1 00:00:00,750 --> 00:00:04,469 Do efficiency and risk management top your generative AI roadmap? 2 00:00:05,130 --> 00:00:09,449 While most law firms still stall over hallucinations, one attorney is 3 00:00:09,449 --> 00:00:13,049 already billing more after automating his workflow to draft court ready 4 00:00:13,049 --> 00:00:15,360 arguments before anyone's first coffee. 5 00:00:15,930 --> 00:00:18,779 My name is Alexandre Nevski and this is Innovation Tales. 6 00:00:19,189 --> 00:00:21,769 Navigating Change one story at a time. 7 00:00:21,799 --> 00:00:25,999 We share insights from leaders tackling the challenges of today's digital world. 8 00:00:26,269 --> 00:00:29,779 Welcome to Innovation Tales, the podcast exploring the human 9 00:00:29,779 --> 00:00:31,639 side of digital transformation. 10 00:00:32,900 --> 00:00:36,560 So who is the lawyer turning predawn coffee into billable hours? 11 00:00:37,040 --> 00:00:41,510 Meet Dr. Don Ray Simmonds, an attorney with over a decade of experience 12 00:00:41,510 --> 00:00:45,470 specializing in intellectual property and the owner of one of the oldest 13 00:00:45,470 --> 00:00:47,150 trademark law firms in the UK. 14 00:00:48,065 --> 00:00:52,475 With only a basic understanding of coding, he stitched a set of LLM prompts in drag 15 00:00:52,475 --> 00:00:57,365 and drop tools like Make.com and n8n to spin up Bron AI, a SaaS that draft 16 00:00:57,365 --> 00:01:01,535 arguments in minutes and earns revenue while he spends more time with his son. 17 00:01:02,555 --> 00:01:06,875 In today's episode, we unpack Don's low-code blueprint, the guardrails 18 00:01:06,875 --> 00:01:11,315 that keep hallucinations out of client documents and why documenting your prompts 19 00:01:11,315 --> 00:01:13,865 now, could save an IP lawsuit later. 20 00:01:14,675 --> 00:01:17,790 Without further ado, here's my conversation with Dr. Simmonds. 21 00:01:18,326 --> 00:01:19,376 Don, welcome to the show. 22 00:01:19,639 --> 00:01:20,519 Hello, Alex. 23 00:01:20,519 --> 00:01:21,539 Thank you for having me. 24 00:01:22,209 --> 00:01:23,284 Oh, it's great to have you. 25 00:01:23,854 --> 00:01:28,684 Now, I want to start with a story I know our audience will find fascinating. 26 00:01:29,644 --> 00:01:33,394 A couple of years ago, if I'm not mistaken, you started an AI business. 27 00:01:33,904 --> 00:01:36,184 Can you tell us what happened and what led you there? 28 00:01:36,807 --> 00:01:41,127 Yes, and I hope they find it fascinating, but it's definitely changed my 29 00:01:41,127 --> 00:01:43,557 life, so I can say that for sure. 30 00:01:44,127 --> 00:01:49,422 I am a trademark attorney and I've been a trademark attorney for 10 years. 31 00:01:50,142 --> 00:01:55,290 And for the last five years I was asked to do what we call office action 32 00:01:55,290 --> 00:02:00,630 responses, which means give you, I'm gonna give you layman's sort of explanation. 33 00:02:00,930 --> 00:02:01,330 Yes, please. 34 00:02:01,630 --> 00:02:07,350 If you apply for a trademark and you get refused, you get sent a 35 00:02:07,350 --> 00:02:09,060 notice, a letter or something. 36 00:02:09,780 --> 00:02:12,180 This is called an office action response, and they're 37 00:02:12,180 --> 00:02:14,540 telling you why you're refused. 38 00:02:15,700 --> 00:02:19,545 And it could be because there's a trademark already that is the 39 00:02:19,545 --> 00:02:23,025 same as the one you're registering or something along those lines. 40 00:02:23,895 --> 00:02:30,525 Usually when that happens, depending on where you file, you argue, you have legal 41 00:02:30,525 --> 00:02:36,145 arguments presented saying, Hey, this is why you should accept the trademark. 42 00:02:36,525 --> 00:02:44,235 So clients usually will ask me to respond and say, Hey, argue in our 43 00:02:44,235 --> 00:02:45,905 favor, and hopefully they will accept. 44 00:02:46,565 --> 00:02:51,605 However, it got to a point that I was good enough that I was being 45 00:02:51,605 --> 00:02:56,525 hired by another law firm to do their office action responses. 46 00:02:56,525 --> 00:03:00,605 So not only I was dealing with mine, which wasn't that mine weren't, I 47 00:03:00,672 --> 00:03:04,865 wasn't getting many, the idea is to hopefully get none when you apply. 48 00:03:05,345 --> 00:03:09,545 But if you work with a bigger firm, they will have more clients and 49 00:03:09,545 --> 00:03:11,745 thus they will have more responses. 