WEBVTT - The AI playbook of ELMO Software’s President, Joseph Lyons

0:00:00.160 --> 0:00:04.880
<v Speaker 1>This episode is sponsored by Elmo the Complete AI Workforce Platform.

0:00:05.240 --> 0:00:09.720
<v Speaker 1>It unifies HR, payroll and rostering on one platform with

0:00:09.920 --> 0:00:14.040
<v Speaker 1>native AI that turns connected data into trusted answers. You

0:00:14.080 --> 0:00:18.880
<v Speaker 1>can act on the best prepared person in the room

0:00:19.120 --> 0:00:23.560
<v Speaker 1>wins right. But when your day is back to back meetings,

0:00:23.720 --> 0:00:27.320
<v Speaker 1>unread emails, and conversations you haven't had time to think through,

0:00:27.760 --> 0:00:32.600
<v Speaker 1>preparation feels like a luxury you can't afford. But Joseph

0:00:32.680 --> 0:00:35.159
<v Speaker 1>Lyons has a better answer, and it comes down to

0:00:35.240 --> 0:00:40.120
<v Speaker 1>how he uses AI. Joseph is the president of Elmo Software,

0:00:40.520 --> 0:00:43.120
<v Speaker 1>and I love this conversation because he isn't just talking

0:00:43.159 --> 0:00:46.240
<v Speaker 1>about how he uses AI in theory. In this chat,

0:00:46.400 --> 0:00:49.640
<v Speaker 1>he pulls back the curtain to share how he's using

0:00:49.680 --> 0:00:54.600
<v Speaker 1>AI every single day in super practical ways. We get

0:00:54.600 --> 0:00:59.280
<v Speaker 1>into the personalized daily intelligence brief his entire exec team

0:00:59.400 --> 0:01:02.320
<v Speaker 1>wakes up to every morning. We talk about how he

0:01:02.360 --> 0:01:05.280
<v Speaker 1>turned a six hour strategy workshop transcript into something that

0:01:05.400 --> 0:01:08.920
<v Speaker 1>was actually usable and actionable, and the way he now

0:01:09.120 --> 0:01:14.319
<v Speaker 1>roleplays tough negotiations and difficult performance conversations with AI, including

0:01:14.440 --> 0:01:23.160
<v Speaker 1>on his drive to work. Welcome to How I Work,

0:01:23.360 --> 0:01:27.520
<v Speaker 1>a show about habits rituals and strategies for optimizing your day.

0:01:28.120 --> 0:01:34.959
<v Speaker 1>I'm your host, doctor Amantha imber Joe. I want to

0:01:35.040 --> 0:01:39.319
<v Speaker 1>start with this ELT productivity agent that you have built.

0:01:39.400 --> 0:01:42.319
<v Speaker 1>Can you tell me what this is and what it does?

0:01:42.680 --> 0:01:45.720
<v Speaker 2>Yeah, Look, I think we have been fortunate in that

0:01:45.920 --> 0:01:48.919
<v Speaker 2>at Elmo, we've been able to deploy a pretty amazing

0:01:48.960 --> 0:01:52.200
<v Speaker 2>tool called Glean and that's like an enterprise AI layer

0:01:52.240 --> 0:01:54.880
<v Speaker 2>that we've got and it integrates with all of the

0:01:55.000 --> 0:01:57.240
<v Speaker 2>various systems that we have. So we've got, you know,

0:01:57.320 --> 0:01:59.800
<v Speaker 2>obviously our g suite, We've got Slack, which is used

0:02:00.000 --> 0:02:05.240
<v Speaker 2>extensively Salesforce, a range of different tools. And then what's

0:02:05.280 --> 0:02:08.240
<v Speaker 2>exciting about that platform is that it's actually allowed us

0:02:08.240 --> 0:02:11.440
<v Speaker 2>to build agents in natural language. It doesn't mean that

0:02:11.480 --> 0:02:13.600
<v Speaker 2>you need to have an engineering background, you don't need

0:02:13.680 --> 0:02:16.680
<v Speaker 2>to understand how to code. You're pretty much able to

0:02:16.680 --> 0:02:19.120
<v Speaker 2>describe what you want the agent to do, and then

0:02:19.160 --> 0:02:22.160
<v Speaker 2>it connects to the data sources and runs. So what

0:02:22.240 --> 0:02:24.520
<v Speaker 2>we've been able to do for our ELT is actually

0:02:24.520 --> 0:02:27.600
<v Speaker 2>build a productivity agent and then it allows us to

0:02:27.800 --> 0:02:29.040
<v Speaker 2>pretty much run.

