by a webhook from another system.)

• You can act from the notification alone: each run arrives by email, Slack, or in-app with a summary that tells you how things stand.

• Every answer stays available: runs are saved in the routine, so you can revisit any report or continue the conversation with a follow-up question

• Your own analysis becomes repeatable: save the way you analyze something as a Skill once, then put it on a schedule

All of that is what makes Routines a critical part of our agentic analytics platform.

Without agentic analytics, analysis must be done by a person. Routines delivers it to people instead.

In the old way, someone pulls the numbers, figures out what moved and why, writes it up, and sends it around. The team talks about it a few days later. The next round happens when that skilled person has time for it again, usually a week or a month out. So analysis happens as often as one person's calendar allows, only the people on that email ever see it, and anything that goes wrong in between doesn't get caught.

With analysis done on a schedule using Routines, a Tuesday problem gets flagged on Tuesday and the plan changes that week instead of next month.

And anyone can be on the delivery list, including people one function over, so the teams that support each other work from the same explanation at the same time. For example, with the right Routine setup, marketing can see that sales is converting leads at a lower rate than last month and can dig into lead quality or push volume up before anyone raises it in a meeting.

We've been testing it internally for weeks. Our team members are more informed, and we spend meetings deciding what to do instead of reviewing what happened.

Routines is live today. Try it out on your account, or start a trial: https://help.databox.com/start-a-trial

If there's a report you'd like to fully automate, tell us about it below. We'll set it up for the first five people who ask, no charge for our time.

Skills + Routines are my new best friends. As a product person, I have to do a lot of analysis and a Skill helps me define how I want the analysis done and with a Routine I can make it into a daily workflow in a single step. No more clicking through another round of setup.

Big congrats to the team 💪

The biggest pain killer for me was that every run pulls fresh from your live connected data and not a cached copy of the last answer. That way i can truly trust the "Nothing changed today" result and not just assume that it is a static placeholder.

Pairing this with custom webhooks means we can tie our automated reporting directly into our existing workflows. congrats @zigapotocnik

@zigapotocnik The set it and forget it part is cool, everyone has a dashboard, but someone still has to remember to check and figure out what changed... How do you decide which insight are actually worth ?

I'd like to point to 2 details:
- this allows you to turn any existing chat or artifact into a Routine with one click, which means no rewriting instructions or prompts
- every Routine run is saved and searchable, so digging up 'what did last Tuesday's check say' doesn't mean scrolling back through chat history
It's small things like these two that make this feature super useful.

I really like is how easy creating a routine is.

You can ask Genie in plain English (or your own language) and it works out the instructions for you. Or you can go to an existing chat or Artifact and turn it into a routine with a single click.

Either way, you end up in the same place: taking something you already wanted to do and making it run on its own on a fixed schedule. No rewriting and rebuilding prompts, no setting up everything up from scratch every time you wand an analysis report or just a simple answer to a recurring question.

One comparison I think is cool here is Databox Routine vs cron job. A cron job just runs a script on a timer, doesn't do any reasoning and just gives you the same type of analysis every time. A Routine decides each time what is worth focusing on and telling you. It's a different analysis every time. Sounds like a small thing, but is why I trust it to run analysis autonomously instead of me doing it manually myself.

I set up my routine for paid ads instead of building the weekly monitoring report by hand every Monday or manually checking my dashboards every day. Ask once, 'summarize performance every day at 9am and it's already in Slack when I need it. Built from the same connected data as my dashboard, so there's nothing to double check before sending it along. Literally took me a few mins and I just pointed ai analyst to the right dashboard.

Hey hunters!

Every company has analysis that someone redoes by hand every week. The Monday pipeline check. The weekly channel report. The monthly board deck. It isn't hard work; it's just work nobody has time for, so it's the first thing that slips when the week gets busy.

Routines fix that. You give your AI Analyst a prompt and a schedule. It runs against live data, prepares the analysis or report you asked for, and delivers it to Slack, email, or in-app. Not a dashboard refresh. An actual written answer, on time, every time.

This is the fourth piece of the same puzzle in the last few months. MCP brought your data into the LLM tools you already work in. Skills gave the analyst repeatable expertise. Artifacts made the output build in seconds and shareable. Routines make it run without you having to ask.

That's the bet we're making: analytics stops being a tool you operate and becomes an agent working in the background, on top of a semantic data layer that actually understands your business.

Try it out; it's free for 14 days - https://databox.com

Would love your feedback.