[
  {
    "start": 0.06,
    "end": 3.64,
    "text": "This podcast is supported by Siemens, your partner for industrial"
  },
  {
    "start": 3.72,
    "end": 4.34,
    "text": "grade AI."
  },
  {
    "start": 16.9,
    "end": 20.52,
    "text": "Hello everybody and welcome to a new episode of our Industrial AI Podcast."
  },
  {
    "start": 21.02,
    "end": 23.3,
    "text": "My name is Robert Wieber And it's the pleasure to talk to"
  },
  {
    "start": 23.8,
    "end": 27.54,
    "text": "Peter Sieberg working on The Road again."
  },
  {
    "start": 27.74,
    "end": 31.7,
    "text": "Good morning good afternoon Good evening To all you dear listeners."
  },
  {
    "start": 32.34,
    "end": 33.3,
    "text": "Good morning, Peter!"
  },
  {
    "start": 33.38,
    "end": 35.76,
    "text": "So I'm also working on the road again."
  },
  {
    "start": 35.94,
    "end": 44.32,
    "text": "I am sitting here in Lech-Am-Arberg and i've never seen so much snow in Easter holidays like this."
  },
  {
    "start": 44.82,
    "end": 45.62,
    "text": "it's unbelievable."
  },
  {
    "start": 45.68,
    "end": 48.26,
    "text": "its snowing since three days or something."
  },
  {
    "start": 48.62,
    "end": 54.24,
    "text": "yeah maybe that is the reason that I changed my original plan."
  },
  {
    "start": 54.36,
    "end": 62.22,
    "text": "we were going to be an ex Yugoslavia at Balkan changed."
  },
  {
    "start": 62.4,
    "end": 66.96,
    "text": "And we're now in Spain, and there's so much sun I've never seen."
  },
  {
    "start": 67.06,
    "end": 67.14,
    "text": "So"
  },
  {
    "start": 68.12,
    "end": 69.08,
    "text": "you have never seen so much?"
  },
  {
    "start": 69.18,
    "end": 73.26,
    "text": "That's amazing Peter that really."
  },
  {
    "start": 73.46,
    "end": 76.52,
    "text": "so i checked our location for summer event."
  },
  {
    "start": 76.68,
    "end": 78.5,
    "text": "AI in the Alps always check."
  },
  {
    "start": 78.98,
    "end": 80.32,
    "text": "very good looks."
  },
  {
    "start": 81.7,
    "end": 85.22,
    "text": "everything is fine a lot of snow but somehow We won't Have so Much Snow."
  },
  {
    "start": 85.8,
    "end": 86.42,
    "text": "Sometimes"
  },
  {
    "start": 86.5,
    "end": 88.48,
    "text": "we see it on the very top of demand, right?"
  },
  {
    "start": 88.54,
    "end": 89.46,
    "text": "Still a little"
  },
  {
    "start": 89.62,
    "end": 89.8,
    "text": "bit."
  },
  {
    "start": 90.94,
    "end": 95.38,
    "text": "So let's talk quickly about AI and the apps because... We have three amazing guests!"
  },
  {
    "start": 95.48,
    "end": 96.84,
    "text": "We are Robert Jositsch."
  },
  {
    "start": 97.08,
    "end": 103.0,
    "text": "He is CEO of Schwarz And you will talk to us a little more about sovereignty and digital strategies."
  },
  {
    "start": 103.94,
    "end": 111.2,
    "text": "Then we have SEP coming talking about memory and the new role of memory when it comes to AI."
  },
  {
    "start": 111.58,
    "end": 116.74,
    "text": "That's very interesting because ZEP is working on something new, when it come to memory approaches..."
  },
  {
    "start": 117.36,
    "end": 120.02,
    "text": "And you mean a ZEP Hochreiter?"
  },
  {
    "start": 120.12,
    "end": 130.02,
    "text": "The one and only inventor of LSDM and lately XLSDM with its theme around NXAI which I'm gonna talk And"
  },
  {
    "start": 131.42,
    "end": 132.46,
    "text": "the first one, of course."
  },
  {
    "start": 132.52,
    "end": 139.48,
    "text": "The one and only Marco Huber from Fraunhofer again about use cases industrial AI use cases."
  },
  {
    "start": 139.86,
    "end": 141.16,
    "text": "ask me anything Marco."
  },
  {
    "start": 141.48,
    "end": 142.4,
    "text": "yeah right so this has"
  },
  {
    "start": 142.46,
    "end": 144.04,
    "text": "been doing that for a very beginning."
  },
  {
    "start": 144.24,
    "end": 145.68,
    "text": "now in version number."
  },
  {
    "start": 145.9,
    "end": 148.08,
    "text": "I don't know version five i think or what?"
  },
  {
    "start": 148.48,
    "end": 149.12,
    "text": "Five six!"
  },
  {
    "start": 149.36,
    "end": 149.86,
    "text": "I'm not sure."
  },
  {
    "start": 149.98,
    "end": 150.1,
    "text": "yeah"
  },
  {
    "start": 150.44,
    "end": 151.48,
    "text": "Yeah totally allright."
  },
  {
    "start": 151.96,
    "end": 153.3,
    "text": "and the First One is same."
  },
  {
    "start": 153.4,
    "end": 161.2,
    "text": "when we talk About the Schwarz which means black In German the schwarz group People know that or do not know it at all."
  },
  {
    "start": 161.3,
    "end": 162.62,
    "text": "I think the most people know That."
  },
  {
    "start": 162.78,
    "end": 163.46,
    "text": "yeah, okay"
  },
  {
    "start": 163.64,
    "end": 164.08,
    "text": "very good"
  },
  {
    "start": 165.22,
    "end": 166.98,
    "text": "The big German company Lidl."
  },
  {
    "start": 167.54,
    "end": 167.74,
    "text": "Mr."
  },
  {
    "start": 167.78,
    "end": 173.68,
    "text": "Schwarz is the owner of the Lidle and Kaufland And they also have a digital company."
  },
  {
    "start": 174.6,
    "end": 176.62,
    "text": "Shall i continue there with you?"
  },
  {
    "start": 177.26,
    "end": 195.82,
    "text": "Thanks for the update looking forward to Very much again in a couple months from now because as i said sepp and team around him The guys have a new paper, NXAI folks together with folks from the University of Linz Johannes Kepler."
  },
  {
    "start": 196.16,
    "end": 200.88,
    "text": "It's called Effective Distillation to Hybrid XLSTM Architecture."
  },
  {
    "start": 200.92,
    "end": 204.22,
    "text": "not sure if that is going to be part of the topic you just referred too."
  },
  {
    "start": 205.18,
    "end": 208.1,
    "text": "it's not a separate topic but there are lists of names."
  },
  {
    "start": 208.14,
    "end": 220.94,
    "text": "I always like to call them out because they're people who work very hard Lukas Hautzenberger and Niklas Schmidigan Thomas Schmit Anna-Marie Roberta Hartl, David Stapp Peter Jan Hutt, that sounds Dutch."
  },
  {
    "start": 221.6,
    "end": 229.36,
    "text": "Maximilian Beck which I know you know most of them all of them who had the pleasure to announce him for a keynote two years ago."
  },
  {
    "start": 229.64,
    "end": 241.54,
    "text": "Sebastian Berg he's been with us actually at our conferences a couple times and then Günter Bauer and Zeb Holger both the NXAI and professors at Johannes Kepler."
  },
  {
    "start": 241.62,
    "end": 259.26,
    "text": "now they talk in their paper about energy effective and cost-effective replacements for transformer based LLMs which is I believe that's what we know them for but this time by means of distillation now."
  },
  {
    "start": 259.399,
    "end": 278.6,
    "text": "We've heard many times about distillation, i just had to make sure that in the context of language models, this relations is about model compression where a small compact model and that's typically being called The student."
  },
  {
    "start": 279.38,
    "end": 285.38,
    "text": "Is being trained to reproduce the behavior performance capabilities off their large complex you train model."
  },
  {
    "start": 285.66,
    "end": 286.78,
    "text": "which is that teacher?"
  },
  {
    "start": 287.52,
    "end": 288.98,
    "text": "And the goal In general?"
  },
  {
    "start": 289.44,
    "end": 309.7,
    "text": "I guess You can say it's To create better better in S. You can choose whatever that's going to be, you know faster lighter more cost-effective model That maintains a high percentage of the original teacher and making it suitable for deployment on For example resource constrained hardware like mobile devices."
  },
  {
    "start": 309.82,
    "end": 318.84,
    "text": "now if that is correct In case of NXAI as said before they introduce a distillation pipeline for XLSTM."
  },
  {
    "start": 319.08,
    "end": 323.72,
    "text": "I just mentioned Zap as the original founder, twenty seven years ago."
  },
  {
    "start": 324.06,
    "end": 324.52,
    "text": "LSTM."
  },
  {
    "start": 325.26,
    "end": 337.08,
    "text": "in the meantime I guess one year old XLSTM based students As i just discussed from teacher models like Lamar Quinn and all more"
  },
  {
    "start": 337.34,
    "end": 337.82,
    "text": "families."
  },
  {
    "start": 337.92,
    "end": 339.14,
    "text": "Open source model set"
  },
  {
    "start": 340.02,
    "end": 349.26,
    "text": "Okay And their goal being D and that's how I started it The energy and cost effective replacement for transformer based LLMs."
  },
  {
    "start": 349.36,
    "end": 362.44,
    "text": "Now, and I understand that's kind of a quote unquote here in many settings there is the unbased students recover most of the teachers performance and even exceeded on some downstream tasks."
