Re-Bidding 26 Trades on a New Plan Series Takes Seven Weeks. Agent Teams Return the Packages the Same Afternoon.

How many weeks pass between approving a plan series and pouring the first slab?

Ask that question out loud at a builder's Monday meeting and watch people look at the floor. The plans came back from the architect in March. The first footing on the new phase goes in around Memorial Day. Somewhere in between, twenty-six trade scopes had to be written, priced, compared and awarded, and the person doing it is one purchasing agent with a laptop and a phone that will not stop.

Nobody defends this. Everybody does it. The bidding is slow for one reason only: it happens one package at a time. Concrete goes out, then framing labor, then the lumber package, then roofing, then windows, then mechanical, then plumbing, then electrical, then insulation, then drywall, then trim, then paint, then cabinets, then tops, then flooring, then tile, then garage doors, then gutters, then flatwork, then final grade, then sod and irrigation, then low voltage, then pest treatment, then the fire sprinkler scope in the jurisdictions that require it. Twenty-six in a row, because there is one of her.

Seven weeks is a normal answer. Nine is not unusual if the plan series has four floor plans with three elevations each and the takeoffs have to be redone for every combination.

What those seven weeks are actually spent on

Not negotiating. Negotiating is maybe six hours of the seven weeks and it is the only part that needs her. The rest is assembly: opening the plan set, finding the sheets that matter for this trade, counting what needs counting, writing a scope of work that says what is included and what the trade partner is expected to supply, attaching the right details, addressing it to five or six bidders, sending it, then chasing the three who did not respond.

Then the part everyone skips. When bids come back, they do not come back comparable. One insulation contractor bid to the 2021 energy code with a blower-door target, one bid the old spec, one excluded the garage ceiling, one priced per square foot of conditioned area and one priced per house. Leveling those five bids into one honest comparison takes an hour per trade if you do it properly, so on week six, with the framer already asking when he starts, it gets done properly for the six biggest trades and eyeballed for the other twenty.

That is where money leaves. Not in the price you negotiated — in the exclusion you did not read, which comes back in August as a variance purchase order.

Twenty-six packages that no longer wait in line

Agent Teams means several assistants split one large job between them and work at the same time, then hand back one merged result — the difference between one purchasing agent writing twenty-six bid packages in a row and twenty-six drafts arriving the same afternoon. It shipped as a research preview alongside Claude Opus 4.6, which also brought the ability to hand over an entire plan set at once instead of feeding it sheet by sheet. Those two things together are what make this practical for a builder: the plans are big, and the job splits cleanly by trade.

flowchart TD
  A["Plan series approved: 4 plans, 3 elevations each"] --> B["Full plan set handed over in one piece"]
  B --> C["Team 1: concrete, framing, roofing scopes"]
  B --> D["Team 2: mechanical, plumbing, electrical"]
  B --> E["Team 3: drywall, trim, paint, cabinets, tops"]
  B --> F["Team 4: site work, flatwork, final grade, sod"]
  C --> G["Merged bid tab, one sheet per scope"]
  D --> G
  E --> G
  F --> G
  G --> H["Purchasing agent reads exclusions and awards"]

Each one works the same way your purchasing agent does, just simultaneously: read the sheets that matter for that trade, pull the quantities, draft the scope of work off your standard language, list the inclusions and exclusions you always argue about, attach the details, and address it to the bidders in your trade partner list for that market. What comes back to her is not twenty-six finished decisions. It is twenty-six drafts, each one wrong in a couple of places, all of them wrong faster than she could have made them right.

The second half is where the real gain is

Sending faster only moves your problem to the bidders. The larger win comes two weeks later when the bids return. Leveling is the definition of a job that is slow because it is serial: five bids across twenty-six trades is 130 documents, each one to be read for scope, exclusions, unit basis and allowances.

Several agents can read all of them at once and hand back one comparison sheet per trade, with every bid restated on the same basis and every exclusion pulled out and listed in plain words: this drywall bidder excludes garage, this one includes level 5 at the great room only, this insulation bidder is quoting to the older code. Your purchasing agent then does the part that is actually her job — deciding who can staff four starts a month in July, who has never once left a house without a punch, and who is worth paying two percent more.

