Industries / Industrial equipment and services / Industrial distribution
Technical sales
Technical sales is helping a customer choose the right product: outside salespeople who visit plants and job sites, inside staff on phones and email, specialists who size and apply products, and counter staff who build items such as hose assemblies.
4
Milestones achieved, from 1893 to 2002.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1893
Advice carries a warranty
Britain's Sale of Goods Act 1893, section 14(1), made it an implied condition of a sale that goods be reasonably fit for a purpose the buyer made known, where the buyer relied on the seller's skill or judgment, "whether he be the manufacturer or not." A seller who picks a product for a stated job answers for that choice even though it made nothing. In the United States the same rule is the implied warranty of fitness for a particular purpose, section 2-315 of the Uniform Commercial Code.
Credit
UK Parliament
1956
A degree in distribution
Texas A&M, then the Agricultural and Mechanical College of Texas, founded an Industrial Distribution degree program in 1956, in the Department of Industrial Education of its College of Engineering. In 2001 it added a distance-learning master's degree in Industrial Distribution, whose first class graduated in 2003.
Credit
Texas A&M University
1985
Reps or salaried sellers
Erin Anderson studied 13 electronic component makers across 159 US sales districts to see when they used their own salaried salespeople instead of manufacturers' representatives, independent agents paid by commission who sell several makers' lines. Makers went direct where a salesperson's results were hard to judge, where product lines were complex and hard to learn, and where a district needed much work besides selling; company size and travel made no difference.
Credit
Erin Anderson, Marketing Science, 1985
2002
A hose that looked right
On 14 August 2002 a chlorine transfer hose burst during rail tank car unloading at DPC Enterprises near Festus, Missouri, and 48,000 pounds (about 22 tonnes) of chlorine escaped over about three hours; 63 people from the community sought medical evaluation. The US Chemical Safety Board (CSB) found that the hose's braid was 316L stainless steel instead of Hastelloy C-276, which looks identical, and that Branham Corporation, the distributor that cut and assembled the hose, had shipping papers saying Hastelloy, relied on visual checks and had no quality system to confirm materials. An X-ray fluorescence test, a common way to check an alloy, would have told the metals apart. The CSB asked NAHAD, the hose distributors' association, to set up continuous marking from maker to user, and closed that recommendation as acceptable in January 2012 after NAHAD added it to its hose assembly guidelines.
LessonA distributor that builds or relabels a product takes on the duty to prove what it is made of; tags and appearance are not proof.
Credit
US Chemical Safety Board; NAHAD
Watch for
General AI misses part questions
Reshape Automation, which sells a grounding system called ReshapeX, says general AI models scored 12% to 14% without web search on its 100 questions from 14 makers' catalogs, covering part search, validation, cross-reference, configuration and technical support, and at best 52% with search (GPT-6 Astra). It estimates that 36.5% of the 349 wrong answers stated a fact, figure or product code with no hedge. Its own ReshapeX answers served as the answer key, so ReshapeX was not scored on equal terms, and no third party reviewed the study.
Achieved whenan outside group reruns the questions, or a public set like them, with its own grading and finds similar scores.
Claimed by
Reshape Automation, October 2026
Open
01
A test for hose orders
A public set of 300 hose requests of the kind a counter or inside salesperson receives (a text message, an email, a photo of the printing along an old hose, called the layline), each with an answer key from people trained in hose assembly: which facts of the trade's STAMPED order checklist are missing, whether the stated fluid, temperature and pressure suit the hose named, and the matching hose in at least two makers' published catalogs. Then score AI tools and makers' selector apps against it. Data only: no hose is built or pressurized.
The briefCredit · unclaimed02
Belt drives sized by AI
A test of 200 V-belt drive design tasks, each stated the way a salesperson hears it (motor power and speed, driven machine speed, distance between shafts, hours of use a day, kind of load), with the correct belt type, pulley sizes and number of belts worked out in a maker's drive design software and checked by hand against its printed tables. Score AI models on them.
The briefCredit · unclaimed
Open 01 · the brief
A test for hose orders
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public set of 300 hose requests of the kind a counter or inside salesperson receives (a text message, an email, a photo of the printing along an old hose, called the layline), each with an answer key from people trained in hose assembly: which facts of the trade's STAMPED order checklist are missing, whether the stated fluid, temperature and pressure suit the hose named, and the matching hose in at least two makers' published catalogs. Then score AI tools and makers' selector apps against it. Data only: no hose is built or pressurized.
Why it is still open
Danfoss's Hose & Fitting Navigator app, first released in December 2024, reads laylines with AI to suggest replacement hoses from other makers, but its listing gives no measured accuracy, and no outside comparison of such tools was found. The CSB's DPC report shows the stakes: the wrong hose looked identical to the right one.
Built with what exists
makers' public hose catalogs and chemical resistance tables; Danfoss's Hose & Fitting Navigator and other makers' online selectors as tools under test; a vision-language model such as Claude Fable 5.2 (see the Note "Three working devices") to read laylines; a spreadsheet answer key.
First step
photograph and transcribe the laylines of 50 used hoses from a farm, garage or school workshop, and write 50 requests from public catalog pages.
Done looks like
300 cases with keys checked by two people, released openly; for each tool, the share of cases where it asked for the missing facts, the share of unsuitable matches it let through, and its reading accuracy on at least 100 layline photos.
Open 02 · the brief
Belt drives sized by AI
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A test of 200 V-belt drive design tasks, each stated the way a salesperson hears it (motor power and speed, driven machine speed, distance between shafts, hours of use a day, kind of load), with the correct belt type, pulley sizes and number of belts worked out in a maker's drive design software and checked by hand against its printed tables. Score AI models on them.
Why it is still open
Benchmarks of AI models on engineering design exist, among them EngDesign (NeurIPS 2025, nine engineering domains), but none of those found tests the choice of catalog components such as belts and pulleys, and Reshape's 2026 index had only five technical-support questions.
Built with what exists
Gates Design Power, which includes the Design Flex Pro belt drive program, for reference designs; makers' printed drive selection tables; Python to check rated power against required power; AI models through their APIs.
First step
work 20 tasks by hand from a maker's drive selection tables and check each in Gates Design Power.
Done looks like
200 tasks and keys released; for each model, the share of designs whose belts carry the required power with the stated service factor, and the share that match the reference design within one pulley size.