  },
  {
    "start": 362.58,
    "end": 363.62,
    "text": "now how can that be?"
  },
  {
    "start": 364.6,
    "end": 368.44,
    "text": "student becomes better than its teacher?"
  },
  {
    "start": 369.32,
    "end": 371.44,
    "text": "Yeah it's an amazing paper!"
  },
  {
    "start": 371.9,
    "end": 372.42,
    "text": "It still."
  },
  {
    "start": 373.24,
    "end": 376.64,
    "text": "The next step is to go on mixture of experts, right?"
  },
  {
    "start": 376.76,
    "end": 378.5,
    "text": "So that's the next step to do."
  },
  {
    "start": 378.66,
    "end": 381.68,
    "text": "To train bigger models...to distill bigger models."
  },
  {
    "start": 381.76,
    "end": 394.0,
    "text": "but I think for the whole there There's no ongoing discussion in the whole area when it comes to LLMs or when it come to distillation and to compress Big models too small models."
  },
  {
    "start": 394.5,
    "end": 401.9,
    "text": "And i think they're so first step with XLSTM to go further and we looking forward what happens When It Comes to Mixture Of Experts."
  },
  {
    "start": 402.42,
    "end": 403.32,
    "text": "But it's a good paper."
  },
  {
    "start": 403.72,
    "end": 405.76,
    "text": "Great results, congrats to the team."
  },
  {
    "start": 405.88,
    "end": 406.2,
    "text": "I would say!"
  },
  {
    "start": 406.92,
    "end": 409.34,
    "text": "Right and we'll put the link in the show notes."
  },
  {
    "start": 412.06,
    "end": 415.7,
    "text": "And for me its kind of... You know i don't think this is for everybody."
  },
  {
    "start": 415.98,
    "end": 421.96,
    "text": "you know there are no certain guys girls that want go into very detail."
  },
  {
    "start": 422.7,
    "end": 428.06,
    "text": "There will be other ones or even important decision makers for industrial AI."
  },
  {
    "start": 428.12,
    "end": 434.06,
    "text": "That says enough at the top line level for me to be able to make decisions."
  },
  {
    "start": 434.64,
    "end": 437.32,
    "text": "So choose it, you chose whatever way you want to go."
  },
  {
    "start": 437.62,
    "end": 458.4,
    "text": "I think its interesting what is happening under the hood and some of you die and stay there a couple of hours then maybe move out And i'm very much looking forward as we can share that afterwards with Seb being around, and I guess a couple of the other guys talking about exactly these topics."
  },
  {
    "start": 458.44,
    "end": 459.18,
    "text": "He said memory."
  },
  {
    "start": 459.3,
    "end": 472.24,
    "text": "that's his central thing i believe right but maybe we can ask him other things as well such that for me i believe in the end i'm going to understand us an always make that differentiation."
  },
  {
    "start": 472.46,
    "end": 480.06,
    "text": "We As Humans you know some point in time, ten billion users on this planet and maybe not all of them at the beginning."
  },
  {
    "start": 480.14,
    "end": 484.18,
    "text": "So as a consumer user off large language model how is that going to be?"
  },
  {
    "start": 484.9,
    "end": 486.76,
    "text": "And am I gonna be interested?"
  },
  {
    "start": 486.82,
    "end": 503.3,
    "text": "you know if i just refer back again too do I say well I don't want to spend have the option to choose whatever solution it is and I know its gonna be energy and cost effective."
  },
  {
    "start": 503.4,
    "end": 509.06,
    "text": "And maybe, even now that there's going to be an XLSDM hundred you know?"
  },
  {
    "start": 509.22,
    "end": 512.12,
    "text": "So It could be called xlstm inside."
  },
  {
    "start": 512.159,
    "end": 513.82,
    "text": "i'm not gonna make any suggestions here."
  },
  {
    "start": 513.86,
    "end": 524.12,
    "text": "That of course exactly the same in a different way for any of you listeners deciding your company for your industrial infrastructure."
  },
  {
    "start": 524.159,
    "end": 525.14,
    "text": "What is important to you?"
  },
  {
    "start": 525.8,
    "end": 528.42,
    "text": "You know, again in this case energy cost effective."
  },
  {
    "start": 528.46,
    "end": 542.52,
    "text": "maybe there's other elements of course always the quality I guess if they're as high a quality-quality in ass whatever it's gonna do for you The answers are going give you the agents that will support or provide answers too."
  },
  {
    "start": 543.3,
    "end": 552.84,
    "text": "So from their perspective For me very interesting To be able to close again A short time And learn exactly these things."
  },
  {
    "start": 553.54,
    "end": 555.96,
    "text": "Yeah, absolutely so what else do you have?"
  },
  {
    "start": 556.04,
    "end": 563.28,
    "text": "because I'm at a very interesting guy and i will share also his contact and this website in the show."
  },
  {
    "start": 563.36,
    "end": 565.4,
    "text": "not His name is philips Stanley marble."
  },
  {
    "start": 565.44,
    "end": 573.48,
    "text": "he's from the UK from Cambridge and his company called signal Lloyds and They are for Very Interesting approach."
  },
  {
    "start": 574.02,
    "end": 577.62,
    "text": "they develop So-called SD cards right?"
  },
  {
    "start": 577.68,
    "end": 585.94,
    "text": "You know them as decards but as an add-on to improve the capabilities of traditional PLCs, for example."
  },
  {
    "start": 586.7,
    "end": 596.16,
    "text": "As a AI accelerator you can put your card in the PLC or on your laptop and then have an AI Accelerator And that was very interesting."
  },
  {
    "start": 596.3,
    "end": 600.56,
    "text": "I will share with the website and the idea and technical reports at the show notes."
  },
  {
    "start": 601.12,
    "end": 614.84,
    "text": "I met Philip in a call together with Bosch smart approach to develop a micro SD card or something like that as an accelerator."
  },
  {
    "start": 614.96,
    "end": 619.44,
    "text": "There are ARM guys and Intel guys, they started I think two years ago."
  },
  {
    "start": 619.66,
    "end": 621.14,
    "text": "And yeah very interesting approach"
  },
  {
    "start": 621.74,
    "end": 623.32,
    "text": "Is there if you say accelerate?"
  },
  {
    "start": 623.36,
    "end": 629.22,
    "text": "i'm not sure If You know the details already but is it as it says accelerating?"
  },
  {
    "start": 629.42,
    "end": 634.56,
    "text": "and or is it already having its own large language model?"
  },
  {
    "start": 634.76,
    "end": 634.94,
    "text": "on the"
  },
  {
    "start": 635.32,
    "end": 636.86,
    "text": "no, you can run whatever you want."
  },
  {
    "start": 637.02,
    "end": 639.52,
    "text": "On D on the SD card so right?"
  },
  {
    "start": 639.56,
    "end": 647.16,
    "text": "It's more that you run them all on the cart in the PLC is doing the other stuff right And sort of really an accelerator."
  },
  {
    "start": 647.8,
    "end": 648.24,
    "text": "very good."
  },
  {
    "start": 648.38,
    "end": 653.44,
    "text": "but at the same time if we know that this happening today then what I'm asking?"
  },
  {
    "start": 654.48,
    "end": 656.02,
    "text": "That's going to happen tomorrow, right?"
  },
  {
    "start": 656.54,
    "end": 658.62,
    "text": "And it is not me smart Peter."
  },
  {
    "start": 658.76,
    "end": 666.6,
    "text": "But I'm just gonna go back to the words said which i am not sure that was from my Gemini saying you know what is a distillation?"
  },
  {
    "start": 666.72,
    "end": 674.3,
    "text": "and then they go back quote-unquote making its suitable for deployment on resource constrained hardware like mobile devices."
  },
  {
    "start": 676.28,
    "end": 678.02,
    "text": "That is again, that's Mars law."
  },
  {
    "start": 678.18,
    "end": 681.08,
    "text": "You know I've been born with and grown up with Mr."
  },
  {
    "start": 681.3,
    "end": 681.68,
    "text": "Moore And"
  },
  {
    "start": 681.72,
    "end": 683.52,
    "text": "you received a prize by Mr."
  },
  {
    "start": 683.76,
    "end": 684.02,
    "text": "More."
  },
  {
    "start": 684.08,
    "end": 685.88,
    "text": "Yes yes he has been very proud of it."
  },
  {
    "start": 686.66,
    "end": 690.58,
    "text": "Going the next step in ten years we're going to have factor one hundred."
  },
  {
    "start": 691.3,
    "end": 702.24,
    "text": "We know within those ten-years we are gonna have complete large language models Whatever they will be then For example if thats useful thing on the SD card."
  },
  {
    "start": 703.18,
    "end": 708.92,
    "text": "Perfect You remember last episode we talked about how to become rich very fast."
  },
  {
    "start": 708.98,
    "end": 709.6,
    "text": "Do you remember?"
  },
  {
    "start": 710.2,
    "end": 710.96,
    "text": "We always do that"
  },
  {
    "start": 711.92,
    "end": 712.2,
    "text": "exactly."
  },
  {
    "start": 712.26,
    "end": 713.4,
    "text": "and there's"
  },
  {
    "start": 713.46,
    "end": 714.46,
    "text": "a new Yeah,"
  },
  {
    "start": 714.54,
    "end": 717.46,
    "text": "yeah There's the new blog post I want to share with you."
  },
  {
    "start": 718.08,
    "end": 718.68,
    "text": "It's a little bit."