50 00:03:13,105 --> 00:03:19,510 So what was happening is I was getting a lot more of these and 51 00:03:19,510 --> 00:03:23,800 my son was getting bigger, so I was spending more time with him. 52 00:03:24,250 --> 00:03:29,310 And thus what was happening is I was not able to keep up with the amount of 53 00:03:29,620 --> 00:03:33,730 office action responses I was supposed to draft, because my, I was spending 54 00:03:33,730 --> 00:03:37,090 more and more time with my son during the summer and I figured, you know what? 55 00:03:37,090 --> 00:03:41,740 I have to figure out a way make it so that I can spend time with him, 56 00:03:41,980 --> 00:03:46,320 but yet at the same time respond to these ever increasing... Being good 57 00:03:46,320 --> 00:03:51,525 at something has its own negative effects, even though it's good for you. 58 00:03:51,545 --> 00:03:54,040 And that was, I was getting more know, work. 59 00:03:54,340 --> 00:03:54,535 Yeah. 60 00:03:54,535 --> 00:03:55,450 Good problems to have. 61 00:03:55,450 --> 00:03:55,930 Definitely. 62 00:03:56,230 --> 00:03:59,560 So the idea was, okay, in my head I was thinking, how can I make 63 00:03:59,560 --> 00:04:00,940 this, how can I deal with this? 64 00:04:01,270 --> 00:04:10,000 And this is when I noticed that Open AI had invited me or some groups of 65 00:04:10,000 --> 00:04:12,320 people to join their beta accounts. 66 00:04:12,350 --> 00:04:16,370 So I didn't know it was for, I did not know what they would be giving. 67 00:04:16,590 --> 00:04:22,330 And this was like in September of the same year that they released 68 00:04:22,330 --> 00:04:25,620 ChatGPT, and I think they released it in December or November. 69 00:04:26,100 --> 00:04:30,000 So I applied, got in, didn't think anything of it, and I was still 70 00:04:30,000 --> 00:04:35,160 racking my brain trying to figure out how to get it so that I did my 71 00:04:35,160 --> 00:04:36,990 office action responses quicker. 72 00:04:37,230 --> 00:04:38,580 Normally, it would take me a day. 73 00:04:39,690 --> 00:04:41,880 Maybe a day and a half to draft a response. 74 00:04:41,880 --> 00:04:42,870 These are legal arguments. 75 00:04:42,870 --> 00:04:43,493 It takes time. 76 00:04:43,530 --> 00:04:44,310 You gotta study. 77 00:04:45,060 --> 00:04:47,880 Even if you have a template to work from, you still, it 78 00:04:47,880 --> 00:04:49,200 still takes time to do, right? 79 00:04:49,620 --> 00:04:55,300 So, lo and behold, December, November comes around and they release ChatGPT. 80 00:04:55,710 --> 00:04:58,050 And of course you might know the story. 81 00:04:58,050 --> 00:04:59,130 Everyone freaks out. 82 00:04:59,460 --> 00:05:03,670 The highest subscription rate ever into I think a hundred million 83 00:05:03,670 --> 00:05:05,890 people and the shortest time ever. 84 00:05:06,220 --> 00:05:10,300 So I was one of the earliest people because I had an account with OpenAI. 85 00:05:11,380 --> 00:05:14,950 And that's where the light bulb went off in my head, as they say. 86 00:05:16,210 --> 00:05:21,160 After working with it, I realized that I could use it to draft 87 00:05:21,220 --> 00:05:23,380 or help me draft arguments. 88 00:05:23,440 --> 00:05:25,090 I just didn't know how. 89 00:05:25,450 --> 00:05:28,660 It was like I couldn't, I had to do it in a way that it was automated. 90 00:05:29,395 --> 00:05:33,445 And then I later realized there were other websites that allowed you to 91 00:05:33,445 --> 00:05:36,505 connect to ChatGPT through an API. 92 00:05:36,805 --> 00:05:38,455 One of them was Make.com. 93 00:05:38,755 --> 00:05:43,465 There's all, there are others, like n8n, was Make.com and there was other, 94 00:05:43,465 --> 00:05:45,025 there was another one, Zapier I think. 95 00:05:45,415 --> 00:05:48,325 It was Make.com that gave me the most flexibility at the time, 96 00:05:48,685 --> 00:05:54,535 and I was more, and I had a basic understanding of coding and JSON files. 97 00:05:54,535 --> 00:05:59,005 I kinda learned myself, so I was like, okay, if I can use this platform, I 98 00:05:59,005 --> 00:06:04,415 can actually create a beast that would help me draft even if, even if it 99 00:06:04,415 --> 00:06:06,335 only helps me, I will be very happy. 100 00:06:06,755 --> 00:06:12,995 So I set to work late nights, ripping my hair out, as you can see. 101 00:06:13,935 --> 00:06:20,222 And it took months doing innovation after in different designs and trying to figure 102 00:06:20,222 --> 00:06:21,542 out, and there were a lot of failures. 103 00:06:21,542 --> 00:06:24,152 But by the third month I had something. 104 00:06:24,702 --> 00:06:30,032 It was automating a draft fully some, and then I was like, okay, I can 105 00:06:30,032 --> 00:06:32,052 use this draft personally for me. 106 00:06:32,802 --> 00:06:39,092 And I started using it to see if it work, it would work as that I could send to 107 00:06:39,092 --> 00:06:44,792 the office, the US or wherever the refusal came from, to see if they would respond 108 00:06:44,792 --> 00:06:46,562 in kind and say, Hey, you know what? 109 00:06:46,562 --> 00:06:47,852 Your arguments are valid. 