Why a Daily AI Briefing Worked Where Calendar Reminders Didn't
- Kathleen Spangler
- May 19
- 4 min read
Before the daily briefings existed I was finding that a lot of my team was struggling to keep up with the tasks they needed to make sure were getting done. The team is really pretty good at executing on the tasks tied to a specific beat. The problem was the other tasks. Weekly maintenance work, monthly check-ins, back-burner items that need a small bit of attention every so often. Those were the ones that slipped, and they would just sit there.
I had tried the obvious things. Every task lives in Jira and every team member has a "for you" view that surfaces what's assigned to them. I built calendar events for the recurring ones, and I would send the occasional reminder myself when something had been sitting too long. None of it really moved the needle. We have hundreds of tasks in Jira at any given time because of how many beats we're juggling, and we move so fast every day that keeping up on all of it without help is hard.
The standing rule on my team is that all tasks due in a given week need to be done and submitted by the end of the work week. That's not hard for the team when the tasks are tied to a specific beat with a hard deadline. It's hard when the tasks are the recurring or back-burner items that depend on someone remembering they exist.
The Before
On most days the load is somewhere between twenty and sixty active items in motion across my team. Each person carries their own slice of that, and the tools we had for tracking it were all pull-based. The team member had to go look at Jira to see what was theirs, they had to check the calendar, and they had to remember the back-burner items that didn't have a hard date.
For most of it, it worked. The beat-driven work got done because the beat was a forcing function. The recurring or schedule-driven work was where the slippage lived. A weekly newsletter task that sat for ten days, a monthly competitor analysis that was three weeks late before anyone noticed. Each one is small on its own but it adds up over time.
I'd send Slack pings, I'd add comments to the Jira tickets, I'd bring it up in our Monday call. None of that scaled, and I was the bottleneck.
Why This One
What triggered building this was a two-part thing. A leader had set up a sentiment report that pulls from different areas of the web and automatically sends to him every morning. Same format, same time, every day. The work of going to find the information was gone, the information just showed up.
I'd been carrying my team's task-tracking frustration around for months at that point. Watching that morning sentiment report land in my own inbox is when it occurred to me that the same shape would work for my team. Each person could get their own daily briefing every morning, with what's overdue, what's due today, and what's coming up. They wouldn't have to go look.
How I Authored It
This was something on its own, not a side effect of the dashboard work. I talked to Claude about the frustration and said I knew it already had access to see tasks assigned to individuals within Jira, and asked if it could also work with Clairvoyance and have a message sent through Clairvoyance every morning per employee. It said it did, so we tested it and it worked pretty quickly and pretty efficiently.
Each person gets the same flags. What's overdue, what's due today, and what's coming up in the next week or so. We added a section at the bottom about upcoming beats, because some of the recurring tasks tie to beats in ways that aren't obvious, and knowing that an announcement is coming in eight days reframes a back-burner item as something that needs to move now.
The first version had too much information and not enough prioritization, and I corrected it. After a few iterations of "this section is too long" or "this category should come first" the core was running cleanly within a couple of days. After that it was a matter of letting it run every morning and watching what changed.
The Outcome
The biggest thing I noticed is that tasks aren't going into the abyss anymore. I'm not getting as many team members telling me that they forgot about a task or they were unaware of one, because they have it laid out for them every single morning.
In fact, some of them catch flaws, or they reach out to me when I've become a little bit of a bottleneck and they're waiting on me to move something forward. That's exactly what I want to have happen.
The other thing is the back-burner work. The items that used to slip aren't slipping the same way. Not because the team got better at remembering them, but because the briefing reminds them every day until they're done.
What's Next
The thing that's missing is that right now in Clairvoyance there's no way for me to see if someone has actually looked at the message. It's an upcoming feature that will happen within the Clairvoyance messaging system. I want to be able to see that they've seen it, by having a little icon show up, the little eyeball like Teams, some kind of indicator that they've read the message. That's not something I can fix for them, but it's my next wish.
I can tell from the work itself whether the briefing is landing, because overdue items dropping is a strong signal. But I can't tell, message by message, whether someone read it, skimmed it, or ignored it. The briefing also can't tell me whether someone is acting on it. If they read their briefing and decide not to do the overdue task today, I won't know until the end of the week. That's fine. I don't want to be policing the briefing. It's just where this fix stops.



Comments