BRENNEN

Industries / Food, farming and forestry / Food and beverage processing / Bakeries

Making and shaping dough

Making dough means weighing out flour, water, salt and something to make it rise, usually baker's yeast or a sourdough starter, then mixing and kneading them until the dough is stretchy, dividing it, and shaping, rolling or layering the pieces into loaves, rolls and pastry.

7

Milestones achieved, from 1780s to 1992.

1

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1780s

    Yeast made for bakers

    Until the late 18th century, bakers raised bread mostly with the foam and dregs left over from brewing beer, bought from breweries. In the 1780s and 1790s, compressed yeast began to appear in England, Germany and the Netherlands, at first as pressed blocks of fermented hops, rye and malt that could be as little as 4–6% yeast, and by the early 1800s it was outselling the brewers' leftovers; dried yeast went on sale in Vienna in 1822 (Lahue, Madden, Dunn and Smukowski Heil, 2020, drawing on Frey, 1930). In 1868 Charles and Maximilian Fleischmann and James Gaff built a yeast plant in Cincinnati, Ohio, which the company calls America's first commercially produced yeast, and in the 1940s the company added active dry yeast, which needs no refrigeration. Lesaffre of France says it created the first instant dry yeast, Saf-Instant, in 1973. Today most commercial baking uses a handful of yeast strains sold by makers such as Fleischmann's, Red Star and SAF (Lahue and colleagues).

    pmc.ncbi.nlm.nih.govfleischmannsyeast.comlesaffre.com

    Credit

    yeast makers in England, Germany and the Netherlands; Charles and Maximilian Fleischmann and James Gaff; Lesaffre

  2. 1950s

    Dough mixed without stopping

    In the usual way of making bread, mixed dough sat in bulk for hours while fermentation made it stretchy. In 1926 Swanson and Working described a "no-time" dough worked in a pin mixer instead, but bakers were slow to take it up. John C. Baker's experiments led to the Do-maker, a continuous process he described in 1954, in which ingredients flow steadily into a high-speed "developer" that kneads the dough mechanically in place of the long bulk rise. A 1975 review by P. J. Frazier, N. W. R. Daniels and P. W. Russell Eggitt of the British miller Spillers says Baker solved the practical problems and the process was "accepted by the breadmaking industry". A 2010 report in the US trade magazine Baking Business says that style of continuous mixing "dominated the 1960s" and "mostly disappeared in the 1980s".

    LessonTaking out a slow step means controlling what it did; the Do-maker had no automatic control of the work put into the dough, the measure that Chorleywood's research later fixed (Frazier and colleagues).

    cerealsgrains.orgbakingbusiness.com

    Credit

    John C. Baker

  3. 1961

    Energy in place of time

    Scientists at the baking research laboratories at Chorleywood, England, found that bread came out best when the mixer put 37 to 46 kilojoules (kJ) of work into each kilogram of dough, whatever the recipe. The Chorleywood Bread Process uses a powerful batch mixer to put in about 40 kJ/kg in no more than about 5 minutes (Frazier and colleagues, 1975). It adds ascorbic acid (vitamin C) and a little fat or emulsifier, and skips the bulk fermentation of up to three hours that the traditional method needs (Federation of Bakers). The BBC dates the work to 1961 and says it let bakers use low-protein British wheat; Peter Shewry (2026) writes that it allows flour with up to 1 percentage point less protein than traditional methods. By 2011 more than 80% of British loaves were made this way (BBC), and it is also dominant in Australia, New Zealand and South Africa (Shewry).

    LessonA process that cut costs for large plants also, in the BBC's account, "helped put thousands of" small bakeries out of business.

    cerealsgrains.orgfob.uk.combbc.co.ukpmc.ncbi.nlm.nih.gov

    Credit

    the British Baking Industries Research Association at Chorleywood, later the Flour Milling and Baking Research Association

  4. 1963

    Machines wrap and layer dough

    Torahiko Hayashi, who had opened a small confectionery shop in Kanazawa, Japan, in 1950, built an encrusting machine, which wraps one dough or paste around a filling, as in a filled bun: two turning discs draw the outer dough in and close it round the filling. He finished the first machine in 1961, began selling the N101 in 1963 and founded Rheon Automatic Machinery the same year; the N101 made 3,000 filled sweets an hour, about ten times what a skilled confectioner could. His US patent, filed in 1965, covers a device "for continuously making spherical filled encrustations with dough". In 1974 he developed the MM line, which rolls pie dough thin and stacks the sheets, so that pies and Danish pastries, until then made by hand, could be made in quantity. By 2016 Rheon exported its machines to 121 countries.