110 00:06:47,852 --> 00:06:50,822 We're going to accept them and register the trademark. 111 00:06:51,152 --> 00:06:55,322 And I had to do this first just to make sure if it doesn't work, I can't use it. 112 00:06:55,352 --> 00:06:55,892 Obviously. 113 00:06:55,942 --> 00:07:00,202 If the arguments are not strong enough, then of course they would not accept them, 114 00:07:00,202 --> 00:07:01,612 and then that would be the end of that. 115 00:07:02,092 --> 00:07:04,672 I knew it should work because the arguments are the same that 116 00:07:04,702 --> 00:07:08,212 I would normally write, but this time I did not write them. 117 00:07:08,272 --> 00:07:09,692 It was done by the AI. 118 00:07:10,382 --> 00:07:15,352 So, three months in I send them off, and during the summer of, that 119 00:07:15,352 --> 00:07:17,292 was two years ago, I was using it. 120 00:07:17,562 --> 00:07:19,462 It allowed me more time with my son. 121 00:07:20,192 --> 00:07:24,632 And at the same time it was, it basically, it took what took a 122 00:07:24,632 --> 00:07:27,002 full day down to about two hours. 123 00:07:27,662 --> 00:07:29,152 So I really gained a lot. 124 00:07:29,922 --> 00:07:36,532 And I was like, if this works for me, is it possible that I can make it so 125 00:07:36,532 --> 00:07:39,232 it works for some other attorneys? 126 00:07:39,622 --> 00:07:45,142 It took another couple of three months or so after that and I made a version that 127 00:07:45,522 --> 00:07:51,292 is more, not so much for me, but more standard that almost any attorney can use. 128 00:07:51,742 --> 00:07:56,212 And it's just to sort of summarize the story, from that moment on, I was using 129 00:07:56,212 --> 00:08:01,072 it with the firms that were hiring me, and I was allowing another attorney to use 130 00:08:01,072 --> 00:08:06,052 it as a beta tester because I just wanted to see was there any errors and all that. 131 00:08:06,052 --> 00:08:08,512 So she used it for a while for free just to make sure it worked. 132 00:08:08,692 --> 00:08:09,712 She was happy with it. 133 00:08:10,192 --> 00:08:14,742 And now we're at that stage where, yeah, I have regular attorneys using 134 00:08:14,742 --> 00:08:20,542 it and paying for it and I'm making nice income on the side with it. 135 00:08:20,872 --> 00:08:23,992 And the idea is to expand it to other attorneys to use. 136 00:08:24,112 --> 00:08:30,262 And it's been a whirlwind year but we are very happy and it was making 137 00:08:30,262 --> 00:08:32,062 money from the start, like right away. 138 00:08:32,356 --> 00:08:34,446 Wow, what a wonderful story! 139 00:08:34,732 --> 00:08:40,837 I also wanted to ask you for more general overview of where the latest 140 00:08:40,837 --> 00:08:45,172 wave of artificial intelligence tools, so mostly generative AI, is being 141 00:08:45,172 --> 00:08:46,882 applied in the legal profession. 142 00:08:46,932 --> 00:08:52,952 So the one that you've just described is, I guess drafting some of the documents 143 00:08:52,952 --> 00:08:55,662 that you may file , that's clear one. 144 00:08:55,722 --> 00:09:02,832 I guess it's also quite useful to stay abreast of evolving regulations. 145 00:09:02,872 --> 00:09:09,531 The capabilities of the large language models with the latest ability to 146 00:09:09,651 --> 00:09:13,721 ingest a lot of different documents and then give you a summary, that 147 00:09:13,721 --> 00:09:15,611 is also a very useful use case. 148 00:09:15,611 --> 00:09:21,586 And the other one I've seen is when you have a case and a lot of different 149 00:09:22,106 --> 00:09:26,006 pieces that fit into the case: some evidence documents or whatever. 150 00:09:26,366 --> 00:09:32,266 That it also helps the legal professional to save time in getting 151 00:09:32,266 --> 00:09:34,096 a quick overview of these documents. 152 00:09:34,096 --> 00:09:38,676 So is that like a fair summary of the benefits and the use cases? 153 00:09:38,916 --> 00:09:39,571 Did I miss any? 154 00:09:40,032 --> 00:09:40,692 There are more. 155 00:09:40,812 --> 00:09:44,142 but yeah, that's a great summary of what is out there now. 156 00:09:44,343 --> 00:09:48,433 We do focus mostly on what would happen is the attorney would fill 157 00:09:48,433 --> 00:09:52,693 in a form then the AI would draft the arguments for the attorney. 158 00:09:53,083 --> 00:09:55,953 But there is, the other thing that you've mentioned. 159 00:09:56,513 --> 00:09:58,493 Let's focus on a case. 160 00:09:58,843 --> 00:10:03,943 And the reason why cases are harder to draft is because each case is unique. 161 00:10:04,183 --> 00:10:11,593 So evidence might be used in this, at this stage, or different evidence 162 00:10:11,593 --> 00:10:12,733 might be used at this stage. 