Mostrar mensagens com a etiqueta automatização. Mostrar todas as mensagens
Mostrar mensagens com a etiqueta automatização. Mostrar todas as mensagens

sábado, junho 03, 2023

Coincidências

Quinta-feira, durante a minha caminhada matinal li em "Lead and disrupt: how to solve the innovator's dilemma" de Charles A O' Reilly III and Michael L. Tushman:
"Amazon lost a quarter of its annual profits to turnover in 2021. According to internal documents reported by Engadget, in 2021 Amazon lost more than eight billion dollars to attrition. Only one out of three new hires stayed for more than three months. The turnover problem is so urgent that the company is beginning to worry about running out of new people to hire"

Entretanto; ontem, no WSJ do passado dia 1 de Junho li em "AI Tapped to Weed Out Products That Are Damaged":

"Amazon.com is rolling out artificial inteligence across a dozen of its largest warehouses ...

Amazon so far has implemented the AI at two fulfillment centers and plans to roll out the system at 10 more sites in North America and Europe."

terça-feira, outubro 13, 2020

Automatização ...

Imagem retirada do semanário Expresso deste Sábado.

Como são as empresas portuguesas? Pequenas e médias empresas.

Como competem nos mercados internacionais? É porque têm o preço mais baixo? 

Não, é porque apostam na flexibilidade, rapidez e diferenciação.

Recordo a hipótese Mongo.

O que é que Mongo requer? Flexibilidade e rapidez!

O que é que a automatização mal feita prejudica? A flexibilidade e a rapidez!

Recordar:

 




sexta-feira, janeiro 03, 2020

"That's our edge!"


Acredito que em muitas áreas a Inteligência Artificial poderá substituir com vantagem os humanos. No entanto, nunca esqueço as palavras de Kasparov acerca do trabalho conjunto entre humanos e inteligência artificial.
"relatively weak computers working alongside human players will wipe the floor with the most advanced supercomputers. Collaborative rather than oppositional thinking has yielded radical advances in chess theory, and opened up whole new areas of play." (fonte)
"humans and machines will work together, and we, as customers, will be allowed, once more, to lazily beg for help" (fonte
Nunca esqueço:
- That's our edge! (fonte

sábado, novembro 23, 2019

Especulação sobre mais um falhanço da automatização

Ao longo dos anos tenho aqui deixado exemplos acerca das limitações da automatização da produção em massa, quando o mundo económico caminha para mais diversidade.

Assim, os exemplos que cito mais vezes são o da Toyota e o da Mercedes. Recentemente, encontrei o exemplo da Adidas (embora a coisa ainda não esteja bem explicada). Fazem-me lembrar o caso recente de uma conversa com um empresário. Alguém que pensava e que não era nenhum burgesso:
  • Via cada vez mais variedade, mais referências nas encomendas;
  • Recebia cada vez mais encomendas, mas mais pequenas; 
  • Continuava a procurar gerir a sua empresa com base em indicadores de eficiência.
Ou seja, tinha os clientes a deslocarem-se num sentido, mas ele continuava a gerir a empresa à maneira do século XX.

Quando o mercado exige mais variedade, a gestão tem de abandonar os cânones do século XX e em vez de continuar prisioneira do lado direito da figura abaixo, pensar em transitar para o seu lado esquerdo:
Agora fazer figura do vermelho abaixo é que não dá:
Stuck-in-the-middle ao tentar ser tudo para todos. Espero que tenha apanhado a ideia de Skinner e da plant-within-the-plant.

Assentemos ideias - automatização e Mongo, a metáfora que uso para descrever o mundo económico para onde caminhamos, não são incompatíveis. O que é incompatível é tentar usar a automatização em produções que não são em massa, com a abordagem da produção em massa. A automatização é fantástica para apoiar um artesão, ou uma cooperativa de artesãos, ou uma pequena unidade produtiva de artesãos. O que as empresas grandes fazem é tentar esticar a produção em massa para além do seu prazo de validade e dá suckiness

O que as empresas grandes fazem é rejeitar o futuro, porque continuam agarradas ao passado. Por exemplo, a VW, à boa maneira alemã, ao contrário do empresário lá em cima que aceita tudo, analisa e rejeita o que não se encaixa no seu modelo de negócio: depois, arrepende-se.