0:02:29.320 --> 0:02:31.519
<v Speaker 3>For daily for each of our exec team.

0:02:31.560 --> 0:02:34.320
<v Speaker 2>It connects all of the applications that they've got and

0:02:34.360 --> 0:02:37.959
<v Speaker 2>it pulls together a prioritized daily brief, summarizes all the

0:02:38.000 --> 0:02:40.400
<v Speaker 2>key emails that we need to do, outlines the next

0:02:40.440 --> 0:02:43.000
<v Speaker 2>seven days of meetings, any sort of open action items.

0:02:43.680 --> 0:02:45.760
<v Speaker 3>It surfaces and flags.

0:02:45.320 --> 0:02:47.799
<v Speaker 2>What's high importance, what might need to go to the board,

0:02:48.320 --> 0:02:51.680
<v Speaker 2>any key customer meetings, or any major risks or escalations.

0:02:51.720 --> 0:02:54.919
<v Speaker 2>So it's been a massive timesaver for all of us

0:02:55.040 --> 0:02:58.120
<v Speaker 2>in the exec and it makes sure that we're acting

0:02:58.160 --> 0:02:59.359
<v Speaker 2>on the things that we need to do, and it

0:02:59.440 --> 0:03:02.480
<v Speaker 2>is delivered beautifully into Slack every morning for us.

0:03:02.560 --> 0:03:05.880
<v Speaker 1>That's amazing. And so does every member of the ELT

0:03:06.000 --> 0:03:09.400
<v Speaker 1>get their own personalized version of that report based on

0:03:09.440 --> 0:03:10.800
<v Speaker 1>what they need to know?

0:03:11.280 --> 0:03:12.399
<v Speaker 3>Yes, exactly.

0:03:12.560 --> 0:03:15.440
<v Speaker 1>Wow, that is absolutely amazing. How much time do you

0:03:15.440 --> 0:03:18.080
<v Speaker 1>reckon that has saved you in terms of all the

0:03:18.080 --> 0:03:20.600
<v Speaker 1>things that you need to be across Joe.

0:03:20.360 --> 0:03:23.320
<v Speaker 2>I think it's not just saving time about planning for

0:03:23.440 --> 0:03:26.960
<v Speaker 2>that day, but it's also saving time in between a

0:03:27.080 --> 0:03:29.360
<v Speaker 2>very full meeting schedule because it allows you to have

0:03:29.400 --> 0:03:32.440
<v Speaker 2>a reference guide and a layer to move through the day.

0:03:32.480 --> 0:03:34.040
<v Speaker 3>So I reckon it's saving at least than an hour

0:03:34.120 --> 0:03:34.720
<v Speaker 3>or two a day.

0:03:34.920 --> 0:03:38.800
<v Speaker 1>Now. I know that you know using AI to improve

0:03:38.960 --> 0:03:42.000
<v Speaker 1>how effective we are with meetings is pretty common. But

0:03:42.280 --> 0:03:46.119
<v Speaker 1>tell me about a six hour workshop that I think

0:03:46.200 --> 0:03:49.920
<v Speaker 1>you attended this week, and tell me how you used

0:03:49.960 --> 0:03:52.320
<v Speaker 1>AI for that, because that is a very long workshop.

0:03:52.560 --> 0:03:55.040
<v Speaker 3>Yeah, it is. I mean it's a big time commitment.

0:03:55.040 --> 0:03:57.520
<v Speaker 2>It was an important strategy workshop that we were running,

0:03:57.560 --> 0:03:59.880
<v Speaker 2>and we had a number of senior execs that were

0:04:00.080 --> 0:04:03.360
<v Speaker 2>in there, and there were some important strategic conversations that

0:04:03.400 --> 0:04:06.040
<v Speaker 2>we needed to get through, and I found that Zoom

0:04:06.200 --> 0:04:08.840
<v Speaker 2>AI companion it really only gives you kind of a

0:04:08.880 --> 0:04:12.320
<v Speaker 2>surface level, bullet point recap of what was discussed then,

0:04:12.360 --> 0:04:15.640
<v Speaker 2>particularly when you're talking about a six hour conversation that