163 00:10:12,733 --> 00:10:18,143 So you can't, there's no sort of specific, formula that works very well. 164 00:10:18,623 --> 00:10:24,783 But you do have a platform like NotebookLM by Google, which has the ability to 165 00:10:24,783 --> 00:10:30,903 bring in PDFs, bring in websites, and then you can ask it to draft arguments 166 00:10:31,233 --> 00:10:36,303 based on the evidence that you bring it, which I think is perfect for cases. 167 00:10:36,673 --> 00:10:42,203 And it's also perfect for, so let's say you're drafting agreements or 168 00:10:42,203 --> 00:10:44,543 privacy policies or terms of service. 169 00:10:44,873 --> 00:10:50,373 This, again, is good for platforms like Google, because you can have 170 00:10:50,373 --> 00:10:54,483 a template and then you could say, using this template draft a new 171 00:10:54,483 --> 00:10:58,863 policy, but this is the business information that I want you to take. 172 00:10:59,133 --> 00:11:01,083 And apply to this template that you have. 173 00:11:01,413 --> 00:11:06,183 And it will do it very quickly, much faster than using a paralegal by hand. 174 00:11:06,642 --> 00:11:09,672 These are one of the cases you can use a differently. 175 00:11:10,302 --> 00:11:14,442 And then there's just looking through large, like reading large cases. 176 00:11:14,442 --> 00:11:19,512 Like getting information on an agreement or a case and you want it quickly. 177 00:11:19,822 --> 00:11:23,752 I think one of the guests earlier in the season was also mentioning like 178 00:11:23,752 --> 00:11:30,072 in a corporate environment to identify variations between standard contracts. 179 00:11:30,122 --> 00:11:33,682 Contracts that are supposed to be standard, but then as a result of 180 00:11:34,222 --> 00:11:38,102 business negotiations sometimes you, have to change a clause or two. 181 00:11:38,102 --> 00:11:43,512 So then having an overview of how many differences there are in your contracts, 182 00:11:43,512 --> 00:11:45,072 I think, that was another use case. 183 00:11:45,072 --> 00:11:46,962 But you mentioned also that there might be others. 184 00:11:46,962 --> 00:11:49,847 So I'm really keen to hear more use cases. 185 00:11:49,897 --> 00:11:52,297 There's the one I'm working on right now that's, I think it's 186 00:11:52,297 --> 00:11:55,177 even, yeah, much bigger or better. 187 00:11:55,957 --> 00:12:02,797 The legal services business requires to get better at automating things because a 188 00:12:02,797 --> 00:12:05,497 lot of things we do are very repetitive. 189 00:12:06,487 --> 00:12:12,457 The idea is to have a paralegal that has access to your full database of 190 00:12:12,457 --> 00:12:19,272 clients and your fees, your terms of service, and then also have access to, 191 00:12:19,332 --> 00:12:23,182 if needed, case law, if you need it. 192 00:12:23,212 --> 00:12:24,892 Sometimes you don't want it to do that. 193 00:12:24,982 --> 00:12:28,492 Sometimes you just want it to be the administrative staff 194 00:12:28,492 --> 00:12:30,022 that keeps you on top of cases. 195 00:12:30,022 --> 00:12:34,542 So what would happen is this automated AI not automated, I'd 196 00:12:34,572 --> 00:12:36,102 just call it a live paralegal. 197 00:12:36,562 --> 00:12:41,392 If done right, you would basically have every person on your, in your law 198 00:12:41,392 --> 00:12:45,142 firm able to communicate with her, it. 199 00:12:45,202 --> 00:12:47,992 I call her, but it and 200 00:12:48,152 --> 00:12:49,112 You're not the first one. 201 00:12:50,987 --> 00:12:56,467 Ask questions 'cause I get a lot of questions as the top, the lead attorney 202 00:12:56,947 --> 00:12:58,997 and it takes a lot of time out of my day. 203 00:12:59,417 --> 00:13:04,757 If I have a, an AI paralegal that can answer all the questions that my staff 204 00:13:05,087 --> 00:13:10,607 has and she responds, or it responds right away, that frees up my time. 205 00:13:10,997 --> 00:13:18,107 And this day and age, yes, we have, n8n specifically has a format that 206 00:13:18,107 --> 00:13:24,677 you could basically send the input via email or directly, but let's say email. 207 00:13:25,307 --> 00:13:31,057 And then once it's received, the AI on that platform, if you set it up correctly, 208 00:13:31,207 --> 00:13:36,217 will choose what is the best tool in order to respond to whatever question is asked. 209 00:13:36,427 --> 00:13:37,927 It will be the one deciding. 210 00:13:38,097 --> 00:13:38,897 You will not be. 211 00:13:39,017 --> 00:13:43,477 It will say, oh, I have to send, I have to draft an email for this client. 212 00:13:43,507 --> 00:13:44,897 Alright, this is the client. 213 00:13:44,897 --> 00:13:45,887 This is their address. 