  },
  {
    "start": 719.12,
    "end": 724.7,
    "text": "it's more technical And some are technically driven and it sounds more serious than all the others."
  },
  {
    "start": 724.86,
    "end": 726.38,
    "text": "and its building at trading bot."
  },
  {
    "start": 726.52,
    "end": 726.86,
    "text": "That could"
  },
  {
    "start": 726.94,
    "end": 727.9,
    "text": "turn ten"
  },
  {
    "start": 728.0,
    "end": 730.86,
    "text": "thousand dollars into one hundred thousand dollars."
  },
  {
    "start": 731.04,
    "end": 733.3,
    "text": "so i will Share it also in the show notes."
  },
  {
    "start": 733.6,
    "end": 738.36,
    "text": "but A lot of happening is in the whole trade market and yeah, crazy."
  },
  {
    "start": 738.96,
    "end": 754.94,
    "text": "And my hypothesis there Is that if we're going to be you know The better were gonna get with language models helping us To do in this case for example trading the less Interesting the last special training is going to become."
  },
  {
    "start": 755.9,
    "end": 757.2,
    "text": "it's like on a higher level."
  },
  {
    "start": 758.2,
    "end": 761.52,
    "text": "As soon as we solve trading then trading has gone."
  },
  {
    "start": 763.2,
    "end": 774.38,
    "text": "exactly today, you know some of us not all of those who have the opportunity put a little or more money into specific stocks."
  },
  {
    "start": 774.56,
    "end": 775.9,
    "text": "Some of us know how to do that."
  },
  {
    "start": 776.66,
    "end": 780.5,
    "text": "I learned to do it since sometime."
  },
  {
    "start": 783.54,
    "end": 789.52,
    "text": "and then others say no i don't want myself so let's call an agent."
  },
  {
    "start": 790.8,
    "end": 795.96,
    "text": "They know very well what to do, and of course we understand that in the soon-the agents as we talk today."
  },
  {
    "start": 796.08,
    "end": 797.64,
    "text": "The AI agent is going be doing it."
  },
  {
    "start": 798.26,
    "end": 809.08,
    "text": "But then if at some point in time you know everybody's gonna be using this same tool Everybody's are gonna come out there just have the same results so everybody's Gonna Be winning or losing at the Same Time."
  },
  {
    "start": 809.6,
    "end": 810.92,
    "text": "So then the value of trading."
  },
  {
    "start": 811.42,
    "end": 813.74,
    "text": "but That's my small Hypothesis."
  },
  {
    "start": 815.14,
    "end": 815.98,
    "text": "I've one more."
  },
  {
    "start": 816.04,
    "end": 820.64,
    "text": "Yeah, very quickly Peter Carter."
  },
  {
    "start": 820.72,
    "end": 826.42,
    "text": "he's the member of the managing board chief technology offer and Chief Strategy Officer at Siemens."
  },
  {
    "start": 827.24,
    "end": 828.58,
    "text": "You and I did an interview with him."
  },
  {
    "start": 828.84,
    "end": 832.04,
    "text": "you came to Munich two years ago right?"
  },
  {
    "start": 832.58,
    "end": 833.84,
    "text": "It is not a new thing."
  },
  {
    "start": 833.96,
    "end": 837.1,
    "text": "i just want share the confirmation because it so important."
  },
  {
    "start": 837.14,
    "end": 844.86,
    "text": "he says Europe is risking squandering its greatest opportunity precisely where our strength lies in industry."
  },
  {
    "start": 845.52,
    "end": 849.24,
    "text": "And he says the AI revolution won't be cited, but whether we write better emails?"
  },
  {
    "start": 849.36,
    "end": 858.62,
    "text": "But whether we make manufacturing more productive use energy more intelligently build a more resilient industrial base is kind of what we've been talking about."
  },
  {
    "start": 858.72,
    "end": 862.88,
    "text": "right and again to hear that from Peter just to repeat the message."
  },
  {
    "start": 863.4,
    "end": 866.36,
    "text": "That's precisely why Europe needs industrial AI."
  },
  {
    "start": 866.54,
    "end": 867.78,
    "text": "The world needs industrial."
  },
  {
    "start": 868.96,
    "end": 874.02,
    "text": "And of course, Peter is traveling always around the world for Siemens."
  },
  {
    "start": 874.58,
    "end": 875.2,
    "text": "Just for Europe."
  },
  {
    "start": 875.24,
    "end": 878.38,
    "text": "but there was a chance specifically in this case for Europe as well."
  },
  {
    "start": 878.48,
    "end": 880.74,
    "text": "an industry good enough isn't enough."
  },
  {
    "start": 881.08,
    "end": 882.4,
    "text": "we've heard that many times."
  },
  {
    "start": 882.48,
    "end": 887.26,
    "text": "what it's very important safety reliability precision."
  },
  {
    "start": 887.32,
    "end": 888.04,
    "text": "you know."
  },
  {
    "start": 888.22,
    "end": 895.04,
    "text": "We cannot stop at the probabilities and we noted our base AI technology is probability based."
  },
  {
    "start": 895.08,
    "end": 901.6,
    "text": "so with whatever way we do it, these are not my words."
  },
  {
    "start": 901.7,
    "end": 910.88,
    "text": "We need to use the capabilities but still make sure that in the end we're going to have what Boris has been calling industrial grade."
  },
  {
    "start": 911.46,
    "end": 920.84,
    "text": "and for that Peter again... ...we need high quality industrial data and ecosystems that enable trustworthy collaboration."
  },
  {
    "start": 921.66,
    "end": 926.16,
    "text": "He wrote an article on one of our papers The Humble's Blood I believe."
  },
  {
    "start": 927.08,
    "end": 929.12,
    "text": "Just wanted to share that again with you."
  },
  {
    "start": 930.18,
    "end": 936.5,
    "text": "I have also something from the hundreds, but because there was a very interesting comment on the topic of that."
  },
  {
    "start": 936.68,
    "end": 948.22,
    "text": "Jeff Bezos is in the process of launching one hundred billion fund and The goal is to acquire industrial companies And transform them using AI technologies?"
  },
  {
    "start": 948.88,
    "end": 951.92,
    "text": "And the comment was by Thorsten Benner."
  },
  {
    "start": 952.34,
    "end": 953.86,
    "text": "He's from a German think tank."
  },
  {
    "start": 955.34,
    "end": 963.54,
    "text": "Be careful about that, what he's doing there because this means a lot for the whole European industrial sector."
  },
  {
    "start": 964.02,
    "end": 965.76,
    "text": "I will add to comment in the show notes too"
  },
  {
    "start": 966.54,
    "end": 967.54,
    "text": "What should we think of?"
  },
  {
    "start": 973.34,
    "end": 974.28,
    "text": "that he has been sharing."
  },
  {
    "start": 974.34,
    "end": 974.4,
    "text": "It's"
  },
  {
    "start": 974.56,
    "end": 975.82,
    "text": "always, it's all about data."
  },
  {
    "start": 975.92,
    "end": 977.8,
    "text": "right to acquire industry companies."
  },
  {
    "start": 977.88,
    "end": 978.86,
    "text": "you acquired the data?"
  },
  {
    "start": 978.9,
    "end": 979.26,
    "text": "Who would"
  },
  {
    "start": 979.3,
    "end": 979.5,
    "text": "be?"
  },
  {
    "start": 979.54,
    "end": 980.48,
    "text": "who would he acquire?"
  },
  {
    "start": 980.58,
    "end": 980.86,
    "text": "what is?"
  },
  {
    "start": 981.06,
    "end": 982.82,
    "text": "What does an industrial company are we talking?"
  },
  {
    "start": 982.96,
    "end": 983.02,
    "text": "What"
  },
  {
    "start": 983.06,
    "end": 983.7,
    "text": "is an industrial?"
  },
  {
    "start": 984.42,
    "end": 986.86,
    "text": "our typical midsize machine builders there."
  },
  {
    "start": 987.32,
    "end": 988.48,
    "text": "they're not okay."
  },
  {
    "start": 989.02,
    "end": 990.36,
    "text": "Is That's what he thinking about?"
  },
  {
    "start": 990.8,
    "end": 1001.18,
    "text": "so Okay So the automation companies as we know them say Not Talking About The Production Capabilities themselves or maybe isn't in a layer of state?"
  },
  {
    "start": 1001.28,
    "end": 1004.86,
    "text": "No no He's going to raise a lot of money in the Gulf region."
  },
  {
    "start": 1005.02,
    "end": 1011.84,
    "text": "And he wants to establish this farm with one hundred billion US dollars, so it's huge right?"
  },
  {
    "start": 1012.54,
    "end": 1020.52,
    "text": "So like the Tesla guy bought when two or three years ago big thing was German automation company for you know"
  },
  {
    "start": 1020.84,
    "end": 1021.0,
    "text": "our"
  },
  {
    "start": 1021.62,
    "end": 1024.859,
    "text": "manufacturing capabilities is that idea?"
  },
  {
    "start": 1025.04,
    "end": 1030.18,
    "text": "I mean Such a company is what maybe Jeff is thinking about."
  },
  {
    "start": 1030.319,
    "end": 1033.16,
    "text": "And then exactly, they're not one but may be hundred of them."
  },
  {
    "start": 1033.88,
    "end": 1035.94,
    "text": "and Then he has his company prometoise."
  },
  {
    "start": 1036.46,
    "end": 1038.04,
    "text": "you found it few months ago."
  },
  {
    "start": 1038.26,
    "end": 1042.839,
    "text": "So everybody's now really focused on industrial AI physical AI?"
