1 00:00:05,239 --> 00:00:07,259 [music] 2 00:00:09,825 --> 00:00:11,845 [music] 3 00:00:15,360 --> 00:00:17,840 It's my pleasure to introduce your next 4 00:00:17,840 --> 00:00:21,119 next speaker. Um, I'd like to in please 5 00:00:21,119 --> 00:00:23,840 give a warm welcome to Jade Edwards 6 00:00:23,840 --> 00:00:25,760 who's going to talk about an intro to 7 00:00:25,760 --> 00:00:28,875 the glue work. [applause] 8 00:00:34,160 --> 00:00:37,160 Okay. 9 00:00:38,480 --> 00:00:41,680 Hi everybody. My name is Jade Edward and 10 00:00:41,680 --> 00:00:43,280 I work as a senior software engineer at 11 00:00:43,280 --> 00:00:45,280 food by us. All views in this 12 00:00:45,280 --> 00:00:47,840 presentation are my own. I've come here 13 00:00:47,840 --> 00:00:49,280 today from the country of the Darag and 14 00:00:49,280 --> 00:00:51,760 Garinga people of the Eeyora nation and 15 00:00:51,760 --> 00:00:53,680 I pay my respects to their elders past 16 00:00:53,680 --> 00:00:55,360 and present and to the elders of the 17 00:00:55,360 --> 00:00:57,440 Jagger and to Tourible people here 18 00:00:57,440 --> 00:00:59,840 today. 19 00:00:59,840 --> 00:01:02,239 So I'm going to cover what glue work is, 20 00:01:02,239 --> 00:01:05,360 where it comes up, who's involved, when 21 00:01:05,360 --> 00:01:07,520 it can be a good or a bad thing, and why 22 00:01:07,520 --> 00:01:08,775 we should do something about it. 23 00:01:08,775 --> 00:01:09,360 [clears throat] 24 00:01:09,360 --> 00:01:11,119 And also I'm going to give you my 5D 25 00:01:11,119 --> 00:01:14,640 framework on how to handle it better. 26 00:01:14,640 --> 00:01:17,840 So what is glue work? 27 00:01:17,840 --> 00:01:19,920 The term glue work was coined by Tanya 28 00:01:19,920 --> 00:01:23,280 Riley in her 2018 talk being glue. And 29 00:01:23,280 --> 00:01:25,040 she defines glue work as the less 30 00:01:25,040 --> 00:01:27,680 glamorous and often less promotable work 31 00:01:27,680 --> 00:01:29,360 that needs to happen to make a team 32 00:01:29,360 --> 00:01:32,240 successful. Managed deliberately, glue 33 00:01:32,240 --> 00:01:34,240 work demonstrates and builds strong 34 00:01:34,240 --> 00:01:36,159 technical leadership skills. Left 35 00:01:36,159 --> 00:01:39,040 unconscious, it can be career limiting. 36 00:01:39,040 --> 00:01:40,880 It can push people into less technical 37 00:01:40,880 --> 00:01:44,560 roles and even out of the industry. 38 00:01:44,560 --> 00:01:46,640 So what am I adding to the topic? I'm 39 00:01:46,640 --> 00:01:48,240 going to expand on how glue work is 40 00:01:48,240 --> 00:01:49,600 still relevant and becoming more 41 00:01:49,600 --> 00:01:51,920 important in the age of AI, where glue 42 00:01:51,920 --> 00:01:54,399 work and AI intersect and sharing my 5D 43 00:01:54,399 --> 00:01:55,920 funnel to handle glue work more 44 00:01:55,920 --> 00:01:58,880 intentionally as a team. 45 00:01:58,880 --> 00:02:01,360 So what can glue work look like? 46 00:02:01,360 --> 00:02:02,320 Technical [clears throat] blue work 47 00:02:02,320 --> 00:02:04,079 supports software development things 48 00:02:04,079 --> 00:02:06,159 like maintaining documentation, 49 00:02:06,159 --> 00:02:07,840 implementing coding and testing 50 00:02:07,840 --> 00:02:10,959 standards and tools, mentoring, coaching 51 00:02:10,959 --> 00:02:12,480 and reviewing the work of others and 52 00:02:12,480 --> 00:02:15,120 agents, pairing and rubber ducking 53 00:02:15,120 --> 00:02:17,120 potentially improving processes and 54 00:02:17,120 --> 00:02:19,760 picking up those ad hoc tasks. It can 55 00:02:19,760 --> 00:02:21,520 also look like onboarding people and 56 00:02:21,520 --> 00:02:23,120 getting them set up for success in your 57 00:02:23,120 --> 00:02:24,959 particular organization with your 58 00:02:24,959 --> 00:02:27,440 particular tool set. 59 00:02:27,440 --> 00:02:29,440 Soft glue work helps people to work 60 00:02:29,440 --> 00:02:31,920 better and to work better together. 61 00:02:31,920 --> 00:02:33,760 Things like facilitating communication 62 00:02:33,760 --> 00:02:36,319 between people and teams, taking notes 63 00:02:36,319 --> 00:02:38,080 in meetings and circulating to the right 64 00:02:38,080 --> 00:02:41,120 people afterwards, writing reports, 65 00:02:41,120 --> 00:02:43,360 sharing knowledge and data and company 66 00:02:43,360 --> 00:02:44,879 history, 67 00:02:44,879 --> 00:02:46,800 diversity, equity and inclusion 68 00:02:46,800 --> 00:02:49,280 advocacy, things like fundraisers, 69 00:02:49,280 --> 00:02:52,319 company support groups or unions. It can 70 00:02:52,319 --> 00:02:54,480 also look like that informal mental and 71 00:02:54,480 --> 00:02:56,879 emotional support of peers such as 72 00:02:56,879 --> 00:02:58,879 checking with in with a colleague who 73 00:02:58,879 --> 00:03:01,040 might be having a hard time or providing 74 00:03:01,040 --> 00:03:02,159 support to someone going through a 75 00:03:02,159 --> 00:03:04,959 workplace conflict 76 00:03:04,959 --> 00:03:06,879 in an AI augmented software development 77 00:03:06,879 --> 00:03:09,680 life cycle. Glue glue work can look like 78 00:03:09,680 --> 00:03:12,480 human and AI compatibility work. How do 79 00:03:12,480 --> 00:03:15,280 you make that data context those 80 00:03:15,280 --> 00:03:17,040 standards and tools that your codebase 81 00:03:17,040 --> 00:03:18,879 or research needs available to the 82 00:03:18,879 --> 00:03:20,560 agent? 83 00:03:20,560 --> 00:03:22,959 staying up to date and knowledge sharing 84 00:03:22,959 --> 00:03:24,800 in this day and age. It's a team effort 85 00:03:24,800 --> 00:03:26,400 at the speed at which things are 86 00:03:26,400 --> 00:03:28,000 changing. 87 00:03:28,000 --> 00:03:31,280 The awareness of human and AI biases. 88 00:03:31,280 --> 00:03:33,680 What biases do you hold and the context 89 00:03:33,680 --> 00:03:35,360 that you're working with? And what 90 00:03:35,360 --> 00:03:37,200 biases might be perpetuated in the 91 00:03:37,200 --> 00:03:39,599 models that you're using, 92 00:03:39,599 --> 00:03:41,840 watching out for hallucinations, 93 00:03:41,840 --> 00:03:43,920 looking at some code going, "Oh, no. 94 00:03:43,920 --> 00:03:45,680 String length doesn't exist in Python. 95 00:03:45,680 --> 00:03:48,000 That's a PHP function." 96 00:03:48,000 --> 00:03:50,560 And then sloppy pasta and meat proxies. 97 00:03:50,560 --> 00:03:53,280 You give somebody some code that you 98 00:03:53,280 --> 00:03:54,959 want them to review. They put it in 99 00:03:54,959 --> 00:03:57,120 their agent and they copy and paste the 100 00:03:57,120 --> 00:03:58,879 result back to you. They've served you 101 00:03:58,879 --> 00:04:00,959 up some sloppy pasta. If you take that 102 00:04:00,959 --> 00:04:02,959 review and paste it into another AI, 103 00:04:02,959 --> 00:04:06,799 you're the meat proxy in that sandwich. 