That's what I observed as well. Finding that generic but valuable use-case is quite tough. Selling it is on a totally different level. The impact it can have should be well worth it though
Not sure. I know that the conglomerate Outokumpu once had a pilot plant with high-Tc superconducting magnets deployed in the Amazon region somewhere ('Ice in the jungle'? Hah, hold my beer, try liquid-helium cooled superconducting magnets in the jungle).
I know because a colleague joined the company, and one of his first assignments was to diagnose unexpectedly large helium losses. A quick FFT later of the recorded Dewar flask levels revealed a 24-hour periodicity, and further analysis found that for a couple of hours every day, the full heat of the tropical sun was finding its way to the tin roof of the mine shed housing the superconducting magnet systems.
I haven't heard anything more about the technology or its economics/scaling since then. But I also certainly haven't heard anything more about its use with landfill-derived feedstocks, which seems like a reasonable move.
My thoughts exactly. Knowledge transfer in manufacturing / industrial environments is something that I'm working on.
- Language models / NLP applications for processing large amount of technical text data (SOP, documentation, technical data, machine text logs, voice to text, video data processing for speeding up corrective action, training, onboarding and highlighting areas of improvement / bottlenecks), digitising documents and extracting failure reasons / equipment names / spare parts / processes involved and making associations between them for pareto analysis, better search or process improvement recommendations
- Recommending the next steps to fix something / remote intervention / do something etc. Lowering the expertise threshold required for technicians, electricians, mechanics or reliability engineers to be effective.
- Enabling operators to become data scientists by enabling to train AI models via their day to day activities / analysis. Building better UX in general and providing simple tools that even a toddler could use.
I'm a partner in a factory and I believe this is an incredibly important area, and the requirements are fairly different than normal "just put it on Confluence" workplaces, in a way that most tech people don't understand and usually completely miss the mark when they're doing product dev.
- Your team is out on the floor. Their hands have grease on them. Using tablets sounds great until you're trying to use it with a glove on it, or your hands are dirty, and it's hard to get grease off tablets. But they need the info out on the floor. Also, it can be noisy on the floor.
- The team tends to be very visual. They don't like tapping on computers a lot. Literacy ranges from pretty good to kinda OK. Sometimes they refuse to get (or wear) reading glasses for whatever personal reason.
- They're working on proprietary hardware, but technicians with the right knowledge are not nearby to come in and look at it. You really need to be able to see the issues visually. Sometimes even hear them. AR might be interesting here. (I spend $10k to fly a tech out for a few days to look at a machine. The bigger issue is that I lose $10k a day from one machine being down, and a tech might not be available to fly out for a week.)
- Predictive maintenance. The fancy sensors and whatnot mostly don't work. Tech people try it in a clean, quiet office and it works, and they can raise money on it from clueless VCs, so money keeps getting set on fire with smart AI machine learning magic motor sensor companies.
- Preventative maintenance. How to schedule, how to verify it was done, how to check whether it revealed an issue that needs a follow-up. Getting people to do it, and verify it was done, can be a challenge, but there are huge returns to preventative maintenance (for example: checking gearbox oil levels, verifying lubrication line function, checking valve temperatures.)
- Diagnosing machine problems. Using prior problem documentation helps team members see most likely issues. But many of these people don't really want to sort through a database of prior similar issues because they "know" what the problem is. How do you provide this information to them in a way that feels more approachable to them?
I could go on forever. Manufacturing is an interesting environment because downtime is usually hundreds to tens of thousands of dollars of hard cost per hour, depending on the operation, and they will spend quite a bit of money to stop it from going down, but culturally there's a vast gulf between the white collar SF tech bros and what actually happens in manufacturing plants, so innovation tends to be more limited.
Predictive/preventive maintenance is actually a big thrust behind my current company, Dials.
HOAs, which we serve, are run by busy volunteers, yet expected to perform almost insane financial gymnastics, planning 30 years of major component replacement, e.g. common area roofs, piping, asphalt resurfacing. This involves (a) estimating each component's lifetime (total and remaining), (b) getting a cost estimate, and (c) coming up with a plan to spread paying for it out over however many years before it's needed, breaking that up between the units in the HOA, and collecting the funds, month after month.
People blame cultural issues ("people won't pay for maintenance") or "laziness" but the truth is, it's just too damn hard to do predictive/preventative without a very accurate inventory of what you have. You need to get all of this into a cloud environment, and then somehow expose it so that either internal staff or external vendors (more common) can see exactly what you have, bid on fixing it, and track status and work in a fine-grained way.
Our ultimate goal is doing the entire inventory automatically using computer vision (partner and I used to work in self-driving) and having enough data around that we can price and estimate everything accurately.
Nobody wants to pay for this as a standalone product so we just decided to build a payment collection product (for monthly dues), start with that, and build it up. It's going pretty well and we'd love to get more people on it. Email's in my profile in case you want to chat
Tracking stuff is hard. I wonder why QR codes won't work in this case, or something similar, or super basic otherwise / stickers or codes at first. Might be more annoying to generate and maintain them initially. CV could work really well to keep track of inspection steps as well, or to recommend what you should do next, and how to do it
We considered this, but it's another step. What I'm talking about is going to be hard and take a while, but feels like the "endgame" for how this is going to be done--automated, done with phones, no extra work.
