A Co-Pack RFQ Lands Friday at 4:50 and the Priced Packet Takes Six Days. In 2026 You Hand Over the Goal Instead.

How many co-pack inquiries did you not answer last quarter?

Not turn down — not answer. The broker's email that arrived on a Friday afternoon in March asking whether you could run a 12-ounce jarred sauce, 6/12 case pack, forty thousand cases in year one, samples wanted before Expo season. It sat in the inbox until the following Thursday, by which point the brand had been quoted by two other co-packers and stopped replying. Most owners cannot tell you that number, because a quote that never got built leaves no trace in the ERP.

Nobody was lazy. In a plant this size the person who builds a quote is the controller, the operations manager or the owner — and building one properly means touching a costed bill of materials, four packaging vendors, a line-rate assumption, a changeover estimate, an overhead rate and a freight quote before a single number gets written down. That does not fit in the gaps of a Friday.

Everything a real co-pack quote has to contain before you can send it

A serious quote for a jarred, acidified sauce is not a price. It is a packet, and a brand that has been burned before asks for all of it: a costed formula at current ingredient pricing, turnkey and tolling numbers side by side, minimum run quantity, case pack and pallet configuration with Ti-Hi, lead time from purchase order to ship, who owns the packaging inventory, the changeover and cleaning charge on short runs, and what happens to the price if tomato paste moves.

Underneath that sit assumptions only your plant can supply. Kettle batch size and cases per batch. Line rate on the glass line, which is a different number for a 12-ounce jar than the 24-ounce you normally run. Crew size and loaded labor rate. Scrap and giveaway on fill weight. Warehouse days before pickup. And the regulatory reality: an acidified food needs a scheduled process from a process authority and the filings that go with it, and somebody has to say whether that cost and that calendar time sit with you or the brand.

Then the paperwork the brand's quality person asks for the moment they like the price — your SQF or BRCGS certificate with its expiry, an insurance certificate at their required limits, a sample specification sheet, an allergen statement for the lines it would run on, and your co-manufacturing agreement. Half the quotes that die do not die on price. They die because the packet came in four emails over nine days and the brand decided you looked disorganized.

The six-day quote, and where the days actually go

Walk the calendar of a real one. Friday 4:50 p.m., inquiry arrives. Monday, the operations manager emails purchasing for current pricing on paste, oil and the acidulant. Tuesday, purchasing replies on two of three; the glass vendor wants annual volume before quoting. Wednesday, the controller pulls last year's actual line rate for the 12-ounce run and argues about whether that rate is fair given the new labeler. Thursday, the freight number lands. Friday, the spreadsheet gets built, the owner changes the overhead absorption, and the packet goes out the following Monday — day eleven, in a business where the founder emailed three co-packers the same week.

The workaround everyone pretends is fine is the ballpark: quote off the last similar item, add ten percent, fix it later. That wins some business and it wins the wrong business, because the item you copied is the one whose changeover runs twice as long.

flowchart TD
  A["Goal handed over: price 12 oz jarred sauce, 40,000 cases, turnkey and tolling"] --> B["Job split four ways"]
  B --> C["Formula costed at this week's ingredient pricing"]
  B --> D["Glass, closure, label, case and pallet quotes assembled"]
  B --> E["Line rate, crew size, changeover and CIP time pulled from last year's runs"]
  B --> F["Freight to the brand's DC and warehouse days priced"]
  C --> G["Cost per case built, turnkey and tolling side by side"]
  D --> G
  E --> G
  F --> G
  G --> H["Packet assembled: quote, spec sheet, allergen statement, SQF certificate, agreement"]
  H --> I["Owner sets the margin and sends it"]

What actually changed: you hand over a goal, not a task list

Until this year the assistant was a chat box. You asked, it answered, you did the next step yourself. Useful for the cover email; useless for the packet.

Anthropic shipped Claude Cowork on 12 January 2026 and expanded it to phone and web in July. OpenAI shipped ChatGPT Work on 9 July 2026 running on GPT-5.6. Both do the same new thing: you give a goal, connect it to your files and the programs you already use, and it breaks the job into steps, works on its own for a couple of hours, and hands back finished work — a completed spreadsheet, a priced quote packet, a set of filled forms. Both were built for people who do not write software, which in a food plant means your controller and your operations manager.