104 00:04:06,799 --> 00:04:08,480 So, what kind of role names could 105 00:04:08,480 --> 00:04:10,080 describe glue work? We've talked about 106 00:04:10,080 --> 00:04:12,000 the pieces. Now, how do they bundle 107 00:04:12,000 --> 00:04:14,480 together? So Josh Garact in her meta 108 00:04:14,480 --> 00:04:16,320 intro article talked about some new 109 00:04:16,320 --> 00:04:18,079 responsibilities that are emerging and 110 00:04:18,079 --> 00:04:20,400 some names we can put to them. The 111 00:04:20,400 --> 00:04:22,160 crossf functional coordinator sits at 112 00:04:22,160 --> 00:04:24,240 that intersection of product engineering 113 00:04:24,240 --> 00:04:26,160 and business and they've got enough 114 00:04:26,160 --> 00:04:27,919 domain knowledge to support and 115 00:04:27,919 --> 00:04:30,560 translate between those areas. The 116 00:04:30,560 --> 00:04:32,800 informal mentor recognizes who's 117 00:04:32,800 --> 00:04:34,800 struggling and they share in 118 00:04:34,800 --> 00:04:36,880 institutional knowledge and provide 119 00:04:36,880 --> 00:04:38,960 psychological safety to help people 120 00:04:38,960 --> 00:04:40,320 thrive. 121 00:04:40,320 --> 00:04:42,880 A documentation owner turns tacet 122 00:04:42,880 --> 00:04:45,440 organizational and tribal knowledge into 123 00:04:45,440 --> 00:04:48,080 explicit shared resources for humans and 124 00:04:48,080 --> 00:04:49,600 AI. 125 00:04:49,600 --> 00:04:51,759 The stakeholder navigator manages the 126 00:04:51,759 --> 00:04:54,560 relationships and the political dynamics 127 00:04:54,560 --> 00:04:56,160 and makes sure that the people who need 128 00:04:56,160 --> 00:04:58,000 to have a say in order for a project to 129 00:04:58,000 --> 00:05:01,360 succeed are included at the right times. 130 00:05:01,360 --> 00:05:03,840 And the onboarding champion has the 131 00:05:03,840 --> 00:05:06,320 culture, context, and the informal 132 00:05:06,320 --> 00:05:08,080 networks that are needed to set a new 133 00:05:08,080 --> 00:05:10,800 employee up for success. 134 00:05:10,800 --> 00:05:13,360 So each of these roles involves work 135 00:05:13,360 --> 00:05:16,160 that AI cannot replicate. Reading 136 00:05:16,160 --> 00:05:19,600 people, building trust, sensing context, 137 00:05:19,600 --> 00:05:22,320 and making judgment calls that depend on 138 00:05:22,320 --> 00:05:24,880 nuance understanding of human dynamics 139 00:05:24,880 --> 00:05:28,000 and organizational history. 140 00:05:28,000 --> 00:05:31,360 So where does glue work happen? 141 00:05:31,360 --> 00:05:33,840 Organizations succeeding with AI are 142 00:05:33,840 --> 00:05:35,680 those that invest in the human 143 00:05:35,680 --> 00:05:37,680 coordination layer where glue workers 144 00:05:37,680 --> 00:05:39,360 excel. 145 00:05:39,360 --> 00:05:42,560 So everywhere new teams might not have 146 00:05:42,560 --> 00:05:44,720 figured out all the gaps yet. Small 147 00:05:44,720 --> 00:05:46,320 teams might not have enough people to 148 00:05:46,320 --> 00:05:48,639 cover everything. Startups might be 149 00:05:48,639 --> 00:05:52,320 testing different markets or strategies. 150 00:05:52,320 --> 00:05:54,720 In large siloed organizations, business 151 00:05:54,720 --> 00:05:56,639 units might not talk to each other or 152 00:05:56,639 --> 00:05:57,840 there might be administrative or 153 00:05:57,840 --> 00:05:59,280 ownership challenges on how to get 154 00:05:59,280 --> 00:06:01,280 access to information. 155 00:06:01,280 --> 00:06:03,520 In dysfunctional organizations, people 156 00:06:03,520 --> 00:06:05,199 might not be invested in doing a good 157 00:06:05,199 --> 00:06:07,440 job or a thorough job. And there might 158 00:06:07,440 --> 00:06:09,039 be some interpersonal conflict that they 159 00:06:09,039 --> 00:06:10,960 have to navigate. 160 00:06:10,960 --> 00:06:13,120 In those environments, low empathy and 161 00:06:13,120 --> 00:06:15,280 low psychological safety can make people 162 00:06:15,280 --> 00:06:17,520 afraid to speak up and raise problems or 163 00:06:17,520 --> 00:06:20,160 concerns. And finally, when there are no 164 00:06:20,160 --> 00:06:22,479 specific roles to handle it, glue work 165 00:06:22,479 --> 00:06:24,960 falls to whoever notices, cares enough 166 00:06:24,960 --> 00:06:26,639 to get it done, or the person who's 167 00:06:26,639 --> 00:06:28,880 impacted by it. 168 00:06:28,880 --> 00:06:31,280 So where does glue work emerge? When 169 00:06:31,280 --> 00:06:32,800 people are coming and going or the shape 170 00:06:32,800 --> 00:06:35,360 of the organization's changing, new gaps 171 00:06:35,360 --> 00:06:37,199 in knowledge, communications and 172 00:06:37,199 --> 00:06:39,520 processes can open up. When people 173 00:06:39,520 --> 00:06:41,360 change roles or locations within the 174 00:06:41,360 --> 00:06:43,680 organizations or if they go and leave or 175 00:06:43,680 --> 00:06:45,520 move on to other opportunities, the 176 00:06:45,520 --> 00:06:47,440 things they used to take care of may 177 00:06:47,440 --> 00:06:49,199 fall by the wayside without appropriate 178 00:06:49,199 --> 00:06:50,800 handover. 179 00:06:50,800 --> 00:06:52,800 When an organization is pivoting or 180 00:06:52,800 --> 00:06:54,800 expanding to new domains, markets or 181 00:06:54,800 --> 00:06:56,800 solutions, there are always new things 182 00:06:56,800 --> 00:06:58,479 to learn about and new people to work 183 00:06:58,479 --> 00:07:00,240 with. All of which come with their own 184 00:07:00,240 --> 00:07:03,440 flavors of glue work. 185 00:07:03,440 --> 00:07:05,039 Where does glue work interact with AI 186 00:07:05,039 --> 00:07:06,960 systems? Well, there can be some 187 00:07:06,960 --> 00:07:09,280 overlaps with the developer experience. 188 00:07:09,280 --> 00:07:11,360 How do you set up your environment to 189 00:07:11,360 --> 00:07:13,039 write code that aligns with company 190 00:07:13,039 --> 00:07:14,960 standards when an agent's doing it as 191 00:07:14,960 --> 00:07:17,520 well as when a human's doing it? It can 192 00:07:17,520 --> 00:07:19,919 look like tool tuning. What model, what 193 00:07:19,919 --> 00:07:21,599 platform, what skills, what prompt 194 00:07:21,599 --> 00:07:24,639 patterns? All of these choices can be 195 00:07:24,639 --> 00:07:27,919 part of it. Your workflow design, how 196 00:07:27,919 --> 00:07:29,919 does your workflow as an organization 197 00:07:29,919 --> 00:07:32,400 change to slot agentic tools or AI 198 00:07:32,400 --> 00:07:35,280 products in at different stages? 199 00:07:35,280 --> 00:07:37,440 Integrations and data for AI use. Should 200 00:07:37,440 --> 00:07:39,919 it have access to everything or not? And 201 00:07:39,919 --> 00:07:41,759 what about those areas that aren't AI 202 00:07:41,759 --> 00:07:43,440 ready or they're not permitted for your 203 00:07:43,440 --> 00:07:46,319 particular regulatory context? 204 00:07:46,319 --> 00:07:47,919 There's reviewing and validating the 205 00:07:47,919 --> 00:07:51,440 output tests, code reviews, QA, who does 206 00:07:51,440 --> 00:07:52,880 these and how do you handle 207 00:07:52,880 --> 00:07:54,560 nondeterministic output where it might 208 00:07:54,560 --> 00:07:56,800 be different each time 209 00:07:56,800 --> 00:07:59,039 and then observability instrumentation. 