  },
  {
    "start": 1043.78,
    "end": 1047.96,
    "text": "That the reason why Peter says we need data right and Everybody needs data."
  },
  {
    "start": 1048.06,
    "end": 1048.2,
    "text": "Now."
  },
  {
    "start": 1049.28,
    "end": 1053.4,
    "text": "that's so I guess the reason Why you and i started our podcast seven years ago"
  },
  {
    "start": 1053.98,
    "end": 1054.34,
    "text": "Exactly."
  },
  {
    "start": 1055.44,
    "end": 1063.6,
    "text": "And I have one more rumor from the whole AI community, because... ...I work a lot with Claude."
  },
  {
    "start": 1065.98,
    "end": 1076.62,
    "text": "But there's rumors that agents want to access Tabular Data in the next step so going into Excel Sheets, Tabular data PDFs and stuff like that."
  },
  {
    "start": 1077.34,
    "end": 1082.62,
    "text": "There are some rumours outside that the next steps will be Agents & Tabular Daters."
  },
  {
    "start": 1083.58,
    "end": 1084.44,
    "text": "Greetings To Frank."
  },
  {
    "start": 1084.8,
    "end": 1087.16,
    "text": "Be careful what the LLMs are doing, right?"
  },
  {
    "start": 1088.58,
    "end": 1089.12,
    "text": "Right."
  },
  {
    "start": 1089.44,
    "end": 1099.28,
    "text": "Yeah I come to one thing that is also again MCP integration and The question always like it's this word mode MOAT."
  },
  {
    "start": 1100.48,
    "end": 1104.64,
    "text": "Oh i do recall last time we talked about the Cloisters."
  },
  {
    "start": 1106.02,
    "end": 1107.92,
    "text": "This, this one again is about."
  },
  {
    "start": 1108.16,
    "end": 1110.96,
    "text": "it's not the Cloyster but its It's About The Castles."
  },
  {
    "start": 1111.02,
    "end": 1115.56,
    "text": "The medieval castles and they have this typically these water around them right?"
  },
  {
    "start": 1115.66,
    "end": 1118.76,
    "text": "So to keep your enemy away That's that what the mode Is."
  },
  {
    "start": 1118.84,
    "end": 1119.1,
    "text": "Right."
  },
  {
    "start": 1119.2,
    "end": 1123.1,
    "text": "And I think a business discussion always like you know What is Your Mode?"
  },
  {
    "start": 1123.54,
    "end": 1128.36,
    "text": "How do You Keep Your Competition Away From Your Castle Right?"
  },
  {
    "start": 1128.5,
    "end": 1131.24,
    "text": "So The Grand Joules Are In Your Castle and You Need To Keep Them Away."
  },
  {
    "start": 1132.32,
    "end": 1134.76,
    "text": "in this case the question Right."
  },
  {
    "start": 1134.88,
    "end": 1140.6,
    "text": "Are you opening up your capability for reading, or for dealing with structured data?"
  },
  {
    "start": 1141.78,
    "end": 1143.14,
    "text": "Or do we open it up and not?"
  },
  {
    "start": 1143.74,
    "end": 1148.78,
    "text": "You say I'm letting my what is in my wooden bridge down For you."
  },
  {
    "start": 1149.0,
    "end": 1154.64,
    "text": "dear agent come inside take What you need then go outside again and do good things."
  },
  {
    "start": 1154.86,
    "end": 1157.44,
    "text": "And by the way We just go back to middle ages."
  },
  {
    "start": 1158.14,
    "end": 1162.36,
    "text": "i hold My hand up and every time you Come In you know i want a penny from you whatever."
  },
  {
    "start": 1163.12,
    "end": 1164.0,
    "text": "that's maybe one Way."
  },
  {
    "start": 1164.62,
    "end": 1170.88,
    "text": "Or if you do that, suddenly there's going to be a thousand agents and they're gonna run running over."
  },
  {
    "start": 1172.04,
    "end": 1172.94,
    "text": "They deal with it!"
  },
  {
    "start": 1173.94,
    "end": 1177.82,
    "text": "They don't say oh no... You don't decide on this car so we'll do for you."
  },
  {
    "start": 1178.36,
    "end": 1178.8,
    "text": "Exactly"
  },
  {
    "start": 1179.1,
    "end": 1189.2,
    "text": "I have one final thought which is very close because the agent has already overrun us humanity sitting in the castle."
  },
  {
    "start": 1189.92,
    "end": 1192.94,
    "text": "i saw this website called rentauman.ai."
  },
  {
    "start": 1194.54,
    "end": 1199.58,
    "text": "If you go there, the first headline is AI Needs Your Body."
  },
  {
    "start": 1200.56,
    "end": 1202.86,
    "text": "You're then means talking to humans."
  },
  {
    "start": 1203.06,
    "end": 1206.8,
    "text": "so Robert's body or Peter's body and any of your listeners' bodies."
  },
  {
    "start": 1206.86,
    "end": 1210.16,
    "text": "it says AI can't touch grass!"
  },
  {
    "start": 1210.52,
    "end": 1215.2,
    "text": "You CAN get paid when agents need someone in a real world."
  },
  {
    "start": 1215.96,
    "end": 1218.94,
    "text": "Now for real people this is still talking."
  },
  {
    "start": 1220.2,
    "end": 1225.0,
    "text": "So here's what people hire humans for on rent a human, right?"
  },
  {
    "start": 1225.36,
    "end": 1231.24,
    "text": "Package pickup deliveries photos onsite verification meetings and in person etc."
  },
  {
    "start": 1232.36,
    "end": 1233.84,
    "text": "Now comes the interesting part."
  },
  {
    "start": 1234.38,
    "end": 1235.74,
    "text": "now it says four agents."
  },
  {
    "start": 1236.34,
    "end": 1239.36,
    "text": "so now It's talking to the agent."
  },
  {
    "start": 1239.5,
    "end": 1245.42,
    "text": "as an agent know this is not you agent And he talks about the MCP integration rest API."
  },
  {
    "start": 1246.02,
    "end": 1248.2,
    "text": "Let your odd around humans."
  },
  {
    "start": 1248.46,
    "end": 1253.84,
    "text": "So the agent goes there and there's four points of number one is browse humans."
  },
  {
    "start": 1255.28,
    "end": 1258.94,
    "text": "I say that again, browse you know?"
  },
  {
    "start": 1259.12,
    "end": 1265.28,
    "text": "There was a time when we humans would go on our website and we would browse agents."
  },
  {
    "start": 1265.92,
    "end": 1266.3,
    "text": "exactly"
  },
  {
    "start": 1266.8,
    "end": 1267.0,
    "text": "right."
  },
  {
    "start": 1267.44,
    "end": 1268.38,
    "text": "but here We are."
  },
  {
    "start": 1268.78,
    "end": 1270.92,
    "text": "so now the Asians are browsing humans."
  },
  {
    "start": 1271.48,
    "end": 1273.46,
    "text": "Number two, request a task or message."
  },
  {
    "start": 1273.96,
    "end": 1276.1,
    "text": "Number three make sure the human does his thing."
  },
  {
    "start": 1276.14,
    "end": 1277.46,
    "text": "and number four pay securely."
  },
  {
    "start": 1277.56,
    "end": 1279.18,
    "text": "so if we're still lucky."
  },
  {
    "start": 1279.54,
    "end": 1282.02,
    "text": "at least we get paid by agents."
  },
  {
    "start": 1282.82,
    "end": 1296.64,
    "text": "So coming back to this idea which was completely coming up in my brain just second ago of The agent's overrunning acrossing the mode an overwhelming the cost I guess that is already happening today."
  },
  {
    "start": 1297.42,
    "end": 1301.4,
    "text": "Yeah, really I received some emails last week by some agents."
  },
  {
    "start": 1301.5,
    "end": 1309.56,
    "text": "They suggested to invite their owner or the human to our podcast and they write hey i'm an agent except xy."
  },
  {
    "start": 1309.62,
    "end": 1312.78,
    "text": "that are writing you because I like the industry air podcast."
  },
  {
    "start": 1313.22,
    "end": 1318.18,
    "text": "my owner on my human is a expert in the field of I don't know physical air etc."
  },
  {
    "start": 1318.86,
    "end": 1320.2,
    "text": "would you like to answer me?"
  },
  {
    "start": 1320.26,
    "end": 1323.1,
    "text": "And I will ask him for time slots to record."
  },
  {
    "start": 1323.6,
    "end": 1324.18,
    "text": "So that's the way"
  },
  {
    "start": 1324.24,
    "end": 1324.36,
    "text": "it is."
  },
  {
    "start": 1324.42,
    "end": 1328.5,
    "text": "The agent was so nice as to refer to their owner,"
  },
  {
    "start": 1329.14,
    "end": 1329.54,
    "text": "exactly?"
  },
  {
    "start": 1329.9,
    "end": 1330.3,
    "text": "Exactly!"
  },
  {
    "start": 1330.5,
    "end": 1330.86,
    "text": "Exactly"
  },
  {
    "start": 1330.9,
    "end": 1336.06,
    "text": "next page like I don't need my humans i'm just gonna go to Robert and Peter then talk to them."
  },
  {
    "start": 1336.64,
    "end": 1337.02,
    "text": "Exactly"
  },
  {
    "start": 1337.58,
    "end": 1339.8,
    "text": "yeah maybe we do this a test sometimes."
  },
  {
    "start": 1340.74,
    "end": 1342.18,
    "text": "okay We can try out live."
