Case study · An internal build
We built an agent that reads a six-month program's schedule and task list every morning, tells the two operators what matters today, and writes the weekly status report for client leadership. Here is what it does, where people stay in charge, and what it would take to run on your project.
A six-month client program was being tracked by hand, through emails and deliverables. We built an agent that reads the schedule every morning and tells us what it means for today, then templated it so it can run on any project with a defined plan.
The base brief was built in under an hour and sent the next morning.
A week of manual tracking removed, by our own estimate
Recovered over the six-month program, by our own estimate
To build the base daily brief
Weekly status report for client leadership, progress derived from the schedule
A six-month program with a schedule, a task list, milestones, 24 contract deliverables and a client leadership team to report to. The two operators tracked it by hand: reading back through emails and deliverables to work out what was due, what was owed by whom, and what had quietly moved. By our own estimate that took about 90 minutes a week, and the small items were the ones that slipped: a draft that needed to go out before a review cycle, a question to the client that nobody had asked yet.
The schedule and task list were already well defined, which is the foundation none of this works without. What was missing was something that read them every morning and told us what they meant for today.
Started from the plan we already had: a schedule as the single source of truth, with every task derived from its dates.
Built a morning brief that reads the plan and the inbox and lands at 8:00 every weekday.
Added a one-page weekly status report for client leadership, with progress derived from the schedule.
Gave the team a project snapshot: the whole plan as one page, rebuilt every morning.
Kept AI to the reading and the judgment. Sending runs on plain code and a fixed clock, so the brief arrives on time.
The schedule and task list hold the facts. The agent reads them, and the people who own the program decide what to do about them. The work splits like this:
Every weekday at 8:00am Eastern, the two operators open a brief that shows exactly which tasks are overdue and which are due today, and who owns each one. Below that: what email shows is already finished, which questions are waiting on the client, what has slipped, and how far the schedule has drifted from the work plan the client is holding. A one-sentence headline at the top names the most valuable thing to act on first. A one-page project snapshot, rebuilt each morning, shows the whole plan to anyone on the team. Once a week the same data produces a one-page status report for client leadership, with progress bars and stoplights derived from the schedule rather than set by hand.
The 90 minutes a week of tracking is gone. Over a six-month program that is about 40 hours, by our own estimate. What it catches is the small things: sequencing (the draft before the review before the delivery), questions sitting unanswered in the client's court, and items that moved without anyone saying so.
AI does the reading and the judgment: scanning the inbox and writing the headline. Sending runs on plain code and a fixed clock, so the brief arrives at 8:00 whether or not the AI part has finished.
Any project with a defined schedule, tasks, milestones and owners can run on the same agent, and one agent can serve several projects at once. The precondition is the plan itself: if your schedule lives in people's heads, that is the first week of work, and it is work worth doing on its own. Setup is a one-time baseline of the tasks already finished. From then on, the brief arrives each morning and the status report each week, and your project manager reads, decides and sends.
The agent never closes a task, moves a date or sends anything to the client. It reads, compares and reminds. Your team decides.
If you have a schedule and a task list, we can show you what a morning brief would say about it. Start with a free AI Readiness Snapshot or a conversation.