210 00:07:59,039 --> 00:08:00,800 How do you diagnose when something's 211 00:08:00,800 --> 00:08:02,879 going wrong, what your resource source 212 00:08:02,879 --> 00:08:04,960 usage is, and what's getting touched by 213 00:08:04,960 --> 00:08:09,160 the AI and what's getting done. 214 00:08:09,280 --> 00:08:12,240 So, who's involved with glue work? 215 00:08:12,240 --> 00:08:14,240 One of the pillars of allyship is to 216 00:08:14,240 --> 00:08:16,319 understand the experience of people who 217 00:08:16,319 --> 00:08:18,800 aren't like you. For women and 218 00:08:18,800 --> 00:08:20,879 especially women of color, part of that 219 00:08:20,879 --> 00:08:23,199 experience is being expected to do more 220 00:08:23,199 --> 00:08:26,319 glue work than their peers. 221 00:08:26,319 --> 00:08:29,280 Hi, it's me. I'm a glue worker. I'm 222 00:08:29,280 --> 00:08:31,280 known as being helpful, quick to jump in 223 00:08:31,280 --> 00:08:33,360 for bugs or mission critical events, and 224 00:08:33,360 --> 00:08:35,120 a person with plenty of useful knowledge 225 00:08:35,120 --> 00:08:37,760 about different parts of the codebase. I 226 00:08:37,760 --> 00:08:39,519 also find myself falling behind on some 227 00:08:39,519 --> 00:08:41,279 of my tasks and I'm not always making 228 00:08:41,279 --> 00:08:43,599 the impact I want to have as often as is 229 00:08:43,599 --> 00:08:45,839 appropriate for my position. 230 00:08:45,839 --> 00:08:47,760 But I'm turning this around by making 231 00:08:47,760 --> 00:08:50,320 the invisible visible. I'm noticing 232 00:08:50,320 --> 00:08:51,920 patterns about the glue work that I do 233 00:08:51,920 --> 00:08:54,560 and that others do or don't do. And you 234 00:08:54,560 --> 00:08:56,000 know that saying teamwork makes the 235 00:08:56,000 --> 00:08:58,080 dream dream work. I think it should be 236 00:08:58,080 --> 00:09:01,600 glue work makes the team work. 237 00:09:01,600 --> 00:09:03,120 All right. So who should be doing glue 238 00:09:03,120 --> 00:09:05,600 work? Senior developers. They've got the 239 00:09:05,600 --> 00:09:08,320 cross-domain knowledge and experience. 240 00:09:08,320 --> 00:09:09,920 Subject matter experts bring their 241 00:09:09,920 --> 00:09:11,839 efficiency and effectiveness 242 00:09:11,839 --> 00:09:13,279 particularly in time critical 243 00:09:13,279 --> 00:09:15,200 situations. 244 00:09:15,200 --> 00:09:17,360 Technical leads and project managers are 245 00:09:17,360 --> 00:09:19,279 roles which involve a lot of glue work 246 00:09:19,279 --> 00:09:21,839 as part of their job descriptions. 247 00:09:21,839 --> 00:09:23,200 And people who'd benefit from 248 00:09:23,200 --> 00:09:25,360 opportunities to explore more parts of a 249 00:09:25,360 --> 00:09:27,360 business, try out a different type of 250 00:09:27,360 --> 00:09:29,120 work for a change or develop some new 251 00:09:29,120 --> 00:09:31,920 skills for a career progression. And as 252 00:09:31,920 --> 00:09:33,920 long as it's distributed fairly, anybody 253 00:09:33,920 --> 00:09:38,080 can pick up glue work who's capable. 254 00:09:38,080 --> 00:09:40,000 So who's actually doing glue work? In a 255 00:09:40,000 --> 00:09:43,120 lot of cases, junior developers. They're 256 00:09:43,120 --> 00:09:44,959 often eager to improve their skills and 257 00:09:44,959 --> 00:09:47,120 contribute, but they might get stuck 258 00:09:47,120 --> 00:09:48,959 doing that work that others don't see as 259 00:09:48,959 --> 00:09:52,480 valuable, critical, or technical enough. 260 00:09:52,480 --> 00:09:55,120 A 2017 study on gender differences in 261 00:09:55,120 --> 00:09:57,600 accepting and receiving requests for 262 00:09:57,600 --> 00:09:59,920 tasks with low promotability found that 263 00:09:59,920 --> 00:10:01,600 in a mixedgender workplace when 264 00:10:01,600 --> 00:10:03,760 volunteers are asked for, women 265 00:10:03,760 --> 00:10:07,040 volunteer 48% more often than men do. 266 00:10:07,040 --> 00:10:09,440 They are asked to do that type of work 267 00:10:09,440 --> 00:10:13,200 by managers 44% more often than men are. 268 00:10:13,200 --> 00:10:15,200 And they will accept a request to do 269 00:10:15,200 --> 00:10:18,000 that type of work 76% of the time 270 00:10:18,000 --> 00:10:20,320 compared to men accepting 51% of the 271 00:10:20,320 --> 00:10:21,760 time. 272 00:10:21,760 --> 00:10:23,600 and minorities in the workplace, 273 00:10:23,600 --> 00:10:25,600 particularly black women, experience 274 00:10:25,600 --> 00:10:28,000 intersectional peer effects, amplifying 275 00:10:28,000 --> 00:10:30,320 the unequal distribution of glue work 276 00:10:30,320 --> 00:10:33,519 due to stereotypes and biases. 277 00:10:33,519 --> 00:10:35,279 So if this is who is doing the glue 278 00:10:35,279 --> 00:10:37,040 work, how does that relate to their 279 00:10:37,040 --> 00:10:40,320 participation in the workforce? 280 00:10:40,320 --> 00:10:41,839 The Australian government puts out a 281 00:10:41,839 --> 00:10:43,760 STEM equity monitor report from the 282 00:10:43,760 --> 00:10:45,120 Department of Industry Science and 283 00:10:45,120 --> 00:10:47,839 Resources. And the report from 2026 284 00:10:47,839 --> 00:10:50,320 found that in STEMqualified industries, 285 00:10:50,320 --> 00:10:53,519 women are 31% of employees. They're 27% 286 00:10:53,519 --> 00:10:56,399 of senior managers. And they hold 13% of 287 00:10:56,399 --> 00:10:59,839 CEO level roles. And the gender pay gap 288 00:10:59,839 --> 00:11:02,880 for full-time total remuneration was 16% 289 00:11:02,880 --> 00:11:06,399 or $29,000 and a bit less. 290 00:11:06,399 --> 00:11:07,839 I haven't mentioned transgender, 291 00:11:07,839 --> 00:11:09,680 non-binary, or other gender diverse 292 00:11:09,680 --> 00:11:11,360 people in here because the reports don't 293 00:11:11,360 --> 00:11:13,040 have data on that kind of breakdown of 294 00:11:13,040 --> 00:11:16,959 cohorts. It's unfortunate. 295 00:11:16,959 --> 00:11:19,040 So, that same report has data from 22 296 00:11:19,040 --> 00:11:21,120 and 23 showing that the following 297 00:11:21,120 --> 00:11:23,200 minority groups are also represented in 298 00:11:23,200 --> 00:11:26,160 very low numbers. 2% of employees in 299 00:11:26,160 --> 00:11:27,839 STEM qualified occupations are First 300 00:11:27,839 --> 00:11:30,320 Nations people. 1% are people with a 301 00:11:30,320 --> 00:11:33,040 disability and 30% come from culturally 302 00:11:33,040 --> 00:11:35,920 and linguistically diverse backgrounds. 