Agents in the Trades
Management

Two-week lookahead, built not typed

Every Friday afternoon, the superintendent opens the master schedule, a delivery tracker, an inspection list, and last week's lookahead, then rebuilds the next two weeks by hand. It takes a chunk...

The problem

Every Friday afternoon, the superintendent opens the master schedule, a delivery tracker, an inspection list, and last week's lookahead, then rebuilds the next two weeks by hand. It takes a chunk of the afternoon. Most of that time is not thinking. It is copying.

The thinking is the valuable part: will the racking crew have material on Tuesday, is the inspector confirmed for the trenching before backfill, can the electrical sub start string wiring if the combiner boxes are still showing a late ship date. The copying crowds that out.

When the lookahead is built tired and in a hurry, errors creep in. A delivery date goes stale. An inspection that was rescheduled on Wednesday still shows the old day. The foremen plan around the document, so a bad line costs a crew a morning.

How the agent workflow works

The agent runs on a schedule, Thursday evening or Friday morning, so the superintendent has a draft before the afternoon starts.

What it reads:

What it produces: A draft lookahead organized by day and by crew or trade. Each activity lists the material it depends on and the inspection it is waiting on, if any. The agent also writes a short flag list at the top, covering items such as:

Where the human checks it: The superintendent reads the flag list first, then edits the draft. Nothing goes to foremen or subs until the superintendent approves it. The agent has no ability to send the lookahead, change the schedule, or contact a supplier. It drafts and flags. The superintendent decides what is real, resequences the work, and owns the final document.

What you need to set it up

Setup time is a few hours for a first working version, mostly spent cleaning the delivery and inspection list and writing the instructions. Expect another few rounds of adjustment over the first month as you see what the agent misses.

Where it breaks

Stale inputs. The agent only knows what is written down. If a supplier called the foreman about a delay and nobody updated the tracker, the draft will confidently show the old date. The flag list will not catch what was never recorded.

Logic the schedule does not hold. Experienced supers carry sequencing in their heads: the crane is shared, the access road floods, the inspector never confirms until the day before. The agent will not know unless you write it into the instructions, and even then it will miss exceptions.

Messy exports. Schedule exports with odd formatting, merged cells, or renamed activities can cause misreads. An activity matched to the wrong delivery is worse than a missing one, because it looks verified.

Over-trusting the flag list. A clean flag list does not mean a clean two weeks. If the superintendent starts skimming, the checkpoint becomes a rubber stamp.

Inspection emails. Confirmations written in loose language, such as "we'll try to get out Thursday," may be read as firm. Treat any inspection date from free text as unconfirmed until you verify it.

The bottom line

This workflow takes the assembly work off the superintendent and leaves the judgment where it belongs. The agent compiles and cross-checks. The superintendent decides.

It is only as good as the schedule, order tracker, and inspection list behind it. Start with one project, compare the draft against what you would have built by hand for a few weeks, and note every miss. Those misses tell you what to add to the instructions and what to keep checking yourself.

Related use cases

← All use cases

One working agent workflow, every day.

Real AI agent use cases from solar, construction, and field operations — what they do, what they cost, and where they break. No hype, no theory.