    gov-online.go.jppatents.google.com

    Credit

    Torahiko Hayashi; Rheon Automatic Machinery

  5. 1971

    Sourdough's microbes identified

    Leo Kline and T. F. Sugihara of the US Department of Agriculture's Western Regional Research Laboratory in Albany, California, studied the "100-year-old" San Francisco sourdough French bread process. With M. W. Miller they found that the yeast doing the raising in five bakeries was Saccharomyces exiguus, and that no baker's yeast was present. They also isolated, "apparently for the first time", the bacteria that make the dough sour with lactic and acetic acids, which need plenty of maltose, a sugar from flour, to grow fast. The bacterium was formally named Lactobacillus sanfrancisco in 1984 from their description and is now called Fructilactobacillus sanfranciscensis.

    LessonA starter's microbes are not set by its city; in 500 starters from four continents, the Global Sourdough Project found "little evidence" of geographic patterns, contradicting the belief behind "San Francisco sourdough" (Landis and colleagues, eLife, 2021).

    pmc.ncbi.nlm.nih.govebi.ac.uklpsn.dsmz.depmc.ncbi.nlm.nih.gov

    Credit

    Leo Kline, T. F. Sugihara and M. W. Miller, USDA

  6. 1986

    Dough enzyme causes asthma

    Bakers add alpha-amylase, a starch-cutting enzyme usually made by the mould Aspergillus oryzae, to flour as an improver. In 1986 X. Baur and colleagues reported in The Lancet on the "role of Aspergillus amylase in baker's asthma". In 1991 J. Brisman and L. Belin described four bakery workers made allergic by the enzyme powder, measured it in the factory air at 0.03 mg per cubic metre, and found that 6 of 20 exposed workers (30%) reacted to it in skin-prick tests; they called it "a potent occupational sensitiser". Britain's Health and Safety Executive (HSE) now lists flour dust and enzyme additives such as amylase as the second most common cause of occupational asthma, and tells bakers to start mixers on slow speed until wet and dry ingredients are combined.

    LessonAn ingredient that helps the dough can harm the people who tip it into the mixer, so improver powders are handled as a workplace hazard.

    ebi.ac.ukebi.ac.ukhse.gov.uk

    Credit

    X. Baur and colleagues; J. Brisman and L. Belin

  7. 1992

    Bromate leaves bread flour

    Potassium bromate is an oxidizing agent that was added to flour to strengthen dough. In 1982 Yuji Kurokawa and colleagues at Japan's National Institute of Hygienic Sciences reported that it caused kidney tumours in rats, and that year Japan cut the most allowed in flour from 50 to 30 mg per kg (Kurokawa and colleagues, 1990). The Joint FAO/WHO Expert Committee on Food Additives (JECFA) had accepted bromate because no residue was detected in baked bread. In 1992, with new safety data and more sensitive tests that did find bromate left in bread, it withdrew the acceptable level, noting that alternatives were available, and it confirmed this in 1995. California's Food Safety Act of 2023 bans the sale of food containing potassium bromate from 1 January 2027. Bans are not always kept: a 2024 study found illegal bromate in 13.9% of 288 bakery and ingredient samples it tested (El-Bahar and colleagues, Food Chemistry).

    LessonAn additive judged safe because none could be found in the loaf was withdrawn once better tests could find it.

    pmc.ncbi.nlm.nih.govinchem.orginchem.orgcalmatters.digitaldemocracy.orgebi.ac.uk

    Credit

    Yuji Kurokawa and colleagues; JECFA; California Legislature

Watch for

  1. Bromate out of US bread

    The American Bakers Association, whose more than 350 member companies run more than 1,200 commercial bakeries, says the "overwhelming majority" of its members already do not use potassium bromate, and that the rest have pledged to phase it out by 31 December 2026. The pledge is voluntary and the announcement names no companies. California's ban on selling food that contains bromate starts on 1 January 2027, and Walmart US is removing it from its own-brand products.