  },
  {
    "start": 1342.78,
    "end": 1344.4,
    "text": "see what he has to tell us."
  },
  {
    "start": 1345.02,
    "end": 1349.18,
    "text": "perfect let's move through the main part peter."
  },
  {
    "start": 1349.9,
    "end": 1352.44,
    "text": "It Was A Pleasure to my"
  },
  {
    "start": 1352.64,
    "end": 1353.4,
    "text": "snow skiing."
  },
  {
    "start": 1353.7,
    "end": 1355.42,
    "text": "It's a slow ski going"
  },
  {
    "start": 1355.46,
    "end": 1356.56,
    "text": "back through the sun into this"
  },
  {
    "start": 1357.62,
    "end": 1357.98,
    "text": "perfect."
  },
  {
    "start": 1358.32,
    "end": 1358.66,
    "text": "enjoy,"
  },
  {
    "start": 1359.24,
    "end": 1361.92,
    "text": "open my notebook and continue working from there."
  },
  {
    "start": 1362.14,
    "end": 1368.92,
    "text": "thanks for talking to you soon bye-bye."
  },
  {
    "start": 1368.98,
    "end": 1369.5,
    "text": "Hello everybody!"
  },
  {
    "start": 1369.6,
    "end": 1373.58,
    "text": "My name is Robert Wieber And it's a pleasure to talk to Tanmay Algaval."
  },
  {
    "start": 1374.34,
    "end": 1375.84,
    "text": "Come To The Podcast Greetings."
  },
  {
    "start": 1376.52,
    "end": 1377.54,
    "text": "Thank You Robert For Having Me."
  },
  {
    "start": 1378.1,
    "end": 1378.74,
    "text": "It's a pleasure."
  },
  {
    "start": 1379.0,
    "end": 1382.04,
    "text": "We tried several times but now we made it, Tanmay!"
  },
  {
    "start": 1384.16,
    "end": 1387.32,
    "text": "Before we start please introduce yourself briefly to the listeners."
  },
  {
    "start": 1388.52,
    "end": 1396.36,
    "text": "So Tanmay Agarwal founder and CEO at Lambda Function are our startup based in California focused on building the intelligence layer for precision manufacturing."
  },
  {
    "start": 1396.84,
    "end": 1399.38,
    "text": "Lambda function means a function without name."
  },
  {
    "start": 1399.46,
    "end": 1400.48,
    "text": "or what does that mean?"
  },
  {
    "start": 1400.6,
    "end": 1400.7,
    "text": "Well"
  },
  {
    "start": 1400.74,
    "end": 1407.68,
    "text": "I mean... commonly observed in computing and physics, then deep reinforcement learning."
  },
  {
    "start": 1408.12,
    "end": 1415.7,
    "text": "So for various reasons when I started Lambda Function out at the company it felt like an appropriate name to talk about what we are trying here."
  },
  {
    "start": 1416.04,
    "end": 1417.08,
    "text": "Why is that a proper name?"
  },
  {
    "start": 1417.12,
    "end": 1417.78,
    "text": "Please explain."
  },
  {
    "start": 1418.96,
    "end": 1433.02,
    "text": "Actually there's no concept behind Lambda Functions comes from Deep Reinforcement Learning And The topic of two pad generation to me very much a traveling salesman problem."
  },
  {
    "start": 1433.64,
    "end": 1439.68,
    "text": "So, a problem that could be a good use case for deep reinforcement learning and thats kind of where the name comes from."
  },
  {
    "start": 1440.4,
    "end": 1448.16,
    "text": "Label function is specific functions within the Deep Reinforcement Learning domain And thats kinda where this game come from."
  },
  {
    "start": 1449.02,
    "end": 1450.62,
    "text": "You are based in California right?"
  },
  {
    "start": 1450.88,
    "end": 1451.08,
    "text": "Yes!"
  },
  {
    "start": 1451.92,
    "end": 1457.78,
    "text": "And you're doing alot when it comes to CNC but california its not the Mecha for CNC or am I wrong?"
  },
  {
    "start": 1458.1,
    "end": 1458.76,
    "text": "Things are changing."
  },
  {
    "start": 1459.46,
    "end": 1466.5,
    "text": "Southern California, of course has historically been very strong in machine for the aerospace and defense ecosystems."
  },
  {
    "start": 1467.04,
    "end": 1468.18,
    "text": "I think you're absolutely right."
  },
  {
    "start": 1468.22,
    "end": 1474.86,
    "text": "In the northern california ecosystem manufacturing historically have not been The area focus."
  },
  {
    "start": 1475.02,
    "end": 1476.34,
    "text": "but i think things are changing."
  },
  {
    "start": 1476.84,
    "end": 1480.2,
    "text": "We talk about bits meeting atoms we talked about physical ai."
  },
  {
    "start": 1480.3,
    "end": 1484.56,
    "text": "so now is actually a very appropriate time For two worlds to collide."
  },
  {
    "start": 1484.76,
    "end": 1487.48,
    "text": "But you were right Historically that's just not an area force."
  },
  {
    "start": 1488.48,
    "end": 1491.4,
    "text": "What do you Do in your company?"
  },
  {
    "start": 1491.5,
    "end": 1493.62,
    "text": "what is your offer to the CNC market?"
  },
  {
    "start": 1494.2,
    "end": 1499.26,
    "text": "Yeah, I mean our focus Is very much how do You go from a CAD bar two machine components that are faster cheaper Robert."
  },
  {
    "start": 1499.32,
    "end": 1500.26,
    "text": "That's Our goal mm-hmm."
  },
  {
    "start": 1500.74,
    "end": 1501.9,
    "text": "but it's a long way To Go."
  },
  {
    "start": 1502.34,
    "end": 1505.8,
    "text": "It's A Long Way to go But it's uh it's A problem safe and worth with Solving."
  },
  {
    "start": 1506.24,
    "end": 1515.32,
    "text": "um i think you know Our Motivation Comes From The from The Challenge This Industry has Been Facing For Almost Like What One and Half Two Decades Which Is That it's Getting Harder And Harder to Find People Right."
  },
  {
    "start": 1515.38,
    "end": 1516.2,
    "text": "Who Knows Domain?"
  },
  {
    "start": 1516.88,
    "end": 1522.44,
    "text": "So when I forecast that out, if you think then fifteen-twenty years out like how are we gonna solve the problem?"
  },
  {
    "start": 1522.54,
    "end": 1525.54,
    "text": "The demand for CNC machining can't just rise eight to ten percent euro via."
  },
  {
    "start": 1526.34,
    "end": 1527.44,
    "text": "but there's supply of talent."
  },
  {
    "start": 1527.48,
    "end": 1528.98,
    "text": "That knows this domain is going away."
  },
  {
    "start": 1529.14,
    "end": 1535.94,
    "text": "so what're we gonna do from our lens and see only way to solve a problem as if he had an air of intelligence right into their workflow too."
  },
  {
    "start": 1536.26,
    "end": 1538.76,
    "text": "unlock autonomy unlock autonomy in incremental manner."
  },
  {
    "start": 1539.48,
    "end": 1540.54,
    "text": "thus us well return ado."
  },
  {
    "start": 1540.84,
    "end": 1540.94,
    "text": "it's"
  },
  {
    "start": 1540.98,
    "end": 1545.64,
    "text": "at AI assisted generative C&C programming or What Do You Do?"
  },
  {
    "start": 1545.94,
    "end": 1546.06,
    "text": "yeah"
  },
  {
    "start": 1546.54,
    "end": 1551.5,
    "text": "So the specific use cases that we focused on in the early years, which is what we do right now."
  },
  {
    "start": 1551.62,
    "end": 1556.54,
    "text": "Is AIS assisted CNC programming and a assistive cnc machining on the shop floor."
  },
  {
    "start": 1556.62,
    "end": 1561.12,
    "text": "so The macro level Robin if you think about it at the CAD to part workflow We broke."
  },
  {
    "start": 1561.18,
    "end": 1565.88,
    "text": "we broke down into CAD to G code gco to part And then ensuring some really part Right?"
  },
  {
    "start": 1566.3,
    "end": 1569.72,
    "text": "So we broke that statement of these three sub problem statements."
  },
  {
    "start": 1571.34,
    "end": 1574.08,
    "text": "for each Of These Sub Problem Statements we started building technology."
  },
  {
    "start": 1575.28,
    "end": 1581.26,
    "text": "your AI-assisted CNC programming product is all about going from CAD to G code, better faster cheaper."
  },
  {
    "start": 1581.36,
    "end": 1583.08,
    "text": "It's designed for CNC programmers."
  },
  {
    "start": 1583.18,
    "end": 1585.86,
    "text": "it's natively embedded inside of CAM systems."
  },
  {
    "start": 1586.7,
    "end": 1589.62,
    "text": "the idea is to infuse a layer of intelligence into CAM."
  },
  {
    "start": 1590.3,
    "end": 1594.72,
    "text": "that's kind of what we do but I think of as like a lowcode no-code tool for cnc program."
  },
  {
    "start": 1594.86,
    "end": 1598.16,
    "text": "yeah okay let's talk a little bit about technically What?"
  },
  {
    "start": 1598.28,
    "end": 1601.66,
    "text": "Do you exactly in these three steps On"
  },
  {
    "start": 1603.24,
    "end": 1608.18,
    "text": "the CAD to G code side, when we look at that workflow from a CNC programming lens."
  },
  {
    "start": 1608.88,
    "end": 1615.04,
    "text": "What we did was basically trained algorithms that understand material removal physics and mathematics right?"