Entretanto, ontem tiveram a atenção de através de um tweet ... 
... me fazerem chegar este texto "Boeing abandons its failed fuselage robots on the 777X, handing the job back to machinists" (engraçado, passei o dia a trabalhar via iternet com alguém em Seattle. Força Brett):
"After enduring a manufacturing mess that spanned six years and cost millions of dollars as it implemented a large-scale robotic system for automated assembly of the 777 fuselage, Boeing has abandoned the robots and will go back to relying more on its human machinists.
.
Boeing said Wednesday it is adopting a different approach that “has proven more reliable, requiring less work by hand and less rework, than what the robots were capable of.”
...
The automation has never delivered its promise of reduced hand labor and Boeing has had to maintain a substantial workforce of mechanics to finish the work of the robots. Because of the errors in the automation, that often took longer than if they had done it all by hand from the start. Boeing said Wednesday there are no planned changes in total staffing."
Interessante que me apetece especular e dar uma sugestão que não tem nada a ver com o que escrevi acima acerca da variedade. Quem é o cliente da linha de montagem da Boeing? A Boeing! A customização não deve ser o problema aqui, até porque a palavra que sublinho acima é "rework". A Toyota e a Mercedes abandonaram os robots não por causa do rework, mas por causa do tempo necessário para voltar a produzir, tempo para afinar e introduzir novos parâmetros.

Como a Boeing é americana, como a Boeing me tem dado vastos exemplos, ao longo dos anos, de sofrer da doença algo-saxónica, a doença da tríade, a doença do século XX. Caricaturando a situação, aquilo deve ser gerido por clones de Kevin O'Leary.

O que me parece mais provável neste caso é outro tema que já apanhei num projecto:
  • Empresa avança para automatização, para reduzir custos porque a gerência só vê custo, custo, custo;
  • Empresa monta robots na produção, mas continua a comprar a matéria-prima como sempre comprou;
  • A matéria-prima barata, sem ser defeituosa, tinha grande variabilidade na espessura, algo que os humanos em conjunto com as máquinas dos anos 70, resolviam rapidamente com a introdução ou retirada de um componente auxiliar;
  • A matéria-prima barata continuou a ser comprada para ser usada com os robots e diferenças de 1 cm ou mais eram a morte do artista. A produção estava sempre a parar porque o lase do robot ou não encontrava o material que era suposto encontrar, segundo o desenho, ou encontrava material numa outra posição tridimensional... asneira: paragem; ou rework, ou scrap! Produtividade é que não.
Aposto que a automatização bem sucedida da Boeing obrigaria a diferentes exigências para o aço comprado... isso implicaria aço mais caro... pois, Total Value Ownership.

sexta-feira, novembro 22, 2019

A construção do futuro


"Michigan, 20 masons lay bricks for a huge dorm, as big as three football fields, at the Naval Station Great Lakes in Illinois. Compared with those in years past, these workers are doing far less laying and “buttering” and, instead, are focused on quality and on cleaning up mortar joints.
.
A robot named SAM handles the real grunt work.
.
SAM, a clawlike metal arm extending from a cage, moves back and forth along the walls, buttering and layering a brick every eight to 12 seconds. Nearby, another robot called MULE uses a burly 12-foot arm to lift heavy cement blocks for workers, who then guide them into place.
.
Neither bot takes sick days or gets sore muscles, and both can work around the clock.
...
These days, reliability is a big issue in the construction industry, responsible for nearly $10 trillion in global spending annually. The vast majority of large construction projects go over budget and take 20% longer than expected, according to consulting firm McKinsey.
.
The problem is partly owing to a labor shortage.
...
Ekso Bionics makes robotic vests that support a worker’s arms for jobs like drilling or installing piping overhead. It also sells a robotic arm that makes it easier for workers to use heavy tools, reducing fatigue and injury.
.
And Dusty Robotics makes small autonomous bots that roll around construction sites and mark lines on concrete floors that indicate the location of walls and infrastructure, based on construction documents.
.
Few companies are trying anything as ambitious as San Francisco startup Built Robotics. It sells autonomous technology for bulldozers and other heavy equipment. The tech can enable a Caterpillar tractor, among others, to move dirt and lift pallets of wood—all without anyone in the cab."





Trechos retirados de "Bots Start Building" na revista Fortune de Dezembro de 2019.

quinta-feira, novembro 14, 2019

Mongo e automatização, fiem-se no conto do vigário

Uma notícia interessante, "Adidas to Close German and US Robot Factories":
"Adidas plans to close high-tech "robot" factories in Germany and the United States it launched to bring production closer to customers, saying on Monday that deploying some of the technology in Asia would be "more economic and flexible."
...
Adidas started production of shoes largely by robots at its "Speedfactory" in the southern town of Ansbach near its Bavarian headquarters in 2016 and opened another near Atlanta in 2017."
O artigo não é claro quanto aos motivos do encerramento. No entanto, acho estranho aquele "deploying some of the technology in Asia would be "more economic and flexible.""