The other batch job with the same shape, worth doing the same week: the annual sweep of certificates of insurance and W-9s across 140 trade partners. That is serial only because a human has to open 140 documents and check the expiration date and the additional-insured endorsement. It stops being a two-week job.

What five weeks off the calendar is worth

Assumptions stated, so you can put your own numbers in. Fourteen lots in the phase, developed lot cost of $95,000 each, land and development carried at 9 percent. The purchasing agent has about eight usable hours a week for bidding once you subtract everything else on her desk.

That carry figure is 14 lots times $95,000 times 9 percent, divided by 365, times 35 days. It is the smallest of the benefits and the easiest to defend to a lender. The larger ones are harder to price and more real: you lock a lumber package five weeks earlier in a market that moves, you start into the spring selling season instead of behind it, and every one of the twenty-six trades got leveled instead of six.

Prove it on one phase. Bid it both ways — agents draft, purchasing agent also drafts six packages by hand — and compare the drafts side by side before anything goes out. You will learn more in that afternoon than from any pilot program.

The part that is still entirely your purchasing agent's job

Awarding work is not a spreadsheet decision, and anyone who tells you otherwise has never had a drywall contractor walk off in July. Capacity, crew quality, who answers the phone on a Saturday, who your superintendent will actually tolerate on a job, who is quietly one bad month from not making payroll — none of that is in the bid. Keep the award with the human, always.

Quantities need checking too. An agent counting from plans is good and getting better, but a plan series with three elevations has details that only make sense to someone who has built that plan, and an optional bay window changes six trades at once. Treat the takeoff as a starting count that a person verifies on the two or three trades where a miss is expensive — framing, lumber, and concrete. On paint and gutters, a small error is a rounding difference.

And the relationship damage is real if you get this wrong. Twenty-six bid packages arriving on the same Thursday afternoon from a builder who used to send one a week reads, to a trade partner, like you are shopping the whole town. Call your six core trades before you press send and tell them what you are doing.

Frequently asked questions

Do the bidders know a machine wrote the scope?

They know something changed, because the packages are more consistent than they used to be and every one of them addresses the exclusions your trades have historically snuck in. That consistency is generally welcomed by the good contractors and disliked by the ones who priced you on the gaps. Your name and your purchasing agent's name are still on it, and she read it before it went out.

We only run two plans, not a whole series. Is this worth it?

The calendar gain is smaller, but the leveling gain is identical. Two plans still means twenty-plus trades and eighty-plus returned bids a year. Start with leveling, not drafting.

Can it read our plans if they come as a scanned set?

Usually, if the scan is clean and the sheets are labeled. Where it struggles is a marked-up field set with handwriting over the details, or a plan whose structural revisions live in an email chain rather than on the sheet. If your set has been photocopied three times, fix that first — it is costing you money in the field anyway.

How do I stop this from becoming twenty-six half-finished packages nobody sent?

Give one person the sign-off and a deadline. The failure mode is not bad drafts, it is drafts that pile up because nobody scheduled the review. Block two afternoons on your purchasing agent's calendar and treat them like a bid opening.

Where to start on Monday

Take the last plan series you bid the old way. Hand over the same plan set and ask for the six trades you dread most — mechanical, insulation, drywall, cabinets, flatwork and framing labor — as separate scopes drafted at the same time. Compare them line by line with the packages your purchasing agent wrote in March. That comparison, done once, tells you whether this belongs in your business better than any vendor conversation will.

One side effect worth planning for: when twenty-six packages go out at once, the phone rings for a week. Bidders call with scope questions, superintendents call about start dates, and the calls come in at 6:30 a.m. and 7 p.m. because that is when trade partners are near a phone. CallSphere builds voice and chat agents that answer the office line at those hours, take the question with the trade and the plan number attached, and put the callback on your purchasing agent's calendar — so a bid week does not turn into a week of voicemail.