The Open-Loop Audit: Who Is Waiting on You

Every morning, an agent scans email, calendar, and messages and returns one ranked list: who is waiting on a reply, what was promised and not delivered, and what breaks if it waits another day. The day starts with the expensive items at the top.

The problem

A superintendent's or PM's commitments live in five places at once: an email thread from Tuesday, a text from a foreman, a promise made out loud in the trailer, a submittal someone is holding, an inspection that needs scheduling. Each one is small. Collectively they are the job's real risk list, and nobody holds the whole of it in their head — the items that surface are the ones somebody chases, not the ones that cost the most.

The failure mode is quiet. An RFI answer that sits for three days doesn't announce itself; it shows up later as a crew standing around, a rescheduled inspection, a sub who stopped waiting and moved to another job. By the time a stranded commitment becomes visible, the cheap fix is gone. Most managers compensate by re-reading their inbox defensively, which costs real time every day and still misses the threads that never arrived by email at all.

How the agent workflow works

1. On a morning schedule, the agent scans the connected sources: email inbox and sent folder, calendar, and message channels the manager works in.
2. It extracts open loops — messages that ask a question and got no reply, commitments in the manager's own sent messages ("I'll send that Friday"), deadlines inside seven days with no visible progress, and threads where the other party has followed up more than once.
3. Each loop is ranked by cost of delay, not by arrival order: does it block a crew, an inspection, a delivery, a payment, or another company's schedule? A waiting crew outranks a waiting newsletter.
4. The output is one short list — typically the top five to eight items — with the source thread named for each, how long it's been open, and what the next action is. It lands before the workday starts.
5. Items the manager closes or answers drop off the next run. Items that stay open climb the ranking, so a loop can't quietly age at the bottom of the list forever.

What you need to set it up

• Agent access to email and calendar at minimum; message channels added where the real commitments live.
• A ranking rubric in plain language — what "expensive delay" means on your jobs: blocked crews, inspections, and pay applications usually top the list.
• A fixed delivery time and place (email or chat) early enough to act on.
• A weekly ten-minute review at first: mark what the audit got wrong, so the ranking and extraction rules get corrected while trust is being built.

Where it breaks

The audit only sees what is written down and connected. The promise made verbally in the trailer — often the most important one — is invisible unless it later appears in a message. Extraction is also fallible: the agent will flag threads that were actually resolved by phone, and it can misread who's waiting on whom in a long CC chain, especially when the real decision-maker isn't the sender. Ranking by cost of delay is a judgment, and early versions get it wrong in both directions — trivia ranked urgent, real blockers ranked low because the thread sounded casual. There is also a privacy line to manage deliberately: the agent is reading the manager's own communications, and which channels get connected should be a conscious choice, not a default. Treat the first month's lists as a draft of the manager's attention, reviewed skeptically, not as ground truth.

The bottom line

Best fit for superintendents and PMs juggling multiple crews, subs, and owners at once. It doesn't close loops — it makes sure the loops close in the order that costs the least, and that none of them strand quietly.