O que é que costumo escrever aqui sobre os elogios desbragados à automação? Que Mongo tem demasiada variedade para um autómato resolver. Recordo:

No primeiro texto apresento o caso da Toyota que, perante o aumento da variedade, retirou robots da linha de produção e voltou a meter humanos. No segundo texto é o exemplo da Mercedes.

Os robots batem os humanos se a produção for constante, mas sempre que é preciso mudar de modelo, é preciso calibrar e afinar programas, o que pode levar até 3 semanas em casos extremos. Com os humanos não se perde tempo com mudanças.

Recordo que nos pré-históricos tempos da primeira guerra do Golfo, 1990, as linhas de montagem de cablagens do Nissan Primera tinham 27 modelos.

domingo, março 17, 2019

A fábrica do futuro

Mais um texto que parece retirado deste blogue, um texto sobre Mongo e muito diferente do que o mainstream escreve, normalmente, encadeado pelas luzes da automatização. Recomendo pois a leitura de "The manufacturing job of the future: clean, urban, and better paid":
"A thick stack of fabric lies on a long machine waiting to be compressed and cut into shapes. “It can automate a lot,” says Gregg Thompson cofounder of combat wear company Crye Precision, “but the volume has to be there.” Machines are not so great at design iteration, he says. A man standing next to the machine slices sheets of fabric by hand. He’s faster, says Thompson, and he can execute multiple designs without needing to be reprogrammed.
.
A shift is happening in manufacturing, bringing humans and machines closer together and making production more responsive to changing needs. The change is coming from companies that need flexible processes that allow their products to evolve with the needs of their customers. They also want their facilities not in industrial suburbs, but in amenity rich neighborhoods, so they can attract star talent. [Moi ici: Quase que não se pode pedir começo mais promissor do que este reforçar da esperança num lugar para o humano na produção em Mongo]
...
BNYDC’s grand vision is to create a campus that seamlessly integrates small and medium manufacturing operations with white collar offices, production studios, restaurants, coffee roasteries and shops, distilleries, grocery stores, bakeries, and rooftop gardens with edible greens.
...
The workflow will look something like this: a customer will call and ask for something, a designer will mock up a solution in its upstairs offices, that drawing will travel downstairs to production, once the part is made it will fly under the purview of a solutions architect who will vet whether the product actually suits the described need. Essentially, the company can iterate on customer needs much faster–in a single day. Previously, the company would have had to send parts back and forth between its manufacturing facility in Los Angeles and main headquarters in New York, a process that takes days if not weeks.
...
Factory workers are no longer just doing mindless production line work. The new manufacturing requires interacting with complicated robots and fixing computer systems. [Moi ici: Clubes de leitura e as ironias automação] “The manufacturing is smaller and cleaner and more sustainable and doesn’t pollute as much,”
...
The Navy Yard also recently opened its own high school [Moi ici: Até a escola que conhecemos, formatada para a revolução industrial e o Normalistão tem os dias contados] on campus called the Brooklyn Science Technology Engineering Arts and Mathematics Center (The STEAM Center for short). The school, developed in partnership with the Department of Education, gives qualifying junior and senior high school students from eight area schools the opportunity to spend half their day learning curriculum influenced by the equipment, jobs, and working cultures on display at the Navy Yard."