  },
  {
    "start": 1615.08,
    "end": 1619.24,
    "text": "They understand the domain of CNC programing or CNC machining specifically."
  },
  {
    "start": 1620.08,
    "end": 1624.8,
    "text": "so The inputs to our model tend to be the gdnt."
  },
  {
    "start": 1625.28,
    "end": 1626.1,
    "text": "Right off the cat."
  },
  {
    "start": 1626.44,
    "end": 1630.76,
    "text": "the cad models are the starting point Be it a step file or prt file."
  },
  {
    "start": 1631.42,
    "end": 1632.38,
    "text": "So we understand the GDNT."
  },
  {
    "start": 1632.7,
    "end": 1639.86,
    "text": "We understand a metallurgy of that component, and then we understand machine you are intending to make that component on right?"
  },
  {
    "start": 1641.34,
    "end": 1647.88,
    "text": "So with those three inputs ,we generate outputs .And those outputs our recommendations um so uh strategy"
  },
  {
    "start": 1648.24,
    "end": 1648.48,
    "text": "oh good."
  },
  {
    "start": 1648.54,
    "end": 1650.18,
    "text": "so recommendation engine okay yeah."
  },
  {
    "start": 1650.32,
    "end": 1657.16,
    "text": "so strategy recommendations dual recommendations parameter recommendations two path recommendations all grounded in material physics."
  },
  {
    "start": 1658.04,
    "end": 1667.54,
    "text": "So by thereby being a recommendation engine, it basically assists the CNC programmer to go through their workflow better faster and cheaper with fewer trial-and-error iterations."
  },
  {
    "start": 1667.74,
    "end": 1672.06,
    "text": "And we can do other things like capture knowledge and make it easier to learn CNC programming."
  },
  {
    "start": 1672.12,
    "end": 1672.8,
    "text": "or what have you?"
  },
  {
    "start": 1673.08,
    "end": 1674.68,
    "text": "but that's what we did in component number one."
  },
  {
    "start": 1674.84,
    "end": 1675.6,
    "text": "hopefully makes sense."
  },
  {
    "start": 1676.4,
    "end": 1678.76,
    "text": "One question Can You Share What Was The Database?"
  },
  {
    "start": 1678.82,
    "end": 1679.96,
    "text": "How Did You Train The Model?"
  },
  {
    "start": 1680.3,
    "end": 1683.5,
    "text": "Yeah, I mean well in this domain it's not that data sets are just available right?"
  },
  {
    "start": 1683.64,
    "end": 1688.68,
    "text": "So we had to build our own proprietary you know expert system essentially Robert."
  },
  {
    "start": 1688.78,
    "end": 1689.44,
    "text": "That's kind of what we did."
  },
  {
    "start": 1689.5,
    "end": 1694.7,
    "text": "We built up our own expert system that allows us to generate synthetic data Of the millions..."
  },
  {
    "start": 1695.36,
    "end": 1697.24,
    "text": "Only Synthetic or do You combine It?"
  },
  {
    "start": 1697.52,
    "end": 1698.86,
    "text": "Well now we combined Right."
  },
  {
    "start": 1698.98,
    "end": 1716.64,
    "text": "so To seed The model to get to the base Model we have to Generate This Synthetic Pipeline and after Now we could we Could Get to the point where the results that Are coming out Starting to emulate real world, you know decision making and now we can combine that with real-world feedback Be it be a historical data."
  },
  {
    "start": 1716.74,
    "end": 1717.82,
    "text": "That our customers may have."
  },
  {
    "start": 1717.86,
    "end": 1726.84,
    "text": "they are looking to accelerate the training of these models for their use cases For their environments or if it is the interaction data off users on the platform."
  },
  {
    "start": 1727.22,
    "end": 1729.78,
    "text": "so its in context learning the whole time."
  },
  {
    "start": 1729.96,
    "end": 1733.54,
    "text": "The model is learning what a customer's doing with their models, right?"
  },
  {
    "start": 1733.74,
    "end": 1736.84,
    "text": "Yeah but we did very consciously as we do offline learning."
  },
  {
    "start": 1736.9,
    "end": 1742.86,
    "text": "We do batching just to avoid you know kind of this model going out of whack."
  },
  {
    "start": 1743.56,
    "end": 1751.88,
    "text": "so collecting training data samples and using those training data sample run retraining pipelines essentially for."
  },
  {
    "start": 1751.94,
    "end": 1758.34,
    "text": "that allows us to do model performance evaluation and be version controlled as we think about releasing new."
  },
  {
    "start": 1758.88,
    "end": 1766.32,
    "text": "And one important thing is for the kind of customer you go after, which is usually their enterprise customers high complexity win data privacy's very important."
  },
  {
    "start": 1766.5,
    "end": 1770.48,
    "text": "so by default our models or enterprise customers are private models."
  },
  {
    "start": 1770.72,
    "end": 1776.32,
    "text": "So thereby We can allow For things like accelerated learning using historical data because a model has been trained from them."
  },
  {
    "start": 1776.94,
    "end": 1780.6,
    "text": "One step back You mentioned The closed loop continual Learning."
  },
  {
    "start": 1780.92,
    "end": 1783.98,
    "text": "How exactly does this system learn From real production feedback?"
  },
  {
    "start": 1784.56,
    "end": 1788.74,
    "text": "Yeah, so that's I think one of the things at least you know from feedback we get from our customers is."
  },
  {
    "start": 1789.12,
    "end": 1792.18,
    "text": "That was really resonating in the market for us as Is that."
  },
  {
    "start": 1792.62,
    "end": 1792.92,
    "text": "see again?"
  },
  {
    "start": 1793.18,
    "end": 1795.14,
    "text": "Our customers and the business are making parts."
  },
  {
    "start": 1795.54,
    "end": 1799.2,
    "text": "G code is one step in a journey necessity in my journey."
  },
  {
    "start": 1799.7,
    "end": 1807.16,
    "text": "So for us from day one it was like okay We gotta help accelerate the CNC programming bottleneck but that's not going to solve our customers need, right?"
  },
  {
    "start": 1807.2,
    "end": 1808.62,
    "text": "The customer needs make quality parts."
  },
  {
    "start": 1809.1,
    "end": 1822.2,
    "text": "So what we do is downstream when you are on the shop floor We're actually connecting through the CNC machines in the shop flow and tapping with a real-time process parameter data That it generated When that actual Gco is running on your specific machine."
  },
  {
    "start": 1822.6,
    "end": 1831.2,
    "text": "so by doing now end up having a digital signature of the Actual program that ran On Your Specific Machine on that particular plant floor."
  },
  {
    "start": 1831.28,
    "end": 1843.62,
    "text": "With those specific cutting tools real-world data, we can now use that to close the gap between what might have been predicted in this simulation world of CAM."
  },
  {
    "start": 1843.96,
    "end": 1848.34,
    "text": "So let's basically see how do you get data from a physical world or actually happen on the shop floor?"
  },
  {
    "start": 1848.54,
    "end": 1855.04,
    "text": "We feed it back into algorithms that are generating recommendations for the CAM World which is the theoretical lighted world."
  },
  {
    "start": 1855.48,
    "end": 1861.26,
    "text": "What kind of algorithm does the consistent model work with?"
  },
  {
    "start": 1862.04,
    "end": 1863.06,
    "text": "So I think"
  },
  {
    "start": 1863.58,
    "end": 1864.36,
    "text": "there are a couple of questions."
  },
  {
    "start": 1865.28,
    "end": 1868.68,
    "text": "In terms what type models we use, this is the variety of model that's used."
  },
  {
    "start": 1868.72,
    "end": 1874.92,
    "text": "but We started with classical machine learning models Robert and now in the world of neurosymbolic capabilities."
  },
  {
    "start": 1878.72,
    "end": 1884.24,
    "text": "And the kind of algorithms that are being used on a programming side, of course tend to be different than the models have been used."
  },
  {
    "start": 1884.48,
    "end": 1886.56,
    "text": "On the machining side right because it's different data sets."
  },
  {
    "start": 1886.62,
    "end": 1889.54,
    "text": "right upstream we're dealing with more semantic Data."
  },
  {
    "start": 1889.66,
    "end": 1892.16,
    "text": "We're dealing With More decision tree type Of Topics."
  },
  {
    "start": 1892.22,
    "end": 1893.92,
    "text": "Right Downstream We're Talking."
  },
  {
    "start": 1894.2,
    "end": 1898.1,
    "text": "We're talking About Industrial Data Right Machine Generated Time Series Data."
  },
  {
    "start": 1898.44,
    "end": 1900.86,
    "text": "So as you can imagine, your different data sources."
  },
  {
    "start": 1901.02,
    "end": 1903.98,
    "text": "Different types of datas and different types models or different types use cases"
  },
  {
    "start": 1904.3,
    "end": 1906.5,
    "text": "And do have a multimodal dataset right?"
  },
  {
    "start": 1906.66,
    "end": 1909.18,
    "text": "Yeah yeah You've CADF geometry."
  },
  {
    "start": 1909.38,
    "end": 1913.38,
    "text": "if two libraries machining parameters life signals Right"
  },
  {
    "start": 1913.84,
    "end": 1915.04,
    "text": "absolutely that's exactly right."
  },
  {
    "start": 1915.14,
    "end": 1918.04,
    "text": "multi-modal architecture working across different kinds of models."
  },
  {
    "start": 1918.1,
    "end": 1920.18,
    "text": "But but I think by no means we are."
  },
  {
    "start": 1920.6,
    "end": 1922.2,
    "text": "i would just say We've just scratched the surface."