214 00:13:45,887 --> 00:13:46,847 I know where to find it. 215 00:13:47,177 --> 00:13:47,837 Email address. 216 00:13:47,867 --> 00:13:51,567 I know what type of email they would like because we have templates so 217 00:13:51,927 --> 00:13:54,417 it's gonna do that, and then it's gonna send it right back to you. 218 00:13:54,897 --> 00:13:55,797 That's what you want. 219 00:13:55,827 --> 00:13:57,687 'cause then you're freeing up time. 220 00:13:58,227 --> 00:13:59,037 It's quicker. 221 00:13:59,037 --> 00:14:01,857 The client is happier because they're getting their responses right away. 222 00:14:02,157 --> 00:14:03,897 So that is the next wave. 223 00:14:03,957 --> 00:14:06,117 And I'm, this is something I'm working on. 224 00:14:06,597 --> 00:14:11,707 And after that there's another step, which is you have that paralegal 225 00:14:11,757 --> 00:14:14,407 AI interact with the internet. 226 00:14:15,012 --> 00:14:16,332 There's many ways to do it. 227 00:14:16,332 --> 00:14:20,622 Perplexity has a, what we call a node that you can connect through 228 00:14:20,622 --> 00:14:27,152 n8n, but they also have MCP, which is this new communicative format for 229 00:14:27,312 --> 00:14:29,772 AIs to interact with the internet. 230 00:14:30,622 --> 00:14:34,412 And that would basically open up the world to this paralegal. 231 00:14:34,997 --> 00:14:40,487 For a, a legal professional, you're awfully up to speed with all the 232 00:14:41,007 --> 00:14:43,847 low-code options to build AI agents. 233 00:14:45,352 --> 00:14:49,962 And there's now A2A, which is Google's new protocol for agents 234 00:14:49,962 --> 00:14:51,342 communicating with each other. 235 00:14:51,342 --> 00:14:53,082 So it's going fast. 236 00:14:53,132 --> 00:14:53,792 Yeah, it is. 237 00:14:54,272 --> 00:14:58,232 I don't really want to go into the technical side of things too much but 238 00:14:58,232 --> 00:15:03,292 I do want to pick your brains about the more broader implication of all 239 00:15:03,292 --> 00:15:06,382 this generative AI tech on IP law. 240 00:15:06,652 --> 00:15:12,822 So what are some of the challenges that you see coming front 241 00:15:12,822 --> 00:15:14,742 and center in the last year? 242 00:15:15,112 --> 00:15:20,677 I think there, there are several in different areas at different stages. 243 00:15:21,487 --> 00:15:27,437 First is that it's going to be required for any trainee attorney to know AI. 244 00:15:27,747 --> 00:15:32,767 I do not see how any of them could get into this field without knowing it. 245 00:15:32,867 --> 00:15:37,547 It's gonna be a standard requirement and we are having trouble finding 246 00:15:38,117 --> 00:15:43,887 trainees right now that know even the basics of AI . That is not 247 00:15:43,887 --> 00:15:45,717 something that should be happening. 248 00:15:46,027 --> 00:15:49,777 But I guess the law schools are trying to catch up with that and 249 00:15:49,777 --> 00:15:52,717 then maybe implement courses that will show them how to use it. 250 00:15:53,317 --> 00:15:56,422 Like AI literacy, prompt engineering, and that sort of stuff. 251 00:15:56,537 --> 00:15:58,367 Knowing how to use it, knowing how to prompt. 252 00:15:58,787 --> 00:15:59,867 Why is it important? 253 00:15:59,867 --> 00:16:04,997 Because a lot of law firms have subscription to AI platforms that 254 00:16:04,997 --> 00:16:09,197 require the user to interface using you know, your standard AI 255 00:16:09,197 --> 00:16:11,577 interface to research case law. 256 00:16:12,327 --> 00:16:18,232 That brings me to the next potential issue and their conservative approach to things. 257 00:16:18,382 --> 00:16:20,782 It's part of being in law. 258 00:16:21,112 --> 00:16:26,962 You tend to not want to embrace the new and then instead embrace the old. 259 00:16:27,942 --> 00:16:32,442 The old way of doing things is usually the best way, kind of way, attitude, 260 00:16:32,442 --> 00:16:36,552 but it is, the one thing that is keeping a lot of law firms back 261 00:16:36,552 --> 00:16:39,282 is that this AI technology is new. 262 00:16:39,282 --> 00:16:40,092 It's not proven. 263 00:16:40,092 --> 00:16:43,752 We know there's errors, there's hallucinations, there's all these 264 00:16:43,752 --> 00:16:49,092 things, and thus, law firms tend to shy away from it, even though they 265 00:16:49,092 --> 00:16:51,732 could reduce their time by 80%. 266 00:16:52,732 --> 00:16:54,502 And the reason is they're afraid. 267 00:16:55,282 --> 00:16:58,252 Like, I don't want a fake case law in my argument. 268 00:16:58,252 --> 00:17:00,382 How do I prevent that from happening? 