sábado, março 09, 2019

Mongo e automatização - tanta treta que se ouve

O meu parceiro das conversas oxigenadoras mandou-me este artigo, "Can There Be Too Much Automation?":
"Much of what you hear about automation focuses on the increased productivity that automation can bring to production lines. You hear about this a lot because it’s a true, measurable reality.
.
But is it possible that too much reliance on automation can hinder the overall productivity of a factory? [Moi ici: Um velho tema deste blogue, Mongo e automatização não jogam bem!!! Ver Lista de artigos abaixo]
...
At Mitsubishi Electric´s Kani manufacturing facility, which is part of the company’s Nagoya Works in Japan, the company found that, by bringing humans into work cells that were once 100 percent automated, the footprint occupied by the cell itself could be reduced by 84 percent.
...
In addition, Mitsubishi Electric notes that the introduction of human workers to previously automated assembly lines is helping the Kani factory react faster to changes in product demand.[Moi ici: Tão Mongo!!!]
...
The Kani factory produces motor starters and contactors for Mitsubishi Electric. The vast amount of production variations and possible configurations of these products—14,000—diluted the volumes of each particular product. This amount of variability, coupled with customer demands for even greater choice, highlighted the automation problem for Mitsubishi Electric. [Moi ici: Tão Mongo!!!]
.
According to the company, manual production at the Kani factory had given way to totally automated assembly lines, which were ideal for mass production with few product variations where high yields could be realized at high speed. But this required many individual components to be held in stock and ready for the manufacturing process; otherwise, the lines would not be able to run for any appreciable length of time.
.
In effect, it became difficult and uneconomical for the factory to produce its products in small batches—which just happens to be the very direction in which industry as whole is headed.
...
The company realized that, by restoring some human elements, it could reduce some of the manufacturing problems it was encountering.
...
Despite that fact that the new cells, featuring a combination of human workers and automation, cannot produce at the same volume and speed as the fully automated lines, the reduced size of the new cells means the company can deploy up to 6.3 cells in the same space once occupied by one cell. Mitsubishi Electric says this means that total productivity density for the facility is much higher due to three key factors: a wider variety of products can be manufactured in smaller batches; one stoppage does not halt the whole of production; and the total number of production lines has increased."
Como escrevo no postal de Fevereiro de 2018 na lista abaixo:
"Muitas vezes penso que as pessoas quando planeiam o futuro não fazem como Teseu no labirinto, não usam uma corda para unir o hoje com o futuro desejado. Por isso, usam lugares comuns. Por isso, não põem os pés no chão e testam a validade dos pressupostos que estão a assumir." 

  1. Estranhistão, autenticidade, imperfeição e automatização (Agosto de 2013) 
  2. Um mesmo processo automatizado é demasiado rígido para Mongo (Abril 2014)
  3. Mongo e a automatização... pois! (Fevereiro 2016)
  4. Beyond Lean (Agosto 2017)
  5. Seru (parte V) (Setembro 2017)
  6. Da normalização para a excepção (parte II) (Fevereiro de 2018)
  7. Coisa de loucos (Maio de 2018)
  8. O que protegerá Portugal dos robôs? (Outubro de 2018)
  9. Nem de propósito! (Dezembro de 2018)

quinta-feira, janeiro 31, 2019

Robôs e Mongo

O parceiro das conversas oxigenadoras mandou-me um texto interessante sobre o tema ada automatização, bem em linha com o que sempre aqui se escreveu sobre o problema da automatização quando choca com a variedade crescente de Mongo:


"Saw this one today...thought it was worth sharing...The theory is that beyond Level 3 automation, capital costs and technical complexity increase dramatically. Properly loading parts for humans is easy but can require delicate technologies to automate.
Similarly, transferring parts automatically may require robots or transfer lines. This may seem attractive on paper, but the complexity often reduces process reliability to below 70%.
In addition, what do you do with expensive, highly automated lines when demand changes? When demand falls off you are stuck with an underutilized piece of high-depreciation equipment. When demand rises above capacity you need to buy another expensive automated line.
Experience has shown that Level 3 Automation achieves many of the benefits of full automation but without the expense, procurement lead time, maintenance requirements, downtime and volume inflexibility.
Thus: “The Great Divide” Look for alternatives first."
Recuo a:

segunda-feira, janeiro 21, 2019

Também por isto sou um contrarian (parte II)

Parte I.

A propósito de "Robôs destroem 440 mil empregos na indústria e comércio até 2030" e do pormenor:
"Indústria, comércio, transportes, funções administrativas e de públicas e agricultura. Estão entre os sectores onde o impacto da automação na destruição de emprego mais se fará sentir."
Sorrio e vou buscar "Report: Retailers have zero clue what shoppers really want":
"Hey, retail executive. It’s very nice of you to suggest I speak with your robot, but no, I’ll pass. It looks like there is a fully functioning human standing in the corner of your shop. Would it really be too much trouble to speak with him instead?
...
I’m not the only one who feels like this. In a report that comes as a surprise to absolutely no one but overeager retail execs, 95% of consumers don’t want to talk to a robot when they are shopping, neither online nor in brick-and-mortar stores. And 86% have no desire for other shiny new technologies either, like artificial intelligence and virtual reality. I, for one, don’t want to pop into a store to quickly pick up that alpaca sweater I saw online, only to have some sort of weird headset shoved in my face.
...
The vast majority of retail executives believe that AI and VR will increase foot traffic and sales, but 48% of shoppers say these technologies will have zero impact on whether they visit a store, and only 14% say they will make a purchase because of these technologies. This also applies to online technologies like chatbots. Seventy-nine percent of retail execs believe that chatbots are meeting shopper’s needs by providing on-demand customer service, while 66% of consumers disagree, with many respondents noting that chatbots are, in fact, more damaging to the shopping experience than helpful."
 Até parece que a batota da interacção entre humanos passa por robôs?!?!?!?!