It sounds really interesting to work in a manufacturing plant for a year or two in order to empathize with the industry and learn how to blend in software in a way that actually solves problems like you describe. Or, generally, to penetrate areas where technology solutions don't apply obviously. I wonder how you'd set that kind of arrangement up. If you could design a company around displacing a few cofounders for a few years where the product research is hands on, on the ground, doing the job, I bet there are many people who would be interested in this type of setup. I agree the software industry does way too much "solve for ourselves first" type of product development and it's really discouraging.
I think a key point that people fail to remember or marketing just oversells, is that there is no silver bullet for all the problems, especially for an industry that has so much history, precedent and inertia. People want to try and solve (and from the other side, want perfect full solutions) all problems at once, whilst in reality small improvements in key areas are probably the 80/20 that is needed to bring business value. I think continuous feedback and good "translators" would be key for any product in those industries. Manufacturing people are busy and will tell you what the surface level pain point is, but they don't have the time or maybe don't have the idea fully thought out on what the underlying problem/goal is.
After typing up all that, I realise that most of this is applicable to every industry.
Spot on, many of these challenges are common across the board - from my father's plant to Pfizer and others I got the chance to work with. There is however a massive talent gap when it comes to high quality software / ML people in these industries as well. It's tough to get experts to generate quality data and 'recipes' for others to follow when their KPIs are not aligned. Maintenance and reliability don't seem to be sexy enough areas for management to invest in, especially if the value proposition is anecdotal at best. Would be great to chat about your approach for solving some of the above
I would not just knowledge transfer, but knowledge organization. We have so many different ways to represent knowledge, but it is very hard to access it or know where to look.
I think better training and investigation into best practices of organizing knowledge would benefit all industries.
Former librarian here, now at a tech co. This is exactly the domain of information science, and its a salutary tale in two industries talking past one another. Librarians have deep training in the science, philosophy and psychology of information storage and retrieval. Most of the time you think of things like Dewey numbers on library books buts its much more than that. At the dawn of the second internet age (circa 1991, think gopher, WAIS, Archie and a nascent thing called the web) there was a boatload of discussion around what this Internet thing would mean for information.
Then, the tech bros arrived and after a few abortive attempts to catalog things for themselves (webrings, portals, yahoo) the industry collectively shrugged and decided to ignore the problem, assuming that a search engine would always be able to pluck your favorite needle out of the haystack.
Except that, today, it cant. Intranets are essentially corporate graveyards of content. The public web is a webring of 7 or 8 megasites that vacuum up all searches and make it all but impossible to break out of their domain. That article you read in 2005 about XYZ? Forget it, you're never finding that with Google.
Do you have any pointers to things worth reading? I've always sworn if I started a tech company one of the first 10 employees would be a librarian because it needs to be someone's job to organize the information, and you're right the automatic systems for doing it are horrible.
Search isn't enough because search doesn't help you if the philosophy behind how the information is organized doesn't make sense -- if what you need is spread around between 100 slack messages, emails, and unconnected unmaintained wiki pages. You need someone (or ideally a team) whose job it is to one one hand organize that information themselves, and on the other hand create a framework so it's easy for the non-librarians to put things in the right place.
But right now the majority of tech companies are like libraries without librarians, the patrons are just wandering around sticking the books on random shelves and wondering why nobody can ever find anything.
I'm curious what your thoughts about Cory Doctorow's Metacrap [1] essay which I think summarized a lot of the problems with the semantic, informational organization approaches of the early-mid web. Are you also familiar with research in informational sciences these days?
Yes, please tell us your recommendations for informational retrieval / taxonomy systems - what are the current best practices for the different mediums?
Have you considered going into SEO? Combining SEO expertise w information science sounds like market dynamite. There is probably a 5k/month blog in simply applying information science concepts to SEO in practice, to say nothing of the consulting gigs, etc.
I'd agree with knowledge organization. It's either you have to root through academic texts or try and navigate the spammy internet with no really happy medium. It's almost like there's a complete lack of quality middle ground information. It's either total SEO garbage or very low quality entry level information or incredibly specific/dense academic content and the middle ground is missing.
Definitely, transfer can only happen when knowledge is organised and understandable by a variety of stakeholders, with different backgrounds (education, languages spoken, years of expertise)
Lowering the expertise threshold required for technicians, electricians, mechanics or reliability engineers to be effective.
This is a really interesting application I hadn't considered before. Having lots of blue-collar family, helping new members of the trades upskill fast would take a considerable load off that workforce.
These things are all on the radar of "innovation" types. I don't mean to say they're not interesting, but in the area of applied ML all this stuff is basically as mainstream as it comes (despite being unsupported by any actual research advances).
I'm starting a new job doing exactly these things in order to reduce the carbon intensiveness of heavy industry, specifically cement production. I'm hyped because I think the technical challenges aren't too daunting, and the prize is huge.