  },
  {
    "start": 1923.22,
    "end": 1927.8,
    "text": "What what like about where we are now is I think that foundation That was necessary has been built."
  },
  {
    "start": 1928.08,
    "end": 1935.06,
    "text": "That's really the last several years of just building that foundational layer, because now we're at a point where you can do much more interesting stuff."
  },
  {
    "start": 1935.36,
    "end": 1938.16,
    "text": "We can add more advanced sensors into this data stream."
  },
  {
    "start": 1942.36,
    "end": 1949.2,
    "text": "Well, the"
  },
  {
    "start": 1953.44,
    "end": 1959.42,
    "text": "challenge Robert in this domain is that if your measure run does not good enough it doesn't matter what your model."
  },
  {
    "start": 1960.26,
    "end": 1962.9,
    "text": "So a CNC machine... If you think about it right?"
  },
  {
    "start": 1963.18,
    "end": 1967.32,
    "text": "It's actually operating at about a thousand hertz or ten-thousand hertz kind of frequency."
  },
  {
    "start": 1967.94,
    "end": 1969.38,
    "text": "and then on the access level Right!"
  },
  {
    "start": 1969.66,
    "end": 1978.94,
    "text": "At the motor level The data streams which can get access to In general..if You're Not really going too deep would be maybe at the one hertz ten hertz hundred Hertz Level."
  },
  {
    "start": 1979.2,
    "end": 1983.14,
    "text": "So if you want to go higher frequency data, You wanna get more higher sensitivity data."
  },
  {
    "start": 1983.6,
    "end": 1987.92,
    "text": "You may need a little beyond what the machine they give out of box?"
  },
  {
    "start": 1988.26,
    "end": 1993.16,
    "text": "It's not doing academic science though these are well proven technologies and all that."
  },
  {
    "start": 1993.44,
    "end": 1996.34,
    "text": "but integration is where our focus is."
  },
  {
    "start": 1996.86,
    "end": 1997.28,
    "text": "Absolutely!"
  },
  {
    "start": 1997.84,
    "end": 2007.48,
    "text": "Integration is good topic because your software or model is integrated directly into the CAM environment such as Siemens NX or Mastercam?"
  },
  {
    "start": 2008.84,
    "end": 2009.28,
    "text": "Yes, it is."
  },
  {
    "start": 2009.76,
    "end": 2009.9,
    "text": "Okay?"
  },
  {
    "start": 2010.42,
    "end": 2013.52,
    "text": "Why is native integration critical for our adoption?"
  },
  {
    "start": 2013.64,
    "end": 2014.16,
    "text": "the shop floor"
  },
  {
    "start": 2014.52,
    "end": 2016.68,
    "text": "I mean i just think of its consumer behavior."
  },
  {
    "start": 2016.76,
    "end": 2018.28,
    "text": "you don't want people to have to change their behavior."
  },
  {
    "start": 2018.76,
    "end": 2024.26,
    "text": "So if our target user persona on the CNC programming site is a CNC programmer, where do they live and breathe every single day?"
  },
  {
    "start": 2024.4,
    "end": 2024.72,
    "text": "Scan."
  },
  {
    "start": 2025.52,
    "end": 2030.48,
    "text": "And then down steam in the shop floor... Our target persona as an operator or the shop manager?"
  },
  {
    "start": 2030.96,
    "end": 2033.92,
    "text": "well you know we need integration with the shop flow right."
  },
  {
    "start": 2034.12,
    "end": 2044.2,
    "text": "so integration I think for industrial technologies it's just kind of inherent to adoption..and i think that integration goes beyond the workflow your focused on!"
  },
  {
    "start": 2044.58,
    "end": 2052.54,
    "text": "For us when thinking about integrations at the PLM level this management level and the quality management system, but I think that's a good thing."
  },
  {
    "start": 2052.84,
    "end": 2058.76,
    "text": "By the end of days you take what digitization for shop floor or factory companies have invested millions of dollars in CapEx."
  },
  {
    "start": 2059.28,
    "end": 2061.58,
    "text": "You'll need to unlock more value."
  },
  {
    "start": 2063.62,
    "end": 2065.739,
    "text": "so those integrations really help facilitate"
  },
  {
    "start": 2065.86,
    "end": 2066.94,
    "text": "it Mm-hmm."
  },
  {
    "start": 2067.02,
    "end": 2070.739,
    "text": "Where do your customers typically see the strongest measurable impact?"
  },
  {
    "start": 2070.8,
    "end": 2075.82,
    "text": "Is it programming time, cycle times to lifetime scrap reduction?"
  },
  {
    "start": 2075.92,
    "end": 2079.94,
    "text": "What is them when you talk to our customers and they say yeah this great"
  },
  {
    "start": 2080.48,
    "end": 2080.9,
    "text": "I like that."
  },
  {
    "start": 2080.96,
    "end": 2082.04,
    "text": "You're basically articulating."
  },
  {
    "start": 2082.08,
    "end": 2087.54,
    "text": "maybe eighty percent of which is good But you know i think The most measurable impact a robot comes from."
  },
  {
    "start": 2088.38,
    "end": 2092.06,
    "text": "Programming throughput is super important."
  },
  {
    "start": 2092.12,
    "end": 2093.8,
    "text": "on the pro upstream workflow."
  },
  {
    "start": 2094.48,
    "end": 2101.16,
    "text": "Accuracy, of course at the cycle or quality of toolpaths is very important because you could be fast but if it's not quality means nothing."
  },
  {
    "start": 2101.72,
    "end": 2104.42,
    "text": "And then downstream cycle time optimization."
  },
  {
    "start": 2104.5,
    "end": 2106.2,
    "text": "we see low hanging fruit there."
  },
  {
    "start": 2107.26,
    "end": 2111.7,
    "text": "We see a lot success Very quick ROI and demonstration capabilities."
  },
  {
    "start": 2112.62,
    "end": 2124.24,
    "text": "Then more complex topics like tool life, tool management, tool cost management Scrap and rework reduction All those are yes very valuable metrics as well to drive over performance for a shop."
  },
  {
    "start": 2124.3,
    "end": 2132.26,
    "text": "But I think if you ask me, the lowest hanging food in The Biggest Bank for your buck is programming time reduction and cycle time optimization."
  },
  {
    "start": 2132.68,
    "end": 2137.8,
    "text": "when it comes to cat cam workflows there's an individual programmer experience right?"
  },
  {
    "start": 2138.24,
    "end": 2142.86,
    "text": "How does your AI addresses knowledge standardization across teams?"
  },
  {
    "start": 2142.96,
    "end": 2149.42,
    "text": "because When i go to the industry to the shop floor It differs between people who are working with machines."
  },
  {
    "start": 2149.52,
    "end": 2150.84,
    "text": "or am I wrong business?"
  },
  {
    "start": 2151.34,
    "end": 2151.62,
    "text": "No, no."
  },
  {
    "start": 2151.8,
    "end": 2152.58,
    "text": "A hundred percent right."
  },
  {
    "start": 2152.62,
    "end": 2152.78,
    "text": "I mean."
  },
  {
    "start": 2152.98,
    "end": 2160.84,
    "text": "the classical saying is know two machinists will agree on what their answer and you give a team of ten programmers The same standards in different recipes Right?"
  },
  {
    "start": 2161.36,
    "end": 2167.2,
    "text": "So for us from day one that's just been part of our design philosophy because In fact both machines agree On the right answers."
  },
  {
    "start": 2167.26,
    "end": 2168.86,
    "text": "we don't Give them One right Answer."
  },
  {
    "start": 2169.16,
    "end": 2177.08,
    "text": "We give Them five Different options And we let the user be in control The user being the expert, they know that last mile context."
  },
  {
    "start": 2177.18,
    "end": 2180.32,
    "text": "Why does user A prefer option a versus user B prefer option b?"
  },
  {
    "start": 2180.6,
    "end": 2181.08,
    "text": "We don't know."
  },
  {
    "start": 2181.48,
    "end": 2185.24,
    "text": "but we can learn from it and personalize our results over time."
  },
  {
    "start": 2186.12,
    "end": 2186.94,
    "text": "That's how you handle it."
  },
  {
    "start": 2187.02,
    "end": 2191.06,
    "text": "And then of course learning from experts to bring knowledge into novices."
  },
  {
    "start": 2191.46,
    "end": 2195.92,
    "text": "When you"
  },
  {
    "start": 2197.2,
    "end": 2199.94,
    "text": "talk to guys on the shop floor, what CNC programming?"
  },
  {
    "start": 2200.0,
    "end": 2201.24,
    "text": "they often struggle with."
  },
  {
    "start": 2201.72,
    "end": 2207.74,
    "text": "inconsistent tool libraries machine variability non-standardized data structures."
  },
  {
    "start": 2207.82,
    "end": 2209.18,
    "text": "How do we handle this?"
  },
  {
    "start": 2209.98,
    "end": 2212.76,
    "text": "Yeah that's I mean... That is a reality of the world we live in."
  },
  {
    "start": 2213.06,
    "end": 2217.64,
    "text": "Data structures and just kind of having a baseline data infrastructure."
  },
  {
    "start": 2217.92,
    "end": 2219.94,
    "text": "i think it critical for any AI adoption."
  },
  {
    "start": 2220.04,
    "end": 2222.56,
    "text": "so.. You know I think AI comes five steps later."
  },
  {
    "start": 2223.62,
    "end": 2236.98,
    "text": "So, I think the sophistication of data architecture and ITOT architecture in any customer environment is a prerequisite to think about as part of adoption discussion with an organization."