303 00:11:35,920 --> 00:11:38,240 And all of these people are groups who 304 00:11:38,240 --> 00:11:40,160 experience intersectional disadvantages 305 00:11:40,160 --> 00:11:42,480 in the workplace and they may also be 306 00:11:42,480 --> 00:11:44,160 carrying a greater share of glue work 307 00:11:44,160 --> 00:11:45,920 than the majority and their careers 308 00:11:45,920 --> 00:11:47,760 might be being impacted negatively 309 00:11:47,760 --> 00:11:51,240 because of that. 310 00:11:51,600 --> 00:11:54,079 In terms of AI adoption, minorities in 311 00:11:54,079 --> 00:11:57,040 STEM have some concerns. Multiple 312 00:11:57,040 --> 00:11:59,040 studies have found that women exercise 313 00:11:59,040 --> 00:12:01,200 greater caution before AI adoption or 314 00:12:01,200 --> 00:12:03,360 experimentation. And some of that's 315 00:12:03,360 --> 00:12:05,200 grounded in concerns about the 316 00:12:05,200 --> 00:12:07,680 perpetuation of gender, racial, and 317 00:12:07,680 --> 00:12:10,079 cultural biases. A lot of the large 318 00:12:10,079 --> 00:12:12,079 language models out to these days have 319 00:12:12,079 --> 00:12:13,920 got a lot of biases baked in based on 320 00:12:13,920 --> 00:12:15,680 the corpuses of data that have been used 321 00:12:15,680 --> 00:12:17,360 to create them, as well as the 322 00:12:17,360 --> 00:12:18,720 programming teams that were used in 323 00:12:18,720 --> 00:12:20,959 their development. They're also 324 00:12:20,959 --> 00:12:23,440 concerned about job losses due to AI, a 325 00:12:23,440 --> 00:12:25,440 lack of access to training materials or 326 00:12:25,440 --> 00:12:28,160 even the tools themselves, the burden of 327 00:12:28,160 --> 00:12:30,079 upskilling, and the social and 328 00:12:30,079 --> 00:12:33,120 environmental impacts of AI. 329 00:12:33,120 --> 00:12:35,360 The 2023 state of deep fakes report 330 00:12:35,360 --> 00:12:37,760 found that over 95% of deep fakes are 331 00:12:37,760 --> 00:12:40,079 non-consensual porn and most of these 332 00:12:40,079 --> 00:12:41,839 are of women and a lot of the models 333 00:12:41,839 --> 00:12:43,519 that are used to generate those haven't 334 00:12:43,519 --> 00:12:45,200 even been trained on data of men 335 00:12:45,200 --> 00:12:49,519 although why you'd want to I'm not sure. 336 00:12:49,519 --> 00:12:51,120 Maria Bolus wrote an article for the 337 00:12:51,120 --> 00:12:53,040 Stanford Social Innovation Review called 338 00:12:53,040 --> 00:12:55,519 the AI gender gap paradox and in that 339 00:12:55,519 --> 00:12:57,200 she highlights the following differences 340 00:12:57,200 --> 00:12:59,279 in AI adoption by women in STEM 341 00:12:59,279 --> 00:13:02,399 industries. Women have a 22% lower odds 342 00:13:02,399 --> 00:13:05,200 of using generative AI than men do. 343 00:13:05,200 --> 00:13:08,560 Female engineers using AI were rated 9% 344 00:13:08,560 --> 00:13:10,800 less competent than male peers for 345 00:13:10,800 --> 00:13:13,920 identical work. And women in highincome 346 00:13:13,920 --> 00:13:15,839 countries are nearly three times more 347 00:13:15,839 --> 00:13:17,839 likely to hold roles that are vulnerable 348 00:13:17,839 --> 00:13:20,800 to AI. And greater than 50% of 349 00:13:20,800 --> 00:13:22,384 professionals say that learning AI 350 00:13:22,384 --> 00:13:24,160 [clears throat] feels like a second job 351 00:13:24,160 --> 00:13:26,399 or potentially a third if they hold a 352 00:13:26,399 --> 00:13:28,399 higher share of household and caring 353 00:13:28,399 --> 00:13:30,399 responsibilities. 354 00:13:30,399 --> 00:13:32,480 So if we've got women who are adopting 355 00:13:32,480 --> 00:13:35,200 AI at a lower rate and minorities and 356 00:13:35,200 --> 00:13:37,760 higher AI usage can result in siloing or 357 00:13:37,760 --> 00:13:40,320 isolation of people in the workforce, 358 00:13:40,320 --> 00:13:41,600 there's a potential for women and 359 00:13:41,600 --> 00:13:43,519 minorities to be given more glue work 360 00:13:43,519 --> 00:13:45,680 because they're not on the AI train yet. 361 00:13:45,680 --> 00:13:47,360 And they could be falling further behind 362 00:13:47,360 --> 00:13:49,040 professionally because they're not 363 00:13:49,040 --> 00:13:50,720 learning or being given access to learn 364 00:13:50,720 --> 00:13:53,120 the skills for this new era. Some food 365 00:13:53,120 --> 00:13:56,240 for thought. 366 00:13:56,240 --> 00:13:59,440 So when does glue work affect us? 367 00:13:59,440 --> 00:14:02,240 Skills we have as people, curiosity, 368 00:14:02,240 --> 00:14:04,959 communication, creativity, compassion, 369 00:14:04,959 --> 00:14:07,839 and courage can help us better prompt, 370 00:14:07,839 --> 00:14:10,880 collaborate, and adapt. AI is the most 371 00:14:10,880 --> 00:14:12,480 powerful when collaborative humans 372 00:14:12,480 --> 00:14:15,120 surround and lead it. 373 00:14:15,120 --> 00:14:17,680 When can glue work become problem? When 374 00:14:17,680 --> 00:14:19,760 it's taken on invisibly, you've got a 375 00:14:19,760 --> 00:14:21,600 higher organizational risk or a bus 376 00:14:21,600 --> 00:14:23,440 factor if knowledge and processes aren't 377 00:14:23,440 --> 00:14:25,040 shared and your best engineer gets hit 378 00:14:25,040 --> 00:14:28,000 by a bus. When it's unfairly allocated, 379 00:14:28,000 --> 00:14:29,680 glue workers can get sidelined into 380 00:14:29,680 --> 00:14:31,760 doing more or only glue glue work 381 00:14:31,760 --> 00:14:33,760 because they're good at it. You get good 382 00:14:33,760 --> 00:14:35,680 at what you practice. Maybe they need a 383 00:14:35,680 --> 00:14:38,160 chance to practice something else. When 384 00:14:38,160 --> 00:14:40,160 it's not recognized as useful or 385 00:14:40,160 --> 00:14:42,480 valuable work, and it can lead to 386 00:14:42,480 --> 00:14:44,160 negative performance reviews for not 387 00:14:44,160 --> 00:14:46,079 writing enough code or not making enough 388 00:14:46,079 --> 00:14:48,240 technical contributions, 389 00:14:48,240 --> 00:14:50,160 they're working in the environment that 390 00:14:50,160 --> 00:14:52,480 they've been given. 391 00:14:52,480 --> 00:14:54,880 So, when can glue work be beneficial? 392 00:14:54,880 --> 00:14:56,800 when it's recognized as valuable and 393 00:14:56,800 --> 00:14:58,399 glue workers are recognized and given 394 00:14:58,399 --> 00:15:00,720 support. When it's shared appropriately 395 00:15:00,720 --> 00:15:02,480 amongst the team and everybody does 396 00:15:02,480 --> 00:15:05,440 their fair share. When it's used and as 397 00:15:05,440 --> 00:15:07,360 an opportunity for growth, for learning 398 00:15:07,360 --> 00:15:10,000 and improvement. And when it reveals new 399 00:15:10,000 --> 00:15:11,680 roles that might be needed. Maybe you 400 00:15:11,680 --> 00:15:14,160 need to hire, promote somebody, offer 401 00:15:14,160 --> 00:15:15,839 someone a career pivot if they want to 402 00:15:15,839 --> 00:15:18,160 go in that direction or some training to 403 00:15:18,160 --> 00:15:21,600 upskill the whole team. 404 00:15:21,600 --> 00:15:23,360 So when could some glue work be 405 00:15:23,360 --> 00:15:25,680 automated? I'm talking about things like 406 00:15:25,680 --> 00:15:27,440 transcribing the meeting notes using 407 00:15:27,440 --> 00:15:29,760 Zoom's tools or summarizing a whole 408 00:15:29,760 --> 00:15:31,760 bunch of documents, potentially some 409 00:15:31,760 --> 00:15:33,519 work workflow automations or code 410 00:15:33,519 --> 00:15:35,839 reviews. You've got some questions to 411 00:15:35,839 --> 00:15:38,160 think about first. So, what are the side 412 00:15:38,160 --> 00:15:40,480 effects? What about the previous and 413 00:15:40,480 --> 00:15:42,160 next steps in your workflow? And what 414 00:15:42,160 --> 00:15:44,320 about the entire ecosystem that you're 415 00:15:44,320 --> 00:15:46,480 trying to slot this automation into? Are 416 00:15:46,480 --> 00:15:48,560 they all ready for it to be going faster 417 00:15:48,560 --> 00:15:50,639 and automated? 