I've started looking into CO2e reduction techniques as well. Would be great to discuss. Working with a client in the food space who is doing this just to learn more
Quality, reproducibility, and precision necessarily require removing the human.
If it's something "artisanal", that not necessarily true, but even then, intentional "mistakes" can be added [1]. Having humans for the sake of having humans isn't a charity that non-luxury businesses can support (à la Snow Crash). It'll have to be something that governments subsidize or enforce tyrannically.
Fewer and fewer people are interested in manufacturing jobs, especially the less glamorous ones. Large manufacturers are having a hard time using analytics and more advanced systems because of qualified labour shortages. I've spoken to manufacturers whose technicians can't even write or follow instructions correctly. Sometimes, sending 10 technicians to inspect an asset would results in 10 different opinions about possible issues / failures. All of these could lead to lower quality product and increased unscheduled downtimes, lower revenues etc etc. But, it is definitely important to still allow people to use their brains and come up with better options
I volunteer for rural development organizations providing the kinds of services that suburbs would call dept of water or forestry. This point is very important to reliably onboard volunteers.
If we need excavation, our worst-case scenario is that that we need excavation by someone who also knows
Many companies store huge amounts of documents, but every team does it in its own way. One template can quickly result in thousands of variations. If robust documentation principles are not used from the very beginning (checklists / reduction in free text, visual indications, etc), it will be a nightmare to make sense of that data afterwards. Also, there is no value in generating large amounts of text data unless you can easily scan it and retrieve the information of interest
What I wrote wasn't simply a poorly specified product requirement document ;) Instead it was a general idea.
Better and more granular documentation in a form that can be interpreted by humans and also machine readable would be a desirable outcome for any system. Especially true for systems in which the builders and operators are being replaced or EOLed
I was being a bit tongue-in-cheek because "digital transformation" has become an overused marketing term. But I think the core of it is valid -- a company has an inefficient process due to lack of technical expertise or whatever and you help them fix it.
There's multiple parts to this field of work: networking to find leads, doing discovery to understand a potential client's problem, formulating a technical solution, creating/negotiating contracts, and implementing the solution. My experience in this area was at a company that was large enough so that these pieces were split into different roles within our organization, and I was mostly on the tech/solution implementation side.
Sometimes they have in-house teams that are slowly switching things over. Are there any good blogs/articles/books on starting or running such a consultancy?
Interesting, that is my experience at Pfizer as well, where I lead projects end-to-end, from problem discovery to solution deployment and I pretty much did everything, from talking to coding
I used to work in manufacturing as a process engineer (now a dev) and this is fascinating to learn about. Do you have any articles on your work or any use cases where I can learn more?
[EU/UK/US] Building platform where hardware startups/integrators can build and scale their supply chains, on-demand, and go from prototyping to mass production in days. Looking for the crazy ones who are hungry and top of their game.
It is important to work with someone that you can trust. Someone that can push through hell together with you and not abandon everything at the smallest bump. It takes a strong character, ambition and sense of urgency. When it comes to non-technical / business people, it might be counter-intuitive, but having attended a good university or having worked for a good company can indicate high intelligence and not entrepreneurial drive. This being said, you can still collaborate with technical people that are comfortable with selling the vision, product and services, can hustle and are reliable.
Also, be careful of non-compete agreements by joining other start-ups.
There are several ways in which you can find good co-founders:
1. Don’t prioritise finding a co-founders, just start building something and reach out to people. They might love what you’re doing and come on-board.
2. Get recommendations via your network about people who want to build a company and have complementary skills / similar goals.
3. Increase your visibility, publish your thoughts, network and debate about the future - randomness / serendipity is a wondrous force
4. Accelerators such as Antler and EF - as a last resort.
I’m also looking for smart, relentless people that want to build a better future for generations to come. Don’t hesitate to reach out via linkedin - username: tudordragos
^^THIS! Trust is THE most important factor. Treat this like finding someone to marry, because (sadly) there's a decent chance your startup could outlast your marriage. (Pay attention to that, BTW! Your family is really important, and unless you've already made a huge mistake, no company is worth losing them over.) Any good/honest partner will be willing to tie equity to milestones - never just toss a big chunk of equity to your CEO/cofounder just because others are stupid enough to do that. (Keep in mind the dictum that in most all startup cos., your equity is simultaneously worthless and the only thing of value the company has, other than your team which is usually there because of the potential for equity. As soon as you're funded, pay people fairly - not top of scale, but enough to keep them happy and loving the job.)
And yes, FWIW, I've been on (and been burned on) both sides of these deals, both as a CTO looking for a good CEO to raise money and get us to market (lost my a$$) and as a contract CEO hired to get a company off the ground. The ONLY thing that matters is the people themselves, and more importantly, their character, trustworthiness, and insistence on being fully upright in all things.
It's much harder to find these people than you might think, and it can be really hard to get rid of a bad one if you screw up. (As the Russians say, "Doveryai, no proveryai - Trust, but verify!")
Follow https://understandingrelationships.com/life-coaching-service... - read free book 10-20 times :)
Read 'No more Mr. Nice Guy'.
Invest in relationship knowledge as much as you do in other topics of interest.