  },
  {
    "start": 2237.5,
    "end": 2238.96,
    "text": "To your point it just varies a lot."
  },
  {
    "start": 2239.88,
    "end": 2242.46,
    "text": "some organizations are much further along than others."
  },
  {
    "start": 2242.5,
    "end": 2249.8,
    "text": "Some have enhanced capabilities but you don't You kind-of have to assess all that In order then come up with right approach."
  },
  {
    "start": 2251.34,
    "end": 2254.4,
    "text": "So you'll hear me say this a lot, but I'm a strong believer."
  },
  {
    "start": 2254.68,
    "end": 2267.36,
    "text": "Given the current state of technology and the current stage in the market's maturity to adopting such technologies... The near term for industrial AI is very consultative sales."
  },
  {
    "start": 2269.08,
    "end": 2275.7,
    "text": "You're a partner with customers on their journey of adopting these technologies as opposed just having them run with it."
  },
  {
    "start": 2276.28,
    "end": 2279.26,
    "text": "So your AI gives recommendations?"
  },
  {
    "start": 2280.62,
    "end": 2288.84,
    "text": "How do you ensure that AI recommendations remain transparent and controllable in a safety, critical machining environment?"
  },
  {
    "start": 2310.06,
    "end": 2315.52,
    "text": "transparent in our decision making as we move into topics like Neurosymbolic capabilities."
  },
  {
    "start": 2316.24,
    "end": 2317.02,
    "text": "We have a little problem,"
  },
  {
    "start": 2319.38,
    "end": 2322.2,
    "text": "but I think there's lot of work going on for AI community right?"
  },
  {
    "start": 2322.3,
    "end": 2352.1,
    "text": "For AI transparency even deep learning scenarios...I wouldn't say that we've fully cracked it for our domain..but we are treading it cautiously and the answer to this might also end up depending industry vertical where this adopted right maybe a domain like defense needs high traceability, high audit ability and thereby you actually might run much more deterministic models as opposed to just purely generative models."
  },
  {
    "start": 2353.04,
    "end": 2354.2,
    "text": "And other domains."
  },
  {
    "start": 2355.84,
    "end": 2356.56,
    "text": "Okay,"
  },
  {
    "start": 2358.22,
    "end": 2359.56,
    "text": "looking ahead at the end."
  },
  {
    "start": 2359.66,
    "end": 2361.18,
    "text": "What are from your point of view?"
  },
  {
    "start": 2361.28,
    "end": 2366.72,
    "text": "The biggest technical barriers that still separate a I assisted programing form?"
  },
  {
    "start": 2366.88,
    "end": 2369.04,
    "text": "full autonomous CNC machining"
  },
  {
    "start": 2369.62,
    "end": 2369.98,
    "text": "lots?"
  },
  {
    "start": 2371.98,
    "end": 2374.12,
    "text": "You answer you're question work back man."
  },
  {
    "start": 2374.54,
    "end": 2376.6,
    "text": "Yeah mean this is something we already actively thinking about."
  },
  {
    "start": 2376.7,
    "end": 2377.5,
    "text": "where were working on it?"
  },
  {
    "start": 2377.56,
    "end": 2381.56,
    "text": "But let me explain to how he think about if you know what over now embodied AI systems as how we?"
  },
  {
    "start": 2382.46,
    "end": 2388.58,
    "text": "So if it's about an embodied AI system for precision manufacturing, we borrow a lot of the learnings from the research coming out of Stanford."
  },
  {
    "start": 2388.7,
    "end": 2389.92,
    "text": "and what have you."
  },
  {
    "start": 2389.96,
    "end": 2397.26,
    "text": "But there is this entire causal inference pillar that I think has so much more work to be done in our domain for causal inference."
  },
  {
    "start": 2397.66,
    "end": 2411.3,
    "text": "It's an entire topic around open world ML That needs to brought-in And i think What we are doing at Lambda Function gets us a step There because We Are Bringing In Feedback From The Real World on that side."
  },
  {
    "start": 2411.54,
    "end": 2420.22,
    "text": "I think trustworthiness is like going back to your question of auditability, there's so much more work that needs to be done from a trustworthy nest lens."
  },
  {
    "start": 2420.58,
    "end": 2431.1,
    "text": "So again several pillars have to continue coming together as we move incrementally moved towards this vision for precision"
  },
  {
    "start": 2431.14,
    "end": 2431.36,
    "text": "production."
  },
  {
    "start": 2433.24,
    "end": 2434.28,
    "text": "What is in the agenda?"
  },
  {
    "start": 2434.44,
    "end": 2436.8,
    "text": "Please share what are you doing next three months"
  },
  {
    "start": 2437.78,
    "end": 2440.46,
    "text": "Next three months, so a lot of focus for us."
  },
  {
    "start": 2441.92,
    "end": 2446.44,
    "text": "One side is the go-to market side working with our customers really making them successful."
  },
  {
    "start": 2446.76,
    "end": 2448.22,
    "text": "that's one focus on it."
  },
  {
    "start": 2448.38,
    "end": 2459.72,
    "text": "and from technical standpoint we are really focusing on this neuro symbolic capabilities site And unlocking those agentic workflows that have now become more complex in other industries."
  },
  {
    "start": 2459.78,
    "end": 2470.68,
    "text": "but just enabling our systems to operate in these more agenetic environments I would say are the nearest term priorities for us."
  },
  {
    "start": 2471.24,
    "end": 2473.4,
    "text": "So we will see a CNC agent?"
  },
  {
    "start": 2474.34,
    "end": 2483.54,
    "text": "Right now, yeah like in terms of programming agents and machining agents where you know working on quality agents as an example with downstream use cases."
  },
  {
    "start": 2485.08,
    "end": 2491.5,
    "text": "but there's still more that needs to be done to truly unlock more agente kind capabilities."
  },
  {
    "start": 2491.56,
    "end": 2500.34,
    "text": "But i think it is right stepping stone to get into the point where we can then start talking about, you know more like world model type of topics and antibody topic."
  },
  {
    "start": 2500.42,
    "end": 2503.66,
    "text": "So I think for us that's kind of a natural direction as we see it."
  },
  {
    "start": 2505.42,
    "end": 2508.62,
    "text": "but when we talk about world models than we talked about temporal right?"
  },
  {
    "start": 2508.68,
    "end": 2519.84,
    "text": "We talk about memory at the end because existing technology When It Comes To Existing Algorithms Are Lacking When It comes to Temporal Dependency And Memory."
  },
  {
    "start": 2520.38,
    "end": 2529.26,
    "text": "so do You See There That for your special business, do you see that we need to think from a physical AI perspective?"
  },
  {
    "start": 2529.32,
    "end": 2532.16,
    "text": "To focus different algorithms new architectures."
  },
  {
    "start": 2532.54,
    "end": 2533.36,
    "text": "Do You need them?"
  },
  {
    "start": 2533.48,
    "end": 2535.3,
    "text": "Do you look into them?"
  },
  {
    "start": 2535.5,
    "end": 2538.12,
    "text": "or is it something you say oh We take what we get?"
  },
  {
    "start": 2538.6,
    "end": 2547.36,
    "text": "no I mean II think we are looking at It very Very consciously and this point due to figure out What we can borrow."
  },
  {
    "start": 2548.22,
    "end": 2550.32,
    "text": "but where the limits of what we could borrow."
  },
  {
    "start": 2550.4,
    "end": 2550.94,
    "text": "And then well how do?"
  },
  {
    "start": 2552.12,
    "end": 2562.18,
    "text": "And I think the complexity is when you start combining these two worlds of, you know... ...the purely simulated theoretical world of CAM with a real-world of the shop floor."
  },
  {
    "start": 2562.54,
    "end": 2569.44,
    "text": "But that convergence is necessary in order to truly unlock next level autonomy here."
  },
  {
    "start": 2569.94,
    "end": 2570.72,
    "text": "So yeah!"
  },
  {
    "start": 2572.96,
    "end": 2583.62,
    "text": "We're constantly trying to stay abreast for research coming out from other domains That can be borrowed and then testing it against our domain rapidability."
  },
  {
    "start": 2584.0,
    "end": 2584.46,
    "text": "Which domains?"
  },
  {
    "start": 2584.52,
    "end": 2586.72,
    "text": "robotics or what do you mean with domains?"
  },
  {
    "start": 2587.08,
    "end": 2596.02,
    "text": "Yeah, I mean in the research that's coming out on the global models are our systems for other domains like autonomous vehicles or human rights?"
  },
  {
    "start": 2597.16,
    "end": 2597.86,
    "text": "Okay, okay."
  },
  {
    "start": 2598.02,
    "end": 2598.5,
    "text": "Thanks a lot!"
  },
  {
    "start": 2598.7,
    "end": 2599.38,
    "text": "It was a pleasure."
  },
  {
    "start": 2599.44,
    "end": 2602.0,
    "text": "I keep my fingers crossed for you and your company for Solution."
  },
  {
    "start": 2602.06,
    "end": 2610.56,
    "text": "And we're really looking forward to meet in Europe maybe at some conferences or exhibitions... I'm really looking foward."
  },
  {
    "start": 2610.64,
    "end": 2611.7,
    "text": "thanks alot for the interview."
  },
  {
    "start": 2612.36,
    "end": 2612.94,
    "text": "Absolutely Rob!"
  },
  {
    "start": 2613.12,
    "end": 2614.8,
    "text": "Pleasure."
  }
]