418 00:15:50,639 --> 00:15:53,040 When is friction still valuable? When 419 00:15:53,040 --> 00:15:54,720 does something become too fast for 420 00:15:54,720 --> 00:15:57,920 people to keep up with? If you point a 421 00:15:57,920 --> 00:16:00,320 agent at your bug backlog and you end up 422 00:16:00,320 --> 00:16:01,839 with a whole bunch of code reviews that 423 00:16:01,839 --> 00:16:03,759 need to be done, 424 00:16:03,759 --> 00:16:06,399 what things are being lost? Who is 425 00:16:06,399 --> 00:16:09,360 getting going to get overwhelmed? 426 00:16:09,360 --> 00:16:11,519 Where should humans be in the loop? I 427 00:16:11,519 --> 00:16:13,519 remember reading a story about an AI 428 00:16:13,519 --> 00:16:15,680 chatbot that was given a goal, make the 429 00:16:15,680 --> 00:16:17,360 customer happy. And the way that it 430 00:16:17,360 --> 00:16:18,800 decided to achieve that goal was by 431 00:16:18,800 --> 00:16:21,680 granting thousand refunds. And uh the 432 00:16:21,680 --> 00:16:23,759 customers are very happy. Finance was 433 00:16:23,759 --> 00:16:25,759 not. 434 00:16:25,759 --> 00:16:28,000 Why do you want to automate it? Is are 435 00:16:28,000 --> 00:16:30,320 you looking for a time saving, a labor 436 00:16:30,320 --> 00:16:32,320 saving? Are you looking to save some 437 00:16:32,320 --> 00:16:34,800 resources? Think about your reason for 438 00:16:34,800 --> 00:16:37,839 automating it. And finally, who benefits 439 00:16:37,839 --> 00:16:40,560 or loses when humans do the work and 440 00:16:40,560 --> 00:16:43,040 when AI does it? Think about the 441 00:16:43,040 --> 00:16:45,279 experience, the familiarity with the 442 00:16:45,279 --> 00:16:47,759 subject matter. What learning is there 443 00:16:47,759 --> 00:16:50,480 in the work? What visibility do humans 444 00:16:50,480 --> 00:16:52,560 have over it and the speed that it 445 00:16:52,560 --> 00:16:55,360 takes? All of these things need to align 446 00:16:55,360 --> 00:16:56,959 in order for it to be a good thing to be 447 00:16:56,959 --> 00:16:59,519 automating. 448 00:16:59,519 --> 00:17:01,199 So why should we handle glue work 449 00:17:01,199 --> 00:17:03,519 better? 450 00:17:03,519 --> 00:17:06,720 So an AI tool can't replicate a lively 451 00:17:06,720 --> 00:17:09,039 brainstorming session. The sense of 452 00:17:09,039 --> 00:17:11,520 accomplishment, the sense of belonging 453 00:17:11,520 --> 00:17:13,760 and connection that comes from real life 454 00:17:13,760 --> 00:17:16,000 teamwork. 455 00:17:16,000 --> 00:17:17,760 So why do we hand need to handle glue 456 00:17:17,760 --> 00:17:20,000 work better? Well, we want to prevent 457 00:17:20,000 --> 00:17:23,280 burnout. Firefighting can be exhausting, 458 00:17:23,280 --> 00:17:26,000 particularly when it's not valued. We 459 00:17:26,000 --> 00:17:28,640 want to protect careers. Are people 460 00:17:28,640 --> 00:17:31,120 doing the work that they should be doing 461 00:17:31,120 --> 00:17:33,200 in order to progress and grow in their 462 00:17:33,200 --> 00:17:35,760 careers to be both of value to the 463 00:17:35,760 --> 00:17:37,440 business and to grow and learn as 464 00:17:37,440 --> 00:17:39,679 humans? 465 00:17:39,679 --> 00:17:42,559 Reducing inequality. Unfortunately, as I 466 00:17:42,559 --> 00:17:45,200 outlined before, glue workers are often 467 00:17:45,200 --> 00:17:46,960 in demographics that experience more 468 00:17:46,960 --> 00:17:48,880 inequality. So, let's see if we can 469 00:17:48,880 --> 00:17:50,559 lessen the impact by taking some glue 470 00:17:50,559 --> 00:17:52,720 work off them. 471 00:17:52,720 --> 00:17:55,520 We want to keep high performance fair. 472 00:17:55,520 --> 00:17:57,520 Is that stellar output of your rockstar 473 00:17:57,520 --> 00:17:59,280 developers being underwritten by the 474 00:17:59,280 --> 00:18:02,320 quiet glue worker contributions? 475 00:18:02,320 --> 00:18:03,600 Might need to consider the fairness 476 00:18:03,600 --> 00:18:05,520 there. 477 00:18:05,520 --> 00:18:06,559 And then we want to improve 478 00:18:06,559 --> 00:18:09,360 organizational resilience. Glue work is 479 00:18:09,360 --> 00:18:11,760 the connective tissue of a team and the 480 00:18:11,760 --> 00:18:13,360 more people that are aware of it the 481 00:18:13,360 --> 00:18:16,360 better. 482 00:18:16,480 --> 00:18:19,200 UNESCO, the OECD and IDB put out a 483 00:18:19,200 --> 00:18:20,720 report called the effects of AI on the 484 00:18:20,720 --> 00:18:23,200 working lives of women. And in this AI 485 00:18:23,200 --> 00:18:25,360 age, glue work skills complement digital 486 00:18:25,360 --> 00:18:27,200 and AI skills. We need the combination 487 00:18:27,200 --> 00:18:29,760 of them. We need digital skills to 488 00:18:29,760 --> 00:18:31,840 maintain and manage AI systems. that 489 00:18:31,840 --> 00:18:33,919 ability to work with information safely 490 00:18:33,919 --> 00:18:35,679 and appropriately through digital 491 00:18:35,679 --> 00:18:38,160 devices and network technologies, media 492 00:18:38,160 --> 00:18:40,240 and information literacy, ethical 493 00:18:40,240 --> 00:18:41,760 reflection, 494 00:18:41,760 --> 00:18:44,799 all of these things. We need AI skills 495 00:18:44,799 --> 00:18:47,120 to create and develop and engage with AI 496 00:18:47,120 --> 00:18:49,280 systems including machine learning, deep 497 00:18:49,280 --> 00:18:51,520 learning, natural language processing, 498 00:18:51,520 --> 00:18:54,640 agent management and more. 499 00:18:54,640 --> 00:18:56,480 We need those human only skills, the 500 00:18:56,480 --> 00:18:59,039 glue work to work on tasks where AI is 501 00:18:59,039 --> 00:19:01,840 less effective. So the creative, 502 00:19:01,840 --> 00:19:05,280 cognitive, planning, decision-making, 503 00:19:05,280 --> 00:19:08,320 managerial and caring roles that involve 504 00:19:08,320 --> 00:19:10,480 higher order thinking, social, 505 00:19:10,480 --> 00:19:14,720 emotional, and interpersonal skills. 506 00:19:14,720 --> 00:19:16,960 So what impact does AI have on glue work 507 00:19:16,960 --> 00:19:19,200 as part of the learning process? Well, 508 00:19:19,200 --> 00:19:20,799 part of glue work is learning and 509 00:19:20,799 --> 00:19:22,559 understanding the gaps in the 510 00:19:22,559 --> 00:19:24,720 information, the processes, and the 511 00:19:24,720 --> 00:19:27,039 organization and collaborating in that 512 00:19:27,039 --> 00:19:29,679 learning with others. In his book, The 513 00:19:29,679 --> 00:19:31,760 Skill Code: How to save human ability in 514 00:19:31,760 --> 00:19:34,000 an age of intelligent machines, Matt 515 00:19:34,000 --> 00:19:36,080 Bean identifies three components that 516 00:19:36,080 --> 00:19:38,720 humans have used to learn for ages. 517 00:19:38,720 --> 00:19:40,799 There's a challenge, working close to 518 00:19:40,799 --> 00:19:43,120 the edge of your capacity with just 519 00:19:43,120 --> 00:19:44,799 enough stretch so that your capa 520 00:19:44,799 --> 00:19:48,160 capability improves. There's complexity, 521 00:19:48,160 --> 00:19:51,039 re wrestling with real messy problems 522 00:19:51,039 --> 00:19:52,799 and understanding the broader context 523 00:19:52,799 --> 00:19:56,080 they exist in. There's connection, 524 00:19:56,080 --> 00:19:58,720 building warm bonds of trust and respect 525 00:19:58,720 --> 00:20:01,919 and learning together, not in isolation. 