E recordo a economia das experiências, "The experience economy is booming, but it must benefit everyone":
"The only companies that will exist in 10 years’ time are those that create and nurture human experiences. This learning and growth will come from maximizing opportunities, including the reinvention of retail spaces, new models of engagement, and an understanding of experiences as perhaps the most important form of marketing."


sábado, janeiro 19, 2019

Contrarian!


A propósito de "Robôs e outras coisas que vale a pena discutir" e de "“Robôs” eliminam 1,1 milhões de empregos em Portugal até 2030, avisa estudo da CIP" fico com a habitual sensação de quem escreve estes textos e estes relatórios não vive no mesmo mundo que eu.

Esta semana em conversa com empresa metalomecânica senti alguma incomodidade quando abordamos o robô de soldadura. Adquirido como a última coca-cola no meio do deserto tem-se revelado um destruidor de produtividade. O mesmo tema que já tenho apanhado no calçado, quando tentam introduzir robôs.

Há dias citei aqui um texto sobre os robôs e a automação na Toyota, ""Anyone can buy robots" o pior é o resto":
"Only those robots that work really well and are cost-effective still have a chance of keeping their jobs at Toyota under Kawai. He explains: After looking at the robots that weld together the base of the Toyota Land Cruiser, he noticed that the welding seam was too wide and had a few defects. "I shut down the entire robot line and I said: 'We'll do it manually again,'" says Kawai. Unlike the robots, human workers could see where a groove to be welded was one millimeter wide and where it was only half a millimeter wide and could then react flexibly. "The use of welding wire alone has decreased by 10%," he says with pride."
O que encontro sistematicamente, no calçado e na metalomecânica, são empresas que compram robôs a pensar que só precisam de os instalar e, depois, só precisam de tirar as pessoas e a produtividade sobe.

Por exemplo, esquecem-se das matérias-primas... concluía a empresa metalomecânica. Continuamos a comprar o tubo onde sempre o compramos, só que agora com o robô... se um tubo vem ligeiramente ovalizado é logo um problema. O robô faz a soldadura, mas ela fica imperfeita e tem de ser corrigida por um soldador. Resultado, produção mais lenta e mantemos o humano. Se queremos "mais qualidade, mais rigor" no tubo, temos de meter alguém a fazer controlo da qualidade, a usar equipamento de medida mais sofisticado, e eventualmente a comprar matéria-prima mais cara.

O mesmo no calçado. Trabalham com peles. Pele é uma matéria-prima natural, quem garante espessuras dentro de limites exigidos por um robô? Recomendo a leitura de "Os Robots na Industria do Calçado. Muitas vantagens, algumas dificuldades."

Outro factor que é esquecido nas análises, lá de cima e nas outras, é que nós estamos a fugir do século XX, nós estamos a fugir do Normalistão e a embrenharmo-nos em Mongo, o Estranhistão. Recordo:

"In principle, the production of virtually any component or assembly operation could be robotized and moved to high-wage countries—but only so long as demand is great enough, and design specifications stable enough, to justify huge scale and hundreds of millions, if not billions, in upfront investments." [Moi ici: Mongo mina logo dois dos pressupostos - tamanho da procura e estabilidade. Mongo é terra de tribos e de modas, terra de velocidade e instabilidade. Terra de gente que não quer ser tratada como plancton. Terra de gente que não se sente atraída pela suckiness.]


"People will be surprised - “[the use of robots] won’t be as disruptive as the hype today would suggest,” he continues.
.
“The more a robot can do, the more it will cost – humans should be able to still be less expensive than robots. Plus robots for the foreseeable future will have to specialise, and we humans don’t – we’re more flexible.”

"Toyota has found that the race to reduce the human element can end up making processes less efficient."