0:04:15.720 --> 0:04:19.240
<v Speaker 2>could cover a range of different topics. We did get

0:04:19.240 --> 0:04:21.360
<v Speaker 2>a zoom summary of it, but I found it didn't

0:04:21.400 --> 0:04:24.520
<v Speaker 2>provide us the full detail and get to us a

0:04:24.600 --> 0:04:28.159
<v Speaker 2>clear set of outcome. So I went back into Zoom

0:04:28.240 --> 0:04:32.000
<v Speaker 2>and that actually pulled out the full detailed transcript of

0:04:32.040 --> 0:04:35.520
<v Speaker 2>the day and I uploaded that into Claude.

0:04:35.880 --> 0:04:36.799
<v Speaker 3>What came back.

0:04:36.680 --> 0:04:39.360
<v Speaker 2>Genuinely kind of blew me away, because what it did

0:04:39.480 --> 0:04:42.000
<v Speaker 2>was it captured all of the wisdom and intelligence of

0:04:42.040 --> 0:04:45.679
<v Speaker 2>everything that we discussed, the strategic decisions that were made.

0:04:46.000 --> 0:04:49.000
<v Speaker 2>Also the ones that were still opened. It pulled together

0:04:49.040 --> 0:04:51.000
<v Speaker 2>all of the key themes and the changes that the

0:04:51.080 --> 0:04:54.480
<v Speaker 2>room agreed on. It then tabled for me the next steps.

0:04:54.839 --> 0:04:57.320
<v Speaker 2>It summarized who were the key owners, and then it

0:04:57.400 --> 0:05:01.960
<v Speaker 2>reorganized everything by topic and sort of the chronological order.

0:05:01.800 --> 0:05:02.240
<v Speaker 3>Of the day.

0:05:02.720 --> 0:05:05.920
<v Speaker 2>I then prompted it again to pull together an artifact

0:05:05.920 --> 0:05:08.920
<v Speaker 2>which I could use to share and cascade with the

0:05:09.200 --> 0:05:12.160
<v Speaker 2>team members that weren't in there, and then a full

0:05:12.279 --> 0:05:16.400
<v Speaker 2>email and common structure that I was allowed enabled me to.

0:05:15.880 --> 0:05:18.240
<v Speaker 3>Use that to cascade to the rest of the team

0:05:18.279 --> 0:05:18.560
<v Speaker 3>as well.

0:05:18.680 --> 0:05:21.320
<v Speaker 2>So it was pretty mind blowing and kind of went

0:05:21.839 --> 0:05:24.679
<v Speaker 2>a much larger step further than what you would typically

0:05:24.720 --> 0:05:25.560
<v Speaker 2>get with Zoom.

0:05:25.560 --> 0:05:28.560
<v Speaker 1>That's so interesting and just the value that you got

0:05:28.600 --> 0:05:34.520
<v Speaker 1>from taking that out of Zoom's AI companion and into Claude.

0:05:34.760 --> 0:05:37.920
<v Speaker 1>Can you tell me some of the things that perhaps

0:05:37.920 --> 0:05:41.240
<v Speaker 1>you were thinking about when you were prompting Claude to

0:05:41.400 --> 0:05:44.599
<v Speaker 1>get the best possible output from it, and whether you

0:05:44.640 --> 0:05:48.039
<v Speaker 1>know you used the thinking model or just on AUDO,

0:05:48.120 --> 0:05:49.120
<v Speaker 1>tell me a little bit about that.

0:05:49.279 --> 0:05:52.560
<v Speaker 2>Yeah, so I'd already used Claude to prepare, help me

0:05:52.600 --> 0:05:55.920
<v Speaker 2>prepare the agenda, help me prepare and consolidate some of

0:05:55.960 --> 0:06:00.039
<v Speaker 2>the pre context and materials, and then you know the

0:06:00.080 --> 0:06:03.240
<v Speaker 2>outputs in the objectives of what we had for the day.