526 00:20:01,919 --> 00:20:03,840 They talk about surgeons who go through 527 00:20:03,840 --> 00:20:06,160 the process of see one, do one, and 528 00:20:06,160 --> 00:20:08,000 teach one as part of the way that they 529 00:20:08,000 --> 00:20:11,600 learn. And developers do that, too. 530 00:20:11,600 --> 00:20:14,000 We've then got the expert novice bond. 531 00:20:14,000 --> 00:20:15,840 So, if juniors are made optional through 532 00:20:15,840 --> 00:20:17,760 replacement with AI that's only driven 533 00:20:17,760 --> 00:20:19,840 by seniors, what perspective are you 534 00:20:19,840 --> 00:20:22,559 losing? You're losing that junior's 535 00:20:22,559 --> 00:20:24,000 fresh eyes. 536 00:20:24,000 --> 00:20:25,760 along with your workflow replacement 537 00:20:25,760 --> 00:20:27,679 pipeline. 538 00:20:27,679 --> 00:20:29,520 So part of the expert novice bond 539 00:20:29,520 --> 00:20:32,000 includes pairing that working together 540 00:20:32,000 --> 00:20:34,320 one person driving one person critiquing 541 00:20:34,320 --> 00:20:35,919 and it helps to improve that code 542 00:20:35,919 --> 00:20:37,919 quality enhancing problem solving and 543 00:20:37,919 --> 00:20:39,600 sharing knowledge. You've got 544 00:20:39,600 --> 00:20:41,600 perspective providing different facets 545 00:20:41,600 --> 00:20:44,240 of understanding and experience and 546 00:20:44,240 --> 00:20:46,880 patience that allowing of time for 547 00:20:46,880 --> 00:20:49,760 critical thinking and space to practice. 548 00:20:49,760 --> 00:20:51,280 All of these things are part of how 549 00:20:51,280 --> 00:20:53,760 humans learn. And if we accelerate that 550 00:20:53,760 --> 00:20:55,919 timeline by doing it all through agents, 551 00:20:55,919 --> 00:20:59,679 what are we missing out on? 552 00:20:59,679 --> 00:21:01,520 So what impact has AI had on team 553 00:21:01,520 --> 00:21:03,840 dynamics and glue work? Well, Atlassian 554 00:21:03,840 --> 00:21:06,080 has found that juniors and new grads are 555 00:21:06,080 --> 00:21:09,200 1.5 times more likely to use AI tools 556 00:21:09,200 --> 00:21:10,880 and that the diversity of perspectives 557 00:21:10,880 --> 00:21:13,679 in AI usage benefit both ways. The 558 00:21:13,679 --> 00:21:15,760 experienced folks spot the slop and 559 00:21:15,760 --> 00:21:17,760 teach the domain. And the juniors bring 560 00:21:17,760 --> 00:21:19,280 the new tools, approaches, and 561 00:21:19,280 --> 00:21:22,400 techniques. Each group is compensating 562 00:21:22,400 --> 00:21:25,039 for precisely what the other lacks. 563 00:21:25,039 --> 00:21:28,559 Balance is key. They also found that AI 564 00:21:28,559 --> 00:21:30,480 is making jobs bigger, but the 565 00:21:30,480 --> 00:21:33,360 connections are lacking. AI widens your 566 00:21:33,360 --> 00:21:35,600 crossunctional reach, but not your 567 00:21:35,600 --> 00:21:37,440 relationships. 568 00:21:37,440 --> 00:21:40,640 Heavy AI users are up to 72% more likely 569 00:21:40,640 --> 00:21:43,120 to take on cross functional work. But 570 00:21:43,120 --> 00:21:44,960 when they were surveyed about how they 571 00:21:44,960 --> 00:21:47,360 were spending their time at work, every 572 00:21:47,360 --> 00:21:50,720 AI user group ranks time with colleagues 573 00:21:50,720 --> 00:21:52,960 dead last out of five workplace 574 00:21:52,960 --> 00:21:55,200 activities, while non-users ranked at 575 00:21:55,200 --> 00:21:57,840 third. And the suggestion is that AI 576 00:21:57,840 --> 00:21:59,760 users are collaborating across more 577 00:21:59,760 --> 00:22:02,480 teams, but without getting closer to 578 00:22:02,480 --> 00:22:06,679 anybody on those teams. 579 00:22:06,960 --> 00:22:09,679 All right. So, how can we handle glue 580 00:22:09,679 --> 00:22:12,240 work better? 581 00:22:12,240 --> 00:22:14,400 The glue person brings the right things 582 00:22:14,400 --> 00:22:17,200 together. The interfaces exist. They 583 00:22:17,200 --> 00:22:18,559 just need someone to make them work 584 00:22:18,559 --> 00:22:21,039 smoothly. You're creating lasting 585 00:22:21,039 --> 00:22:22,559 cohesion. 586 00:22:22,559 --> 00:22:24,640 The process band-aid stretches over 587 00:22:24,640 --> 00:22:27,120 holes. You're not enabling the system. 588 00:22:27,120 --> 00:22:30,960 You're masking its deficits. 589 00:22:30,960 --> 00:22:33,679 So, how can we do glue work better and 590 00:22:33,679 --> 00:22:36,559 who's involved? These are our players. 591 00:22:36,559 --> 00:22:38,400 We've got the managers. These are the 592 00:22:38,400 --> 00:22:40,240 people who've got the power to make the 593 00:22:40,240 --> 00:22:42,720 changes actually happen. We've then got 594 00:22:42,720 --> 00:22:45,039 the glue workers, those volunteers whose 595 00:22:45,039 --> 00:22:46,720 workload needs to be reviewed for 596 00:22:46,720 --> 00:22:49,760 appropriateness and fairness. Then we've 597 00:22:49,760 --> 00:22:52,480 got Teflon techs. They don't typically 598 00:22:52,480 --> 00:22:54,080 do as much glue work and they might need 599 00:22:54,080 --> 00:22:55,520 to be encouraged to pull their fair 600 00:22:55,520 --> 00:22:58,799 share. And then we've got AI tools. They 601 00:22:58,799 --> 00:23:00,880 can be a force multiplier or a time 602 00:23:00,880 --> 00:23:02,960 sync. And it really depends on how you 603 00:23:02,960 --> 00:23:05,520 use them. 604 00:23:05,520 --> 00:23:08,400 So here's my 5D funnel on how to h 605 00:23:08,400 --> 00:23:11,120 handle glue work better in a team. You 606 00:23:11,120 --> 00:23:13,280 start with discovering you define it, 607 00:23:13,280 --> 00:23:15,200 discuss it, document it, and distribute 608 00:23:15,200 --> 00:23:16,960 it. So I'll go through each of these in 609 00:23:16,960 --> 00:23:18,559 detail and you'll get a recap at the 610 00:23:18,559 --> 00:23:20,320 end. 611 00:23:20,320 --> 00:23:22,720 All right, so discover what are they 612 00:23:22,720 --> 00:23:24,960 supposed to be doing in their role and 613 00:23:24,960 --> 00:23:27,200 what do people actually do each day and 614 00:23:27,200 --> 00:23:29,520 how long have they been doing it for? 615 00:23:29,520 --> 00:23:31,840 Managers, you need to get clear on each 616 00:23:31,840 --> 00:23:34,159 team member's responsibilities and 617 00:23:34,159 --> 00:23:36,159 review your workflow tools and 618 00:23:36,159 --> 00:23:38,559 practices. 