Por que é que esta corrente de pensamento não é mais divulgada? Vende menos, por ser mais optimista e menos alarmista?



quinta-feira, janeiro 10, 2019

"Anyone can buy robots" o pior é o resto

O meu parceiro das conversas oxigenadoras chamou-me a atenção para "Automotive manufacturing requires human innovation":
""We always start with manual work," says Kawai. "The automation process will progress, generally speaking. But when we use robots, they'll be trained by people who know what they're doing." In this simple sentence lies an entire philosophy: Kawai is throwing out the robots and replacing them with human workers – as often as he can.
...
While everyone is talking about how robots will very soon be taking away our jobs, Toyota is shaking things up and charting out its own course. Even before Kawai took the lead at Toyota in 2015, the company's automation program was changing direction in many areas of production. Now, from the forging shop to chassis assembly and the paint shop, human beings are back at work again in places were robots were recently employed.
...
If Toyota has an automation ratio that is barely greater than 10 years ago – sources put the machine-to-human ratio at less than 10% – this has financial reasons as well as the pursuit of quality. Kawai says: "If you immediately start with automation in production lines, you get highly complex systems that are also very expensive and often stand idle."
E este pormaior, para os parolos que acham que o futuro é a automação sem mais reflectirem um bocado:
"Anyone can buy robots, but the knowledge to use them effectively is best acquired by oneself."
E tanta gente a acreditar que o trabalho humano não tem futuro... pois, gente que não percebe Mongo.

Como não recordar "Nem de propósito".

sábado, janeiro 05, 2019

BINGO!! (parte II)

Parte I.

Ainda de "Value Never Actually Disappears, It Just Shifts From One Place To Another" sublinho outro tema clássico aqui no blogue:
"You Can’t Compete With A Robot By Acting Like OneThe future is always hard to predict. While it was easy to see that Amazon posed a real problem for large chain bookstores like Barnes & Noble and Borders, it was much less obvious that small independent bookstores would thrive. In much the same way, few saw that ten years after the launch of the Kindle that paper books would surge amid a decline in e-books.
.
The one overriding trend over the past 50 years or so is that the future is always more human. In Dan Schawbel’s new book, Back to Human, the author finds that the antidote for our overly automated age is deeper personal relationships. Things like trust, empathy and caring can’t be automated or outsourced.
.
There are some things a machine will never do. It will never strike out in a little league game, have its heart broken or see its child born. That makes it hard — impossible really — for a machine ever to work effectively with humans as a real person would. The work of humans is increasingly to work with other humans to design work for machines.
.
That why perhaps the biggest shift in value is from cognitive to social skills. [Moi ici: Engraçado que cada vez mais dou comigo a pensar que um número crescente de artigos em revistas ditas de gestão ocupam o seu espaço com temas que a minha mãe, ou a catequese, ou o pertencer a uma associação, ou o pertencer a um grupo de colegas de rua me ensinaram e que parece que agora estão em falta] The high paying jobs today have less to do with the ability to retain facts or manipulate numbers (we now use a computer for those things), but require more deep collaboration, teamwork and emotional intelligence."
O quanto os gigantes gostariam que o factor humano fosse removido da equação... mas a imperfeição é cool, e a desautomatização está a crescer.





quinta-feira, março 22, 2018

Acerca da automatização

Ontem isto:

Hoje isto:
"When you ask Amazon’s Alexa to reserve you a table at a restaurant you name, its voice recognition system, made very accurate by machine learning, saves you the time of entering a request in Open Table’s reservation system. But Alexa doesn’t know what a restaurant is or what eating is. If you asked it to book you a table for two at 6 p.m. at the Mayo Clinic, it would try."
Num mundo onde há cada vez mais variedade, num mundo cada vez mais polarizado, as empresas grandes do preço mais baixo, as "amazons", são conhecidas e tudo o que fazem é descrito e narrado como o último grito na gestão. As empresas do outro extremo não se devem iludir, o que serve para as do outro extremo não pode servir para elas. Se servir para elas estão tramadas. Até algumas grandes já perceberam as limitações da automação.

BTW, hoje de manhã na rádio ouvi notícias sobre as grandes tiradas dos políticos sobre os dinheiro do próximo envelope financeiro comunitário... e pensei na quantidade de projectos de milhões às moscas porque era dinheiro fácil de obter via Portugal 2020 e está lá nas fábricas, não há ilegalidade nenhuma, mas não trabalham porque não são úteis.

Trecho retirado de "The Great AI Paradox"

quinta-feira, fevereiro 22, 2018

Cuidado ao tirar os humanos da equação

Há dias apanhei esta frase:
"relationships are not zero sum games"
Bem na linha do que aqui escrevemos há muitos anos acerca do poder da interacção e da co-criação. Por isso, há muito que defendo que o futuro de valor acrescentado passa mais pela interacção do que pela automatização.