269 00:17:01,522 --> 00:17:02,992 That's not so hard to do. 270 00:17:03,362 --> 00:17:06,522 It's just that you definitely have to start doing some work 271 00:17:06,522 --> 00:17:08,172 there and learn how to do it. 272 00:17:09,352 --> 00:17:13,692 And what about the impact on specifically intellectual property, 273 00:17:14,197 --> 00:17:17,577 whether it's trademarks or patents or something like that. 274 00:17:18,057 --> 00:17:25,397 My very basic understanding is that there is like challenges I suppose, 275 00:17:25,907 --> 00:17:29,737 I'm not sure how to call 'em, with tools like Midjourney, obviously, 276 00:17:29,787 --> 00:17:35,577 or Dall-E or the newest model from ChatGPT to do text to image, which 277 00:17:35,577 --> 00:17:41,127 raise a whole bunch of questions about who does a generated image belong to, 278 00:17:41,187 --> 00:17:44,247 if it was then based on some inputs? 279 00:17:44,737 --> 00:17:45,607 Where do we stand today? 280 00:17:45,907 --> 00:17:49,887 All right, so what I recently heard is that the ruling, at least in the 281 00:17:49,887 --> 00:17:58,462 US is that the creation of something by an AI can be attributed to the 282 00:17:58,582 --> 00:18:05,582 person that created the AI, as long as it's proven that they had enough of 283 00:18:05,582 --> 00:18:10,252 an input or influence in the creation of whatever it is they created. 284 00:18:10,475 --> 00:18:11,825 What does enough mean? 285 00:18:12,485 --> 00:18:14,735 That's what the courts are gonna have to decide. 286 00:18:14,735 --> 00:18:20,335 But they're saying that yes, AI itself can't own anything. 287 00:18:20,335 --> 00:18:24,460 And maybe there will be a time years down the road, I would 288 00:18:24,460 --> 00:18:26,560 say, it might happen, who knows? 289 00:18:26,560 --> 00:18:30,670 But right now, if you use AI to help you create something, you have to 290 00:18:30,670 --> 00:18:36,120 show that you did enough in that realm to make it attributable to you. 291 00:18:37,620 --> 00:18:40,260 I know that doesn't really clearly, but 292 00:18:41,025 --> 00:18:44,075 No, of course, but this is where we are today. 293 00:18:44,585 --> 00:18:47,735 And, yeah, it feels a bit disorientating I suppose. 294 00:18:47,758 --> 00:18:54,328 How long do courts take to get to a position for people who 295 00:18:54,328 --> 00:18:58,408 are going to be using these tools to know where they stand? 296 00:18:58,438 --> 00:19:02,458 Are we talking about something that will unroll over just a year or two? 297 00:19:02,458 --> 00:19:03,808 Or does that take 10 years? 298 00:19:03,808 --> 00:19:05,758 How long, like in the past, how did it work? 299 00:19:05,883 --> 00:19:10,413 There is no standard length, but it usually takes many years because of 300 00:19:10,413 --> 00:19:18,003 the fact that you are learning how it affects different realms differently. 301 00:19:18,333 --> 00:19:23,593 So it could be one thing to create artwork, another thing to create music, 302 00:19:23,923 --> 00:19:30,623 another thing to write a book It affects so many different areas so differently. 303 00:19:30,623 --> 00:19:36,173 You will not have a definitive and maybe never have a definitive 304 00:19:36,173 --> 00:19:37,983 answer of how to attribute rights. 305 00:19:37,983 --> 00:19:43,503 But, I guess it shouldn't stop people from creating for that reason. 306 00:19:43,903 --> 00:19:49,673 Unfortunately, the legal field is very slow because all these things have to 307 00:19:49,673 --> 00:19:55,283 percolate through the courts to finally get a judgment that everyone could 308 00:19:55,283 --> 00:19:57,893 read and say, oh, this is how I react. 309 00:19:58,223 --> 00:20:00,393 But it takes such a long time. 310 00:20:00,773 --> 00:20:01,253 Yeah. 311 00:20:01,598 --> 00:20:06,128 So it sounds like for people who are using these tools to create content 312 00:20:06,128 --> 00:20:11,423 that is maybe published, there's gonna be gray areas, if I understood 313 00:20:11,428 --> 00:20:16,268 you correctly, for a few years, may, maybe many years, and maybe forever. 314 00:20:16,748 --> 00:20:19,613 If you are a content creator, what are the gray areas you need to be aware of? 315 00:20:20,613 --> 00:20:24,458 If you're going to use AI, that is gonna be a gray area right away. 316 00:20:24,903 --> 00:20:32,373 I would try and make it so that I am the one giving as much input as possible when 317 00:20:32,643 --> 00:20:35,643 I create something and documenting it. 318 00:20:35,693 --> 00:20:37,793 I did this part of this creation. 319 00:20:38,453 --> 00:20:41,453 Try and be more specific when you're using it. 320 00:20:41,858 --> 00:20:46,343 Get as much information in there when you're doing it, just to show how much 321 00:20:46,373 --> 00:20:48,323 you've been a part of the process. 322 00:20:48,653 --> 00:20:51,908 And not just say make me a photo of of whatever. 