March 30, 2026
Your to-do is a spell—cast it well
Tickets Are Prompts
Write the goal, not 20 tiny chores—bots obey your words
TLDR: AI helpers follow tickets literally, so tiny, narrow tasks lead to fragmented, buggy work; write clear outcomes instead. Commenters split between “finally said,” “we’ve always done this,” and “the headline misleads—tickets aren’t prompts, but we should still write them like prompts.”
“Tickets are prompts” lit up the comment section like a fire alarm. The piece warns that AI “agents” (think tireless office robots) copy our worst habits: narrow tickets lead to narrow fixes, new bugs, and a swarm of micro-tasks. The author tried a classic, tidy ticket and watched it snowball into three more, burying the real goal. Moral: write the outcome in two lines and let bots find the steps—because bots don’t have hallway chats to fill in missing context.
The crowd split instantly. Pakvothe cheered that the ticket is the prompt—better writing equals better results. Chrysoprace sighed that good tickets are unicorns and engineers still have to translate fuzzy requests into reality. pjm331 highlighted the key move: give the biggest justifiable chunk, don’t pre-slice it. Then xg15 threw cold water, calling the headline misleading: tickets aren’t prompts…but since people treat them that way, we need to write them like prompts. Notpushkin rolled eyes and dropped a Shape Up link, arguing this is how it’s always worked: let the person doing the work break it down.
Cue the memes: “No hallways for robots,” “JIRA nesting dolls,” and “fragment-shaped work.” The real drama isn’t AI vs. humans—it’s whether we can write better instructions so the bots stop dutifully building the wrong thing faster.
Key Points
- •The article argues that tickets now act as prompts that shape AI agents’ scope and reasoning.
- •An experiment showed that narrowly scoped tickets led agents to fragment work, introduce new bugs, and obscure the intended outcome.
- •Repeated trials produced the same fragmented behavior, indicating a systemic pattern rather than a one-off issue.
- •Agents mirror human ticketing behaviors learned from historical data, compressing timeline inflation from sprints to minutes.
- •The author recommends outcome-focused tickets with broad scope and deferring subtasks until review; split only into smaller outcome-shaped initiatives if necessary.