BRENNEN

Industries / Industrial equipment and services / Industrial distribution

Quoting

Quoting is setting the price a customer pays: list prices less discounts, contract prices, and written quotes for one-off requests and contractor projects.

4

Milestones achieved, from 1977 to 2024.

1

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1977

    Custom parts at catalog prices

    MISUMI, a Japanese trading company founded in 1963 to sell electronic equipment and bearings, began selling press die components in 1965 and published its first catalog of standardized press die parts in January 1977. Before such catalogs, MISUMI says, a buyer drew a blueprint for each part, chose among many small makers and negotiated price and delivery one part at a time; with the catalog the buyer enters a part number and dimensions and orders without a quote.

    misumi.co.jpsg.misumi-ec.com

    Credit

    MISUMI (then MISUMI Shoji Co., Ltd.)

  2. 1999

    Wholesalers fined over prices

    On 26 October 1999 the European Commission fined the FEG, the Dutch association of electrotechnical wholesalers, €4.4 million and its largest member, Technische Unie (TU), €2.15 million. The FEG had sent members lists of gross and net prices calculated by TU, members had discussed prices and discounts at FEG meetings from 1989 to 1993, and wholesalers had pressed suppliers not to supply City Electrical Factors, a British wholesaler that entered the Dutch market in 1989 and complained in 1991. On 21 September 2006 the Court of Justice dismissed the FEG's challenge, leaving the decision in place.

    LessonEach distributor has to set its own net prices; a price list or discount sheet shared among rivals is a cartel even when it is called a recommendation.

    publications.europa.eupublications.europa.eu

    Credit

    European Commission, on City Electrical Factors' complaint

  3. 2017

    Web prices on every item

    In January and February 2017 W.W. Grainger, a US distributor of maintenance and industrial supplies, adjusted list prices across the board and introduced lower web prices on about 450,000 items for medium and large customers without contracts. Customers bought more than it expected, and on 18 April 2017 it pulled the rest of the changes planned for 2018 into the third quarter of 2017, so that every item had a web price. Its shares fell 11.4% that day, by the Motley Fool's count, and its 10-K put a 1.3-point fall in its 2017 gross margin, to 39.3%, mainly down to the pricing actions.

    LessonA visible price cut drew customers faster than planned, and investors marked down the lost margin the same day.

    sec.govfool.comsec.gov

    Credit

    W.W. Grainger, under chief executive D.G. Macpherson

  4. 2024

    A pricing model per salesperson

    Yael Karlinsky-Shichor and Oded Netzer built a model of each salesperson's own pricing at a business-to-business seller of aluminum, from its sales records. In an eight-business-day field experiment covering more than 2,000 quotes and 4,000 product lines, quotes picked at random to show the salesperson the model's price earned 11% more profit per line than the others. Their analysis found the model did better on routine quotes and people did better on unusual, complex ones, and a mix that sends each quote to one or the other did best.

    LessonMuch of the gain came from removing a salesperson's day-to-day inconsistency, not from knowing more than the salesperson.

    business.columbia.edudoi.org

    Credit

    Karlinsky-Shichor and Netzer, Marketing Science, January 2024

Watch for

  1. Quotes drafted from emails

    Proton.ai says its Order & Quote Entry Automation, launched in July 2026, turns customer emails, spreadsheets and PDF files into draft quotes that a sales representative approves before sending. It says that in a trial at one of North America's largest industrial distributors the software found the correct product from description-only requests three times as often as the distributor's own cross-reference system and cut list preparation time by 75%. Distribution Strategy Group notes that the distributor is not named and the results were not independently verified.

    Achieved whena named distributor or an outside tester publishes the test requests, the answer key and the hit rate.

    distributionstrategy.com

    Claimed by

    Proton.ai, July 2026

Open

  1. 01

    Build configured part numbers

    A public test of whether AI tools can write the exact part number of a configured product from a plain request and the maker's published ordering rules: a pneumatic cylinder of a given bore and stroke, a sensor with a chosen cable and connector, a shaft cut to length, an enclosure with given cutouts. 500 requests from at least 20 makers, each with the correct number worked out by two people from the maker's own catalog or configurator.

    The brief
    Credit · unclaimed
  2. 02

    Spec sections to quote lists

    A labeled set of specification sections from public construction projects, such as the panelboard and lighting sections of an electrical specification (Division 26 in the CSI MasterFormat numbering), each marked by hand with every product type, the makers and model numbers named as the basis of design or as acceptable, and whether substitutes are allowed, plus a script that scores an AI tool's reading against the labels. This reading is the first step of a distributor's quote for a contractor's bid.

    The brief
    Credit · unclaimed

Open 01 · the brief

Build configured part numbers

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public test of whether AI tools can write the exact part number of a configured product from a plain request and the maker's published ordering rules: a pneumatic cylinder of a given bore and stroke, a sensor with a chosen cable and connector, a shaft cut to length, an enclosure with given cutouts. 500 requests from at least 20 makers, each with the correct number worked out by two people from the maker's own catalog or configurator.

Why it is still open

Reshape Automation's Industrial AI Accuracy Index 2026 had only 11 configuration questions, 10 of them from one maker's catalog, graded against its own product's answers with no outside review; the best general model with web search scored 22.7% on them. Danilo Valerio and four co-authors argued in September 2026 that language models on their own are unsuited to product configuration, where every answer must obey hundreds of rules.

reshapeautomation.comarxiv.org

Built with what exists

makers' public ordering-code tables and free online configurators; a Python checker for each maker's rules, so that any proposed number can be tested for validity; AI models called through their APIs, with and without web search.

  1. First step

    pick five makers whose ordering-code tables are public, write 20 requests for each, and have a second person check every answer.

  2. Done looks like

    500 requests and answers released under an open license, with scores for at least four AI setups, each run three times, and the share of wrong answers given without any warning.

Open 02 · the brief

Spec sections to quote lists

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A labeled set of specification sections from public construction projects, such as the panelboard and lighting sections of an electrical specification (Division 26 in the CSI MasterFormat numbering), each marked by hand with every product type, the makers and model numbers named as the basis of design or as acceptable, and whether substitutes are allowed, plus a script that scores an AI tool's reading against the labels. This reading is the first step of a distributor's quote for a contractor's bid.

Why it is still open

The nearest outside test, a 2026 paper at the FLAIRS conference on a "Submittals Agent", scored the extraction of submittal requirements from 20 real specifications (94.3% F1) but did not report on named makers or substitution terms.

journals.flvc.org

Built with what exists

project manuals posted on public bid portals by cities, school districts and state agencies; PyMuPDF to split each manual into sections by MasterFormat number; AI models through their APIs; a shared spreadsheet format for labels.

  1. First step

    download ten project manuals from public bid portals and label their panelboard (26 24 16) and interior lighting (26 51 00) sections.

  2. Done looks like

    200 sections from at least 40 projects, labeled by two people with their agreement reported, and published precision and recall on maker names and model numbers, plus accuracy on substitution clauses, for at least three tools.