323 00:20:51,968 --> 00:20:52,208 Yeah. 324 00:20:52,883 --> 00:20:58,318 And would would a content creator, would they gain from doc that you said 325 00:20:58,348 --> 00:21:02,298 document, make sure that they contribute as much as possible and that they 326 00:21:02,298 --> 00:21:03,738 document their what their contribution. 327 00:21:03,978 --> 00:21:05,388 Would you publish this? 328 00:21:05,388 --> 00:21:08,508 Would you make it or send it to some office, or is it just 329 00:21:08,508 --> 00:21:09,918 okay to keep it in your files? 330 00:21:10,218 --> 00:21:10,878 Just keep it. 331 00:21:10,878 --> 00:21:11,328 Yeah. 332 00:21:11,718 --> 00:21:15,858 The whole point is to keep it just in case something comes up, somebody 333 00:21:15,858 --> 00:21:17,948 goes, oh, this is not really yours. 334 00:21:17,948 --> 00:21:23,543 You used AI to create and thus we can use it because AI doesn't have any rights. 335 00:21:23,823 --> 00:21:25,113 We can use it on our website. 336 00:21:25,163 --> 00:21:28,333 And then you would be the one trying to enforce your rights and 337 00:21:28,333 --> 00:21:29,983 say, look, this is what I drafted. 338 00:21:30,013 --> 00:21:34,873 This is what I used to make it, and if you don't stop, we'll go to court. 339 00:21:34,873 --> 00:21:37,573 And that hopefully the court will side with you because you have 340 00:21:37,783 --> 00:21:43,213 evidence showing that you did partake in most of the creative process. 341 00:21:43,888 --> 00:21:44,248 Thank you. 342 00:21:44,248 --> 00:21:47,938 I think it's a, it's very clear and practical advice that people can follow. 343 00:21:48,508 --> 00:21:52,318 So what would be the key takeaway when it comes to the impact of 344 00:21:52,318 --> 00:21:56,128 artificial intelligence on, on business and intellectual property? 345 00:21:56,928 --> 00:22:03,338 Well, the word on the street is, nobody's going to be hired in the future 346 00:22:03,368 --> 00:22:05,528 without some sort of AI background. 347 00:22:05,528 --> 00:22:13,068 So I would try to at least learn about AI and understand how it's 348 00:22:13,068 --> 00:22:14,538 going to impact my business. 349 00:22:14,848 --> 00:22:17,428 You don't have to go back to school or anything, but you 350 00:22:17,433 --> 00:22:19,443 should know how to use a prompt. 351 00:22:19,723 --> 00:22:22,413 That's like the very basic of all things. 352 00:22:22,413 --> 00:22:30,298 And believe it or not, the next wave of hirees or people joining 353 00:22:30,298 --> 00:22:32,638 your company may not be technical. 354 00:22:33,093 --> 00:22:35,338 They're likely going to be English majors. 355 00:22:35,938 --> 00:22:38,518 People who know how to use the language. 356 00:22:39,238 --> 00:22:41,523 That's really where it excels. 357 00:22:41,863 --> 00:22:43,513 Understanding how to use the language. 358 00:22:43,513 --> 00:22:44,833 And I did this with my son. 359 00:22:44,833 --> 00:22:49,161 He's just seven years old, but we built an app using just words. 360 00:22:49,903 --> 00:22:52,723 Because I knew how to prompt . And so there's an app online right 361 00:22:52,723 --> 00:22:55,903 now, and I'll tell you a little anecdote why there's an app. 362 00:22:57,043 --> 00:23:00,093 There's a lot of apps on Apple Store. 363 00:23:00,143 --> 00:23:04,193 But a lot of these apps are, they're free and then they show you a video 364 00:23:04,433 --> 00:23:05,993 every time you want to use, you know, 365 00:23:06,583 --> 00:23:06,873 Yeah. 366 00:23:07,013 --> 00:23:13,213 We love trivia and we were downloading these apps and, yeah, every 10, 367 00:23:13,453 --> 00:23:14,743 15 minutes we'd get a video. 368 00:23:14,743 --> 00:23:16,843 And these videos are not really nice for young kids. 369 00:23:17,263 --> 00:23:19,993 So I got fed up and so did he. 370 00:23:21,223 --> 00:23:25,303 You know what, I think we can create our own app right now with the way things are. 371 00:23:25,483 --> 00:23:28,853 So we set it up used cursor and we just talked. 372 00:23:29,213 --> 00:23:29,873 Well typed. 373 00:23:30,573 --> 00:23:35,293 And we said we wanted a trivia app that we both can use and we could play. 374 00:23:35,963 --> 00:23:38,523 And this is what, and we outlined it and blah, blah, blah. 375 00:23:38,853 --> 00:23:41,703 And then it created it and now it's online. 376 00:23:42,063 --> 00:23:46,623 And my son and I now use that app to have fun playing trivia. 377 00:23:46,863 --> 00:23:48,093 And this is what I mean. 378 00:23:48,393 --> 00:23:51,153 Anyone can make apps, anyone can make things now. 379 00:23:51,893 --> 00:23:55,443 And so your business model might be disrupted by a kid in a room at 380 00:23:55,443 --> 00:23:57,163 home, just talking to his computer. 