0:06:03.279 --> 0:06:06.200
<v Speaker 2>So Claude already had the context of what I was

0:06:06.800 --> 0:06:11.120
<v Speaker 2>hoping to achieve for the entire workshop, and then having

0:06:11.440 --> 0:06:14.880
<v Speaker 2>the full transcript and conversation that was following that it

0:06:14.960 --> 0:06:17.960
<v Speaker 2>already had connected the dots between the conversation and what

0:06:18.279 --> 0:06:20.839
<v Speaker 2>the goals were. So it was then simple to be

0:06:20.920 --> 0:06:24.000
<v Speaker 2>able to prompt it in saying based on everything that

0:06:24.080 --> 0:06:26.839
<v Speaker 2>you knew of what we wanted to achieve the entire

0:06:27.160 --> 0:06:30.359
<v Speaker 2>span of the conversation. And then we were really conscious

0:06:30.400 --> 0:06:32.360
<v Speaker 2>while we were in the meeting, like no one needing

0:06:32.400 --> 0:06:36.400
<v Speaker 2>to take notes as you typically would. We would actively

0:06:37.000 --> 0:06:40.040
<v Speaker 2>talk to the AI and make sure that we were

0:06:40.040 --> 0:06:43.320
<v Speaker 2>clear on capturing actions that needed to be done or

0:06:43.360 --> 0:06:46.680
<v Speaker 2>owners or timeframes. So it did make the summarizing of

0:06:46.720 --> 0:06:50.720
<v Speaker 2>the full day far easier because it did capture everything

0:06:50.760 --> 0:06:51.640
<v Speaker 2>that was spoken.

0:06:51.360 --> 0:06:52.000
<v Speaker 3>Throughout the day.

0:06:52.120 --> 0:06:54.960
<v Speaker 1>I love the idea and something we do at Inventium

0:06:55.200 --> 0:06:58.600
<v Speaker 1>is that if we are gathering in person for an

0:06:58.600 --> 0:07:01.800
<v Speaker 1>off site or something like that, generally Granola is our

0:07:01.839 --> 0:07:04.560
<v Speaker 1>meeting transcription tool of choice and we will just pop

0:07:04.560 --> 0:07:07.240
<v Speaker 1>a phone with granola running in the middle of the

0:07:07.240 --> 0:07:10.240
<v Speaker 1>workshop room, and it just allows us all to be

0:07:10.280 --> 0:07:12.520
<v Speaker 1>so much more present because no one is needing to

0:07:12.560 --> 0:07:15.120
<v Speaker 1>take notes, no one is needing to capture things on

0:07:15.160 --> 0:07:18.920
<v Speaker 1>a whiteboard. And then because we've got that full transcript,

0:07:18.920 --> 0:07:20.680
<v Speaker 1>we can put it into any AI tool that we

0:07:20.720 --> 0:07:22.800
<v Speaker 1>want prompted in ways that are going to get the

0:07:22.840 --> 0:07:26.920
<v Speaker 1>best output. So I love that example, Joe. Now, something

0:07:26.960 --> 0:07:30.200
<v Speaker 1>that I would imagine your job as president of Elmo

0:07:30.360 --> 0:07:33.920
<v Speaker 1>involves a lot is negotiating. Can you tell me how

0:07:34.000 --> 0:07:36.960
<v Speaker 1>you've used AI to help with negotiations?

0:07:37.240 --> 0:07:39.960
<v Speaker 2>Yeah, I mean, we're also fortunate that we've deployed a

0:07:40.000 --> 0:07:43.320
<v Speaker 2>tool called Gong, which is also a pretty mind blowing

0:07:43.880 --> 0:07:47.520
<v Speaker 2>platform because what it does is it captures every single

0:07:47.800 --> 0:07:50.320
<v Speaker 2>interaction that we have with our customers, so every call,

0:07:50.400 --> 0:07:53.240
<v Speaker 2>every meeting, any piece of content that they've engaged with,

0:07:53.360 --> 0:07:56.320
<v Speaker 2>any presentation or even voice messages that may be left.