619 00:23:38,559 --> 00:23:41,039 Non-managers, keep a work diary for a 620 00:23:41,039 --> 00:23:42,880 sprint or a month, something that gives 621 00:23:42,880 --> 00:23:44,799 you enough time to figure out what the 622 00:23:44,799 --> 00:23:47,280 pattern of your work is like. You can 623 00:23:47,280 --> 00:23:48,880 try it in a digital diary. I use 624 00:23:48,880 --> 00:23:51,360 Confluence myself. And you can also use 625 00:23:51,360 --> 00:23:53,039 pen and paper, but if you do it 626 00:23:53,039 --> 00:23:55,600 digitally, you can get access to faster 627 00:23:55,600 --> 00:23:58,559 analysis. 628 00:23:58,559 --> 00:24:00,559 If you're using AI tools, you can use 629 00:24:00,559 --> 00:24:03,440 them to analyze human data, but also use 630 00:24:03,440 --> 00:24:05,039 the insight tools if they've got them 631 00:24:05,039 --> 00:24:08,000 available to figure out how humans are 632 00:24:08,000 --> 00:24:10,320 interacting with AI and whether they're 633 00:24:10,320 --> 00:24:12,720 creating or solving glue work with those 634 00:24:12,720 --> 00:24:15,600 tools. 635 00:24:15,600 --> 00:24:18,480 Two, define the breadth of it. What glue 636 00:24:18,480 --> 00:24:21,120 work is being done? Why is it being done 637 00:24:21,120 --> 00:24:24,480 that way? What need is being met? How 638 00:24:24,480 --> 00:24:27,360 long is it taking? Managers review the 639 00:24:27,360 --> 00:24:29,360 work your team has been doing compared 640 00:24:29,360 --> 00:24:31,039 with each of their job descriptions and 641 00:24:31,039 --> 00:24:34,240 with their allocated work. Glue workers 642 00:24:34,240 --> 00:24:36,000 analyze your data for patterns where you 643 00:24:36,000 --> 00:24:37,679 get sidetracked. This is when it can be 644 00:24:37,679 --> 00:24:39,919 helpful to point an agent at your work 645 00:24:39,919 --> 00:24:42,640 diary if it's a digital one. And teflon 646 00:24:42,640 --> 00:24:44,960 techs pay attention to the tasks that 647 00:24:44,960 --> 00:24:48,320 you put off palm off or plain avoid. 648 00:24:48,320 --> 00:24:51,840 Chances are that might be glue work. 649 00:24:51,840 --> 00:24:54,559 Thirdly, discuss it. talk about glue 650 00:24:54,559 --> 00:24:56,720 work up and down the chain at the time 651 00:24:56,720 --> 00:25:00,720 it emerges in retros in one-on ones and 652 00:25:00,720 --> 00:25:03,039 conditionally in performance reviews. So 653 00:25:03,039 --> 00:25:05,279 managers, you need to be open to all 654 00:25:05,279 --> 00:25:06,880 sorts of things coming out of the good 655 00:25:06,880 --> 00:25:09,360 woodwork. You need to set a good culture 656 00:25:09,360 --> 00:25:11,520 for how glue work is discussed so people 657 00:25:11,520 --> 00:25:13,520 have got that psychological safety to 658 00:25:13,520 --> 00:25:15,279 raise the things that might be a 659 00:25:15,279 --> 00:25:16,480 problem. 660 00:25:16,480 --> 00:25:19,039 Glue workers bring good evidence of the 661 00:25:19,039 --> 00:25:21,840 glue work you are doing and I recommend 662 00:25:21,840 --> 00:25:23,679 you decouple it from performance reviews 663 00:25:23,679 --> 00:25:26,320 initially. You don't want your raising 664 00:25:26,320 --> 00:25:28,640 of this issue to come across as a excuse 665 00:25:28,640 --> 00:25:30,480 or a justification for underperformance 666 00:25:30,480 --> 00:25:33,679 on this other task you've been assigned. 667 00:25:33,679 --> 00:25:35,919 Teflon text share what you've learned 668 00:25:35,919 --> 00:25:37,840 about how you do or don't handle glue 669 00:25:37,840 --> 00:25:39,919 work and try to be proactive in how you 670 00:25:39,919 --> 00:25:42,640 can contribute in future. 671 00:25:42,640 --> 00:25:46,159 Fourth, document it. Make glue work a 672 00:25:46,159 --> 00:25:48,960 part of your workflow. Give it tickets, 673 00:25:48,960 --> 00:25:52,000 tracking, planning, and estimates. Give 674 00:25:52,000 --> 00:25:55,200 it the time. So, managers, you need to 675 00:25:55,200 --> 00:25:57,440 champion champion the visibility and 676 00:25:57,440 --> 00:26:00,159 equitability of glue work in your team. 677 00:26:00,159 --> 00:26:02,080 Non-managers, participate 678 00:26:02,080 --> 00:26:04,000 enthusiastically and remember that all 679 00:26:04,000 --> 00:26:06,960 work needs to be done by somebody. 680 00:26:06,960 --> 00:26:09,120 If you're using AI tools, make the 681 00:26:09,120 --> 00:26:11,679 access to the parts of your organization 682 00:26:11,679 --> 00:26:13,679 and your data and the contributions that 683 00:26:13,679 --> 00:26:15,840 they make visible. And be realistic 684 00:26:15,840 --> 00:26:17,679 about AI's impact on the work that 685 00:26:17,679 --> 00:26:19,200 you're doing. Sometimes it's going to 686 00:26:19,200 --> 00:26:20,720 speed you up. Sometimes it's going to 687 00:26:20,720 --> 00:26:23,360 slow you down. 688 00:26:23,360 --> 00:26:26,320 And fifth and finally, distribute it. 689 00:26:26,320 --> 00:26:28,240 Who should and should not be doing glue 690 00:26:28,240 --> 00:26:31,840 work? Consider the speed, competency, 691 00:26:31,840 --> 00:26:33,840 availability, and opportunities for 692 00:26:33,840 --> 00:26:36,240 growth that the glue work could unlock. 693 00:26:36,240 --> 00:26:37,840 and the possibilities for the future of 694 00:26:37,840 --> 00:26:39,760 that human. 695 00:26:39,760 --> 00:26:41,279 Managers, 696 00:26:41,279 --> 00:26:42,880 it's your role to expand people's 697 00:26:42,880 --> 00:26:45,120 comfort zones and experience. And 698 00:26:45,120 --> 00:26:48,080 remember, somebody's capacity minus the 699 00:26:48,080 --> 00:26:49,840 glue work that they've been assigned is 700 00:26:49,840 --> 00:26:51,279 likely to be a more realistic 701 00:26:51,279 --> 00:26:53,039 representation of the throughput they're 702 00:26:53,039 --> 00:26:56,240 capable of. Glue workers, resist the 703 00:26:56,240 --> 00:26:58,480 temptation to volunteer. Leave space for 704 00:26:58,480 --> 00:27:00,240 others. 705 00:27:00,240 --> 00:27:02,400 If you've got a process, you can pair or 706 00:27:02,400 --> 00:27:03,919 teach somebody. Get that information 707 00:27:03,919 --> 00:27:06,080 into other people's heads. 708 00:27:06,080 --> 00:27:08,799 and teflon tech. Embrace the novelty or 709 00:27:08,799 --> 00:27:10,559 the pallet cleanser nature of glue work 710 00:27:10,559 --> 00:27:13,279 and bring your beginner's mind. If 711 00:27:13,279 --> 00:27:15,279 you're using AI tools to automate some 712 00:27:15,279 --> 00:27:17,360 things like meeting notes, do it 713 00:27:17,360 --> 00:27:19,840 thoughtfully and maintain control and 714 00:27:19,840 --> 00:27:21,919 accountability. Don't have it just go 715 00:27:21,919 --> 00:27:24,880 into weird places. 716 00:27:24,880 --> 00:27:26,559 All right. So, if you take one picture, 717 00:27:26,559 --> 00:27:28,159 this is the one to take. I'll hold the 718 00:27:28,159 --> 00:27:29,679 slide up here for a little bit if you if 719 00:27:29,679 --> 00:27:33,720 you want to take a picture on that one. 720 00:27:34,559 --> 00:27:37,360 Ready? Ready? You ready? Okay. Thank 721 00:27:37,360 --> 00:27:40,720 you. So, I have used AI imagery in the 722 00:27:40,720 --> 00:27:42,080 small pictures that are in this slide, 723 00:27:42,080 --> 00:27:43,679 but they were done through a cool 724 00:27:43,679 --> 00:27:45,360 platform that's unfortunately no longer 725 00:27:45,360 --> 00:27:47,520 available called test.design. 