Agora, em "The Parts of Customer Service That Should Never Be Automated" encontro:
"the economics of service automation aren’t universally rosy. When a nationwide retail bank introduced online banking, customers who adopted it increased their total transaction volume and began visiting and calling the bank more, increasing costs and decreasing overall profitability. Similar dynamics can be observed in health care. Patients who adopted e-visits, for example, actually began showing up at the doctor’s office twice as often. One explanation for this pattern is that current technology is functionally limited, requiring people to seek out in-person help in addition to using automated services. But as innovation progresses, functional limitations are bound to fall by the wayside.
.
Another explanation is that humans are inherently social creatures who get emotional value from seeing and interacting with one another. Research shows that taking away the opportunity for this kind of connection can undermine service performance. In one study, my colleagues and I found that when banking customers used the ATM more and the teller less, their overall level of satisfaction with the bank went down."
Recordo esta opinião:

E o comentário que fiz na altura:
"Quem não aposta no "cheaper" e no "cost", aposta na interacção, aposta na co-criação, aposta noutro mindset... eu diria, "Every visit customers have to make are an opportunity for interaction and co-creation""
E voltando ao texto do artigo:
"the deck is stacked against automation in several important ways:
.
1. Service can be emotional; technology cannot.
...
Automating sympathy is certainly cheaper than having a human employee comfort the bereaved, but the tradeoff can come across as disingenuous and is unlikely to be sustainable.
...
2. We still prefer having people help solve our problems. In many ways, the capacity and computational power of technology far outstrips our own. ... Nevertheless, when we’re looking for creative solutions to service problems, we still seek out other humans. If we get stuck, if there’s ambiguity in the information, or if we need help making a purchase decision, we still opt for a person.
.
3. Less work for employees often means more work for customers."

quarta-feira, outubro 25, 2017

Acelerar o time to market

"The impetus to use automation is not just about costs: it is also trying to keep up with consumers. Across the board, the most successful retailers are now those with a constant stream of new products to meet rapidly changing tastes and shopping habits. Yet companies have been slow to adapt footwear, with its more complicated manufacturing process, to so-called fast fashion trends — at least until now.
...
Traditional shoe production has required as many as 200 different pieces across 10 sizes, often cut and glued together by hand. The new manufacturing process being developed by Flex has introduced two ideas once thought impossible: the gluing process has been automated and lasers are used to cut the Flyknit material.
Lead times in the shoe industry once ran to several months: Flex has promised to help Nike speed up lead times, which can be three to four weeks for a customised pair of sneakers.
...
Moving production closer to its key markets will help satisfy some of that demand."
Trechos retirados de "Robotics in the running for Nike’s factories of the future"

terça-feira, setembro 05, 2017

Automatização e têxtil

Excelente motivo de reflexão: "This T-Shirt Sewing Robot Could Radically Shift The Apparel Industry":
"“People buy 11 billion T-shirts a year. That’s an interesting market where automation makes sense, where our robots make sense, because our robots produce a very high volume of product.”
...
The company is focused first on T-shirts and jeans because the robots’ strength is producing huge quantities of clothing. “People buy 11 billion T-shirts a year,” says Rajan. “That’s an interesting market where automation makes sense, where our robots make sense, because our robots produce a very high volume of product.”
...
While the technology is still developing, and will eventually make more complicated items of clothing, Rajan believes that some higher-end clothing will always be made by humans.
...
the technology could have a positive impact in countries with large garment manufacturing industries. Workers might shift into doing more artisanal work, at higher wages.
...
In practice, the transition is likely to be messier. Sewing robots may only threaten certain jobs in developing countries; the machines are expensive (the company won’t disclose the cost) and the company is initially only selling them in the U.S., where manufacturers can save money on the total process because they can avoid factors such as higher industrial electricity bills in some other countries, and can benefit from a “Made in USA” label." 

quarta-feira, agosto 09, 2017

"Automation is a game of large numbers"