0:07:56.520 --> 0:07:59.520
<v Speaker 2>So it gives us a complete history of the relationship

0:07:59.560 --> 0:08:02.040
<v Speaker 2>with had with either a new customer that we're looking

0:08:02.080 --> 0:08:05.040
<v Speaker 2>to engage with or an existing customer as well. And

0:08:05.080 --> 0:08:07.600
<v Speaker 2>then I guess where it gets really powerful is when

0:08:07.600 --> 0:08:11.239
<v Speaker 2>you feed that history into AI. Before I walk into

0:08:11.280 --> 0:08:14.920
<v Speaker 2>a conversation or a negotiation. Because what I've been able

0:08:14.960 --> 0:08:17.680
<v Speaker 2>to do is extract all of the intelligence from any

0:08:17.720 --> 0:08:19.680
<v Speaker 2>of the pre meetings that have occurred with some of

0:08:19.680 --> 0:08:24.320
<v Speaker 2>our team. I can upload the detail of the partnership

0:08:24.320 --> 0:08:26.880
<v Speaker 2>contract that we're trying to work through. What I can

0:08:26.920 --> 0:08:29.800
<v Speaker 2>then do is prompt and explain kind of what my

0:08:29.840 --> 0:08:34.079
<v Speaker 2>commercial position is, what theirs may be, Prompt and understand

0:08:34.120 --> 0:08:36.400
<v Speaker 2>what is some of the likely objections that might come

0:08:36.440 --> 0:08:40.400
<v Speaker 2>out of the discussion, and then help define pretty much

0:08:40.440 --> 0:08:42.599
<v Speaker 2>a cheat sheet that I can keep hoping during my

0:08:42.679 --> 0:08:45.920
<v Speaker 2>laptop in the call, and effectively it's kind of mapped

0:08:45.920 --> 0:08:48.240
<v Speaker 2>out what the game theory of that negotiation might be.

0:08:48.720 --> 0:08:49.720
<v Speaker 3>I don't just rely.

0:08:49.600 --> 0:08:52.320
<v Speaker 2>On the intelligence though what we've got in Gong obviously,

0:08:52.720 --> 0:08:54.760
<v Speaker 2>make sure that I'm always speaking to the humans that

0:08:54.800 --> 0:08:58.080
<v Speaker 2>have engaged with customers or prospects as well, because it

0:08:58.120 --> 0:08:59.960
<v Speaker 2>is important to get that human overlay as well.

0:09:00.160 --> 0:09:02.800
<v Speaker 1>I'm wondering, Joe, is there an example of a meeting

0:09:02.800 --> 0:09:04.640
<v Speaker 1>you know that you've had in the last few weeks

0:09:04.679 --> 0:09:08.040
<v Speaker 1>that you know maybe has required some tough negotiation and

0:09:08.400 --> 0:09:11.760
<v Speaker 1>just some specific examples of how that has helped you.

0:09:11.800 --> 0:09:16.319
<v Speaker 1>Because it sounds incredibly powerful. I'm very aware of gong.

0:09:16.720 --> 0:09:19.160
<v Speaker 1>We don't personally use it an inventium, but yeah, a

0:09:19.240 --> 0:09:22.400
<v Speaker 1>super cool tool. Yes, I'd love to hear an example

0:09:22.440 --> 0:09:25.920
<v Speaker 1>maybe of where in this AI world do you approach

0:09:26.000 --> 0:09:28.319
<v Speaker 1>the negotiation differently and got to a better result than

0:09:28.559 --> 0:09:29.840
<v Speaker 1>you know three or four years ago.

0:09:30.080 --> 0:09:32.480
<v Speaker 2>Yeah, we had a partnership agreement that we were looking

0:09:32.520 --> 0:09:34.640
<v Speaker 2>to strike and this has only happened just over the

0:09:34.720 --> 0:09:37.480
<v Speaker 2>last couple of weeks. We had what their position was

0:09:37.520 --> 0:09:40.880
<v Speaker 2>on the commercial terms, what mine and the businesses were

0:09:41.360 --> 0:09:45.080
<v Speaker 2>both on pricing, on contract duration, and on some of

0:09:45.120 --> 0:09:49.080
<v Speaker 2>the sort of technical components. So it allowed me to

0:09:49.800 --> 0:09:52.840
<v Speaker 2>prompt it to say, well, where were the points of

0:09:52.880 --> 0:09:57.679
<v Speaker 2>contention and help empathize what the partner's position was, but

0:09:57.840 --> 0:10:00.640
<v Speaker 2>most importantly what ours was, and then what would be

0:10:00.720 --> 0:10:03.880
<v Speaker 2>some recommended solutions that we could come to terms on,

0:10:04.320 --> 0:10:06.840
<v Speaker 2>and most importantly, how to kind of build a talk

0:10:06.920 --> 0:10:09.360
<v Speaker 2>track and an empathy layer so that we could actually

0:10:09.440 --> 0:10:13.000
<v Speaker 2>reach a conclusion. And fortunately we've just had sign off

0:10:13.040 --> 0:10:15.640
<v Speaker 2>and we're moving forward with getting a contract stood up,

0:10:15.640 --> 0:10:16.800
<v Speaker 2>so it did actually work.