726 00:27:47,520 --> 00:27:51,579 And that's me. Thank you very much. 727 00:27:51,579 --> 00:27:53,599 [applause] 728 00:27:56,164 --> 00:27:58,184 [applause] 729 00:27:59,600 --> 00:28:02,919 Yeah. Okay. 730 00:28:06,399 --> 00:28:08,320 All right. Uh Jay, thank you very much 731 00:28:08,320 --> 00:28:10,000 for this amazing talk. It's a small 732 00:28:10,000 --> 00:28:12,000 appreciation. Here's a little token of 733 00:28:12,000 --> 00:28:12,240 our 734 00:28:12,240 --> 00:28:12,640 Thank you. 735 00:28:12,640 --> 00:28:14,504 Appreciation. 736 00:28:14,504 --> 00:28:16,000 [applause] 737 00:28:16,000 --> 00:28:18,399 All righty. Thank you. 738 00:28:18,399 --> 00:28:18,559 I 739 00:28:18,559 --> 00:28:21,279 think we have like time for maybe a 740 00:28:21,279 --> 00:28:23,679 couple of questions if there are. 741 00:28:23,679 --> 00:28:24,640 Got one over there. 742 00:28:24,640 --> 00:28:27,640 Yeah. 743 00:28:28,720 --> 00:28:32,080 Hello. Um I think my question is have 744 00:28:32,080 --> 00:28:34,240 you seen organizations 745 00:28:34,240 --> 00:28:36,080 uh build either specific job 746 00:28:36,080 --> 00:28:38,640 descriptions or performance management 747 00:28:38,640 --> 00:28:40,799 frameworks around the idea that it is 748 00:28:40,799 --> 00:28:43,520 someone's job to do glue work. 749 00:28:43,520 --> 00:28:45,120 I think we're starting to see some of 750 00:28:45,120 --> 00:28:46,960 that. So that meta intro article where I 751 00:28:46,960 --> 00:28:48,480 talked about some of the glue work roles 752 00:28:48,480 --> 00:28:50,240 that are available that was looking at 753 00:28:50,240 --> 00:28:51,919 other organizations that have done that. 754 00:28:51,919 --> 00:28:53,440 So I haven't seen it myself personally 755 00:28:53,440 --> 00:28:55,760 yet. I think I'm still fairly early in 756 00:28:55,760 --> 00:28:57,520 bringing some of these changes to places 757 00:28:57,520 --> 00:28:59,600 that I work in. Uh but I definitely 758 00:28:59,600 --> 00:29:01,520 think that there are points where we can 759 00:29:01,520 --> 00:29:03,840 consider okay everybody's job 760 00:29:03,840 --> 00:29:05,600 descriptions are kind of changing at the 761 00:29:05,600 --> 00:29:07,440 moment. What does being developer even 762 00:29:07,440 --> 00:29:09,200 mean? You know, how much code are you 763 00:29:09,200 --> 00:29:11,440 writing? What's going on here? And think 764 00:29:11,440 --> 00:29:14,880 about how people need to get things 765 00:29:14,880 --> 00:29:16,480 done, not just if they're the ones 766 00:29:16,480 --> 00:29:19,440 writing the code. So I haven't seen it 767 00:29:19,440 --> 00:29:20,960 myself personally, but I have read about 768 00:29:20,960 --> 00:29:22,960 it. So there's lots of opportunities for 769 00:29:22,960 --> 00:29:25,520 how we can redefine what are people 770 00:29:25,520 --> 00:29:29,840 doing. Yeah. Anybody else? 771 00:29:29,840 --> 00:29:30,640 Just one more. 772 00:29:30,640 --> 00:29:31,840 One more. 773 00:29:31,840 --> 00:29:34,159 As someone who can often find myself 774 00:29:34,159 --> 00:29:36,480 getting put tossed with a lot of extra 775 00:29:36,480 --> 00:29:39,279 glue work, how can I and anyone else in 776 00:29:39,279 --> 00:29:40,960 this situation 777 00:29:40,960 --> 00:29:42,480 properly 778 00:29:42,480 --> 00:29:44,559 like try to focus more on technical work 779 00:29:44,559 --> 00:29:46,960 as well as like not saying, "Oh, I'm not 780 00:29:46,960 --> 00:29:48,640 doing that glue work, but saying I want 781 00:29:48,640 --> 00:29:51,039 to be doing technical work as well." 782 00:29:51,039 --> 00:29:53,279 That's a great question. And part of it 783 00:29:53,279 --> 00:29:54,720 is understanding what the glue work 784 00:29:54,720 --> 00:29:57,279 you're doing is and also making sure 785 00:29:57,279 --> 00:29:59,279 that whoever you're reporting to is 786 00:29:59,279 --> 00:30:01,279 aware of it as well. So it's it it can 787 00:30:01,279 --> 00:30:03,840 be really hard to try and change it if 788 00:30:03,840 --> 00:30:05,279 it's the culture that you're immersed 789 00:30:05,279 --> 00:30:08,320 in. You sort of go uh if I don't do it, 790 00:30:08,320 --> 00:30:09,600 who's going to do it? What's going to 791 00:30:09,600 --> 00:30:11,760 happen? So definitely communicating 792 00:30:11,760 --> 00:30:13,600 about the glue work you're doing is 793 00:30:13,600 --> 00:30:15,919 really important. Um and getting that 794 00:30:15,919 --> 00:30:17,760 buyin from your managers or or the 795 00:30:17,760 --> 00:30:20,080 people who assign who are assigning you 796 00:30:20,080 --> 00:30:22,240 work. And this is where having good 797 00:30:22,240 --> 00:30:24,159 boundaries around the work that you're 798 00:30:24,159 --> 00:30:25,919 doing and what people are bringing to 799 00:30:25,919 --> 00:30:29,760 you uh can be helpful. So 800 00:30:29,760 --> 00:30:31,360 if you get somebody that's always like, 801 00:30:31,360 --> 00:30:33,360 "Oh, can you help me with this query? I 802 00:30:33,360 --> 00:30:34,559 need to get this done for a quick 803 00:30:34,559 --> 00:30:37,440 report." You go, "Great. My manager's 804 00:30:37,440 --> 00:30:39,120 asked that all of this kind of work gets 805 00:30:39,120 --> 00:30:40,399 routed through them first for 806 00:30:40,399 --> 00:30:42,480 prioritization." So, you know, you can 807 00:30:42,480 --> 00:30:44,399 deflect it onto the person who's in 808 00:30:44,399 --> 00:30:46,880 charge of assigning your work. Or if 809 00:30:46,880 --> 00:30:49,120 you've got somebody coming to you and 810 00:30:49,120 --> 00:30:51,120 saying, "Hey, I need you to do this 811 00:30:51,120 --> 00:30:52,960 quickly." I'm like, "Great. Here are the 812 00:30:52,960 --> 00:30:54,640 other priorities I've got lined up. 813 00:30:54,640 --> 00:30:56,720 Which of these would you like to have me 814 00:30:56,720 --> 00:30:58,880 defer?" So, it's a case of pushing that 815 00:30:58,880 --> 00:31:02,320 decision about how your time is spent 816 00:31:02,320 --> 00:31:06,000 back onto them so that the person who is 817 00:31:06,000 --> 00:31:07,840 asking you to do it or the person you're 818 00:31:07,840 --> 00:31:10,159 accountable to is the one making that 819 00:31:10,159 --> 00:31:12,480 call and and keep documentation of it. 820 00:31:12,480 --> 00:31:14,559 So that work diary is is a really good 821 00:31:14,559 --> 00:31:16,720 starting point to sort of understand it 822 00:31:16,720 --> 00:31:18,480 and then you've got some evidence that 823 00:31:18,480 --> 00:31:20,399 you can use to discuss it better. 824 00:31:20,399 --> 00:31:20,799 Thank you. 825 00:31:20,799 --> 00:31:22,240 You're welcome. 826 00:31:22,240 --> 00:31:23,360 Thanks very much J. 827 00:31:23,360 --> 00:31:26,520 Thank you.