There is a second piece. Claude Opus 4.6 introduced Agent Teams — several assistants splitting one job and working on different parts at the same time. On a quote that fits exactly: one costing the formula, one chasing packaging quotes, one digging last year's real line rates out of the production records, one pricing freight. They finish together instead of in sequence, which is why the six-day calendar collapses rather than shortens.

The same inquiry, from Friday 4:50 to Monday 9:00

Friday 4:50 p.m. The operations manager forwards the inquiry with one instruction: price this at 40,000 cases turnkey and tolling on our 12-ounce glass line, current ingredient pricing, standard terms, and build the full packet.

Over the next two hours it opens the item master and the costed bill of materials for the closest existing sauce, pulls actual yields and case counts from last year's production records for that line rather than the standard rate, drafts quote requests to the four packaging vendors on file, prices freight to the brand's distribution center by zone, and assembles the packet — quote sheet with assumptions listed, draft spec sheet, allergen statement, current certificate, and the co-manufacturing agreement.

Monday 9:00, the owner opens one document. The assumptions are on the first page and two are wrong: the line rate came from a run before the new labeler went in, and the changeover assumption left out the extra sanitation step this brand's no-soy claim requires. He fixes both, sets the margin himself — that number is never delegated — and the packet goes out before ten. Vendor quotes arriving midweek get folded into a revised number on Wednesday, still ahead of where day one used to be.

What getting the quote out in two days is actually worth

All illustrative: a co-packer taking three serious inquiries a month, average award of 30,000 cases a year at $7.90, contribution margin 19% after ingredients, packaging and direct labor.

The win-rate bump is the softer of the two assumptions, so test it before you believe it — but the volume side is not soft at all. Twelve inquiries a year that get no reply is a fact you can verify this afternoon by searching the inbox. Even if the win rate never moves, quoting ten more programs at 22% is two more awards a year, and on a plant with open line hours in the spring, two programs is what makes the fall profitable rather than merely busy.

The parts of a quote you cannot hand over

Margin. Whatever the packet says, the last line is the owner's number, set with knowledge no file contains: how full the fall calendar already is, whether this founder will be a nightmare about short runs, whether you want to be in glass at all next year.

Capacity honesty is second. An assistant reading last year's records will happily assume the line is open in September because nothing is booked. You know September is when the holiday co-pack work lands and third shift is already thin. Nothing in the data says that.

And technical feasibility stays on the bench. Whether your kettle holds that viscosity at that batch size, whether the fill temperature holds the pH the process authority filed, whether that jar runs on your capper without scuffing — that is your R&D person and your line lead running a trial. A quote built on a formula nobody has run is one you regret at the first production run, not at the bid.

And it will use last year's numbers with total confidence, including the wrong ones. If your standard line rates have not been updated since the labeler changed, the packet is wrong in exactly the way your spreadsheet was wrong.

Frequently asked questions

Our costing lives in a spreadsheet the controller built, not in the ERP. Does that matter?

No, and it may help. A spreadsheet with visible logic is easy to work from. What matters is that somebody can explain the tabs once. The plants that struggle are the ones where the costing logic lives only in the controller's head.

How long does it take before it produces a quote we would actually send?

The first two or three are drafts you rewrite heavily, because you are teaching it your assumptions — how you absorb overhead, whether you quote pallet-in or pallet-out, your terms and pricing-validity window. By the fourth or fifth it produces something you edit rather than rebuild. Plan on two weeks, not a day.

Can it send the vendor emails for the packaging quotes itself?

It can draft them and, if you allow it, send them. Most owners start draft-only for anything leaving the building with the company's name on it, then loosen up for routine repeat requests to long-standing vendors. That is the right order.

Where we come in. The quote packet is half of a co-pack inquiry. The other half is first contact — the founder who calls the main line on a Tuesday evening, gets a voicemail box, and calls the next co-packer on the list. CallSphere builds AI voice and chat agents that answer the plant's phone and website chat around the clock, ask the four questions that qualify a co-pack inquiry — product, pack, volume, timing — and book the call with your operations manager before the caller moves on.