381 00:23:57,521 --> 00:24:03,071 Yeah, it's a great example and I'm glad you, you brought this 382 00:24:03,071 --> 00:24:08,141 up that you described, that with Cursor and a bit of vibe coding, 383 00:24:08,381 --> 00:24:09,941 everyone can do their own apps now. 384 00:24:09,941 --> 00:24:11,566 It's, yeah, super. 385 00:24:12,136 --> 00:24:15,816 And that kind of brings me to one of my two last questions. 386 00:24:16,086 --> 00:24:22,266 So, what book, tool or habit has made the most impact on you in the last 12 months? 387 00:24:22,566 --> 00:24:23,646 There's no question. 388 00:24:23,646 --> 00:24:28,936 I would never have thought that you could do this with words, and 389 00:24:28,936 --> 00:24:31,936 have an app in minutes running. 390 00:24:32,516 --> 00:24:35,266 I was blown away by what it could do. 391 00:24:35,266 --> 00:24:36,376 And this is still early. 392 00:24:36,426 --> 00:24:38,496 I cannot imagine what's coming next. 393 00:24:38,526 --> 00:24:42,756 So yes, Cursor is probably the most impactful. 394 00:24:43,476 --> 00:24:43,696 Wow. 395 00:24:44,186 --> 00:24:44,536 Super. 396 00:24:45,436 --> 00:24:50,806 And the last question that I always ask is, obviously we've been talking about all 397 00:24:51,016 --> 00:24:54,406 the things that are changing, but is there maybe one thing that you would expect 398 00:24:54,406 --> 00:24:56,356 to remain the same in the next 10 years? 399 00:24:57,356 --> 00:25:00,056 Nobody's replacing law attorneys. 400 00:25:00,836 --> 00:25:01,706 You hear this a lot. 401 00:25:01,746 --> 00:25:07,211 That is never gonna happen because a person has to be liable for the 402 00:25:07,211 --> 00:25:10,221 decisions and advice that is given. 403 00:25:10,221 --> 00:25:11,181 Legal advice. 404 00:25:11,871 --> 00:25:16,031 And there's an impact if you get it wrong, whereas if an AI gets it 405 00:25:16,031 --> 00:25:16,961 wrong, what are you going to do? 406 00:25:16,961 --> 00:25:17,621 Shut it down. 407 00:25:17,831 --> 00:25:18,881 You can't sue an AI. 408 00:25:19,161 --> 00:25:21,101 No, that is not gonna change. 409 00:25:21,176 --> 00:25:23,696 So accountability of professionals. 410 00:25:23,696 --> 00:25:27,926 Would you apply this specifically to legal or any professionals? 411 00:25:28,241 --> 00:25:31,271 Anyone who needs a license to operate. 412 00:25:31,901 --> 00:25:35,471 Those jobs are gonna stay because you cannot apply that 413 00:25:35,471 --> 00:25:38,291 to an artificial intelligence. 414 00:25:38,511 --> 00:25:41,511 There's no impact for them to be sued. 415 00:25:41,931 --> 00:25:43,611 Or maybe the company running them, yes. 416 00:25:43,611 --> 00:25:46,931 But still, it will always be attributed to a person. 417 00:25:48,101 --> 00:25:48,581 That's it. 418 00:25:48,731 --> 00:25:50,861 That's the only thing I think is gonna be the same. 419 00:25:51,551 --> 00:25:52,151 Fantastic. 420 00:25:52,721 --> 00:25:54,191 With that, Don, thank you very much. 421 00:25:54,551 --> 00:25:55,031 Thank you. 422 00:25:55,721 --> 00:25:56,361 I had a great time. 423 00:25:57,228 --> 00:26:01,788 Dr. Simmonds reminds us that enterprise transformation isn't reserved for coders. 424 00:26:02,028 --> 00:26:06,408 With only basic coding skills, he automated around 80% of repetitive 425 00:26:06,408 --> 00:26:10,008 trademark work, freeing, billable hours, and family time alike. 426 00:26:10,818 --> 00:26:15,438 Equally valuable is his roadmap for scaling: built hard guardrails 427 00:26:15,438 --> 00:26:19,338 against hallucinations, log every prompt for audit and treat prompt 428 00:26:19,338 --> 00:26:22,098 engineering as a core legal competency. 429 00:26:22,878 --> 00:26:24,858 Don is happy to swap war stories. 430 00:26:24,858 --> 00:26:28,548 Find him on LinkedIn for a free consultation on AI or IP law. 431 00:26:29,178 --> 00:26:30,678 The link is in the description. 432 00:26:31,728 --> 00:26:32,448 What about you? 433 00:26:33,018 --> 00:26:36,138 Are you ready for citizen developers to create revenue ready 434 00:26:36,228 --> 00:26:38,178 AI services in your organization? 435 00:26:39,258 --> 00:26:43,548 As always, we have more exciting topics and guest appearances lined up, so 436 00:26:43,548 --> 00:26:48,048 stay tuned for more tale of innovation that inspire, challenge and transform. 437 00:26:48,588 --> 00:26:50,418 Until next time, peace. 438 00:26:51,010 --> 00:26:53,350 Thanks for tuning in to Innovation Tales. 439 00:26:58,090 --> 00:27:01,780 Get inspired, connect with other practitioners and approach the 440 00:27:01,780 --> 00:27:03,820 digital revolution with confidence. 441 00:27:04,720 --> 00:27:08,170 Visit innovation-tales.com for more episodes. 442 00:27:09,480 --> 00:27:10,500 See you next time.