    Achieved whentests of US breads and rolls by a laboratory or agency outside the industry, or a published audit of members' ingredient lists, find no potassium bromate in products from pledging companies in 2027.

    theshelbyreport.comsupermarketperimeter.com

    Claimed by

    American Bakers Association, January 2026

Open

  1. 01

    Telling when dough is mixed

    A clip-on power logger and a model for the spiral mixers that shops use, which tells the baker when a bread dough is fully developed, tested against the baker's own call (often the windowpane test: stretching a piece thin enough to see light through) and against the baked loaf.

    The brief
    Credit · unclaimed
  2. 02

    Robots that shape real dough

    A general-purpose robot arm, taught from bakers' demonstrations, that pre-shapes and shapes yeasted bread dough into round loaves (boules) and baguettes by hand-style moves, judged by bakers and by the baked loaves.

    The brief
    Credit · unclaimed

Open 01 · the brief

Telling when dough is mixed

Credit · unclaimed

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

What to build

A clip-on power logger and a model for the spiral mixers that shops use, which tells the baker when a bread dough is fully developed, tested against the baker's own call (often the windowpane test: stretching a piece thin enough to see light through) and against the baked loaf.

Why it is still open

Large plants control mixing by measured energy, such as the Chorleywood Process's 40 kJ/kg (Frazier and colleagues, 1975), but shop mixers run on a timer and the baker's touch. In 2003 the American Institute of Baking's "smart mixer" project tried near-infrared readings to "predict readiness" of dough on plant lines. Wang and colleagues (Foods, 2024) used a low-cost current sensor to find the best kneading time from the rise and fall of the motor's load, but on pasta dough in one machine. A search of Europe PMC, arXiv and GitHub on 18 November 2026 found no field test on bread doughs in working shop mixers, checked against bakers and bake results, with open data.

foodengineeringmag.comebi.ac.ukcerealsgrains.org

Built with what exists

A three-phase energy meter such as the Shelly Pro 3EM, or SCT-013 clamp current sensors on an ESP32 microcontroller; a probe thermometer for dough temperature; Python with scikit-learn; loaf volume by rapeseed displacement or a laser volume scanner.

  1. First step

    Fit a meter to one shop's spiral mixer and log 20 batches, writing down the second the baker stops each one.

  2. Done looks like

    200 batches of at least three bread doughs in five shops, with the logger's call within 60 seconds of the baker's in at least 80% of batches, loaves from logger-stopped batches within 3% of the volume of baker-stopped ones, and the data and code published.

Open 02 · the brief

Robots that shape real dough

Credit · unclaimed

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

What to build

A general-purpose robot arm, taught from bakers' demonstrations, that pre-shapes and shapes yeasted bread dough into round loaves (boules) and baguettes by hand-style moves, judged by bakers and by the baked loaves.

Why it is still open

Research robots shape stand-ins for dough. A 2022 study, "Robotic Dough Shaping", had an arm with a rolling pin roll Play-Doh, Plasticine and kinetic sand into flat target shapes; RoboCook (2023) learned to make dumplings and letter-shaped cookies with tools; a 2024 study taught a humanoid robot to roll dough from human demonstrations. None reports a baked loaf. Bakeries that automate shaping buy dedicated machines, such as moulders or Rondo's RONDObot (2017), which Rondo says shapes very wet doughs into loaves and rolls. A search of arXiv and GitHub on 18 November 2026 found no counted, filmed trial of a general robot arm shaping yeasted bread dough and baking the result.

arxiv.orgarxiv.orgarxiv.orgrondo-online.com

Built with what exists

An open robot model such as NVIDIA GR00T 2 (Note: "Five demonstrations per task"); a low-cost arm such as the SO-101 from Hugging Face's LeRobot project, or a Franka Research 3; an Intel RealSense D405 depth camera; a bench-scraper tool on the gripper; a food-safe sleeve over the arm.

  1. First step

    Record 50 demonstrations of a baker pre-shaping 500 g pieces with a bench scraper through the arm's teleoperation controls, and train a first policy on them.

  2. Done looks like

    20 filmed attempts each at boules and baguettes from the same dough at 70% hydration (70 g of water per 100 g of flour), scored blind from 1 to 5 by three bakers against 20 hand-shaped controls, with baked volume and height measured; a published tally, with at least 15 of 20 robot loaves within 10% of the hand-shaped volume as the bar.