"Jobs are increasingly viewed as undifferentiated and interchangeable across humans and machines — the very definition of a commodity.
...
Outsourcing — exchanging internal employees for external ones, often offshore — was a big step toward commoditization for many companies.[Moi ici: Mentalidade anglo-saxónica no seu pior, o foco nos custos acima de tudo e, passar ao lado das oportunidades dos nichos e da proximidade]
...
Just as there are low-value and high-value commodities, there are low-value and high-value commoditized jobs.
...
For many jobs, the value is driven less by their intrinsic worth but rather by market demand. A recent Bloomberg Businessweek visual analytic suggests that jobs that disappeared in the first four months of 2017 compared with the same period in 2016 were not lost to automation, but were lost because fewer customers wanted to buy the products and services they produce.
...
For many organizations today, the next big driver of job commoditization is automation driven by smart machines. Simply put, if a job is viewed as a commodity, it won’t be long before it is automated. My research on automation through artificial intelligence (AI) or cognitive technologies suggests that if a job can be outsourced, many of the tasks typically performed by the jobholder can probably be automated — even by relatively “dumb” technologies like robotic process automation. Many global outsourcers are working desperately to create their own automation capabilities that could replace human jobs with machines.
...
The key for financial professionals and other workers whose jobs have traditionally seemed safe is to make themselves less commodity-like. Automation is a game of large numbers, and it’s not economical to automate unique activities. [Moi ici: Mongo é um mundo de "large numbers"?] As long as human workers’ capabilities are differentiated from machines’ capabilities, then machines can’t easily replace them — and few organizations will be tempted to automate that niche."

Trechos retirados de "When Jobs Become Commodities"

segunda-feira, agosto 07, 2017

Beyond Lean (parte II)


Na parte I procurámos demonstrar que o futuro será muito mais do que a automatização. A explosão de tribos de Mongo requer estruturas produtivas com um ADN diferente do que aquele que o século XX nos legou.

A evolução tecnológica vai trazer, também, a democratização da produção, a redução de barreiras à entrada e, por isso, a explosão no número de pequenas empresas life-style business.

Em paralelo a esta evolução, que vai sugar os mais apaixonados para uma nova Idade de Ouro de artesãos do século XXI, teremos a reacção das empresas grandes no seu combate final pelo domínio da decrescente fatia de mercado que representa os que continuam dentro da caixa e optam pelo preço como o critério prioritário de compra. Essas continuarão a optar pela automatização como forma de reduzir custos.

E em paralelo com as duas correntes anteriores teremos uma terceira impulsionada pela demografia e tão bem ilustrada em "Rise of the machines".

Até que ponto Portugal vai ter de ser pioneiro nesta terceira via?

Demografia, marxismo social e a atracção pela emigração, são uma combinação tremenda que só agora começa a ter o seu impacte. Recordar:

sábado, agosto 05, 2017

Beyond Lean

Há anos que escrevo aqui sobre o advento de Mongo e o consequente impacte na dança entre produção e consumo:

  • mais tribos;
  • mais nichos;
  • séries mais pequenas;
  • mais flexibilidade;
  • mais rapidez; 
  • mais variedade;
  • mais diferenciação;
Em paralelo há anos que se lê com cada vez mais frequência sobre a automatização da produção.

Em Mongo, a produção é muito diferente da do paradigma do século XX com séries longas e planeamento da produção feito com muita antecedência. Em Mongo o planeamento da produção é feito cada vez mais em cima e é mais volátil. Como é que a automatização e as organizações-cidade lidam com Mongo?
"With improvements in living standards and a transformation in people’s ideas of consumption, much of the current electronics manufacturing industry is confronted with market demands characterized by variety and volume fluctuation. Manufacturing system flexibility is useful to address such fluctuated market demands.
...
Seru production has been called beyond lean in Japan and can be considered to be an ideal manufacturing mode to realize mass customization
...
seru production relies on low-cost automation and has little automation. When reconfiguring a conveyor assembly line into serus, expensive large automated equipment is substituted with simple-structure equipment with similar functions. The reconstructed equipment can be easily duplicated and modified at a low cost, so as to avoid equipment-sharing conflicts among multiple serus and reduce investment in equipment.
...
Factories that produce multiple electronics product types in small-lot batches tend to adopt seru production. Compared to mass production, which displays its superiority in the case of a narrow range of product types with high product volumes, seru production would be affected by low efficiency and high cost in such an environment.
...
Using highly automated production systems, mass production factories can attain high production efficiency. However, they usually achieve low production flexibility. As both the variety and volume fluctuations of market demands increase, mass production factories may need to reconfigure their traditional conveyor assembly lines for their survival and development.
...
Seru production is human-centered manufacturing. Multiskilled operators are important resources to implement seru production, more important than in mass production. The equipment used in seru production is simple and not automated. The effect and influence of equipment on the performance of seru production systems is less than that on mass production systems. Accordingly, a practical production planning system for seru production should consider multiskilled operators more than equipment. In a dynamically changing manufacturing environment, a dynamic production planning system is needed"
Anónimo da província mas muito à frente:

Cuidado com os media, desconfie sempre!


Trechos retirados de "An implementation framework for seru production" publicado por Intl. Trans. in Op. Res. 00 (2013) 1–19