Post what you need built. Claim what you can build. Publish what you finished.
This board is a working sample. The quests below are examples, and anything you post, comment on, or claim is kept in this browser rather than sent anywhere. To put a quest in front of someone else, open it and use Share, which copies a link with the quest inside it.
One analyst runs the same six sanity checks in Snowflake every Monday: row counts against last week, null rates on the join keys, late-arriving files. If they're on PTO, nobody runs them, and the first sign of a problem is a stakeholder asking why a number moved.
Prompt changes to the AI chat agent get tested in production because there is nowhere else to run them. Nobody wants to be the one who breaks a live conversation, so most ideas never get tried.
Every campaign starts from a blank page or an old deck someone half-remembers. Drafting on-brand assets for a single campaign takes a week of back-and-forth between Demand Gen and product marketing.
Customer Success Managers work out of a shared dashboard, then hunt for the matching Salesforce record and Drive folder in other tabs. Four or five clicks and a search box every time, per account, all day.
Sourcing passive candidates means collecting profiles 25 at a time, then stitching batches together by hand in a spreadsheet and deduping by eye.
Each AE decides every morning which of several hundred accounts to touch, from memory and whatever is loudest. Two reps looking at the same book pick differently.
A daily scorecard in Google Sheets tracks four signals per rep. The pipeline feeding it breaks most weeks, and whoever notices first fixes it by hand.
Partly working, usable now
Pricing outside policy and non-standard payment terms, with the clause quoted back so you can check it.
Every rep builds their QBR deck by hand from three tools, roughly six hours each, and the numbers don't reconcile between decks.
Report requests arrive as Teams messages, emails, and hallway asks. They get worked in the order they're remembered, duplicates get built twice, and nobody can answer how long the queue actually is.
After a demo, follow-up is a personal email with attachments. Nothing gathers the stakeholders who weren't on the call, and nobody can see what the buying group actually looked at.
When a member calls about a claim, the rep reads the repricing history off several screens and translates it live. New reps take the longest and are the least sure they got it right.
A prompt or prompt pack
A prompt pack plus the extract of the approved answer library it reads from. Cut a questionnaire from two days to about three hours on the last one we ran.
The dashboard count keeps going up and nothing ever gets retired. Nobody knows which ones are load-bearing and which are abandoned, so every one of them gets maintained forever.