0:10:16.960 --> 0:10:20.440
<v Speaker 1>Yeah, awesome. I would imagine that that would also change

0:10:20.679 --> 0:10:24.240
<v Speaker 1>your kind of mental and emotional state going into a negotiation,

0:10:24.280 --> 0:10:26.719
<v Speaker 1>which can obviously be quite stressful situations.

0:10:26.960 --> 0:10:27.160
<v Speaker 3>Yeah.

0:10:27.200 --> 0:10:30.400
<v Speaker 2>Absolutely, I mean I think it allows me to feel

0:10:30.480 --> 0:10:34.760
<v Speaker 2>more confident knowing all of the variables, feeling really prepared,

0:10:35.120 --> 0:10:39.120
<v Speaker 2>knowing what's transpired before, and then knowing what the points

0:10:39.120 --> 0:10:41.760
<v Speaker 2>of contention are and then you know what option AB

0:10:41.920 --> 0:10:43.600
<v Speaker 2>or C might be on how to respond to that.

0:10:43.840 --> 0:10:45.679
<v Speaker 1>Now, something that I know a lot of people use

0:10:45.720 --> 0:10:49.800
<v Speaker 1>AI for is role playing conversations. And I would love

0:10:50.120 --> 0:10:53.120
<v Speaker 1>to know for you as the president of an organization,

0:10:53.440 --> 0:10:56.160
<v Speaker 1>and I feel like people think of, well, you're running

0:10:56.200 --> 0:10:59.000
<v Speaker 1>an organization, like surely you're confident, you don't need to

0:10:59.080 --> 0:11:01.800
<v Speaker 1>like you know, PA for meetings, like you know you've

0:11:01.800 --> 0:11:03.800
<v Speaker 1>got it. But I know that you do use AI

0:11:03.920 --> 0:11:05.800
<v Speaker 1>in this way, Jo, And i'd love to hear like

0:11:05.880 --> 0:11:09.520
<v Speaker 1>a couple of examples of where and how you've used

0:11:09.559 --> 0:11:12.120
<v Speaker 1>AI to prepare for tough conversations.

0:11:12.640 --> 0:11:14.480
<v Speaker 3>Yeah. So it's a good question, Amantha.

0:11:14.520 --> 0:11:18.800
<v Speaker 2>I recently had a pretty difficult performance conversation with a

0:11:18.840 --> 0:11:21.839
<v Speaker 2>team member that I needed to have and wanted to

0:11:21.880 --> 0:11:23.840
<v Speaker 2>make sure that I was really well prepared and that

0:11:23.880 --> 0:11:26.720
<v Speaker 2>I'd thought through how I was going to frame that

0:11:26.760 --> 0:11:31.319
<v Speaker 2>feedback the specific examples, and I quite often use the

0:11:31.440 --> 0:11:34.400
<v Speaker 2>drive into work as time to think. And I have

0:11:34.559 --> 0:11:38.800
<v Speaker 2>connected Clawed Voice directly through so I can actually speak

0:11:38.800 --> 0:11:40.880
<v Speaker 2>to it in my car. So I was able to

0:11:40.920 --> 0:11:44.880
<v Speaker 2>just in a conversational way explain the situation, explain the

0:11:44.920 --> 0:11:48.200
<v Speaker 2>person that I was dealing with, the specific challenge that

0:11:48.240 --> 0:11:51.720
<v Speaker 2>we were having to face into the feedback that I

0:11:51.920 --> 0:11:54.520
<v Speaker 2>needed to provide. And it was amazing what it came

0:11:54.559 --> 0:11:58.040
<v Speaker 2>back with in terms of giving me a pretty cohesive talk.

0:11:57.960 --> 0:11:59.400
<v Speaker 3>Track and how to approach it.

0:12:00.040 --> 0:12:03.640
<v Speaker 2>And I could then refine it based on further prompts

0:12:03.640 --> 0:12:06.640
<v Speaker 2>to make sure that it was engaging, that we got

0:12:06.679 --> 0:12:09.920
<v Speaker 2>to a clear outcome, and that it was a mutually

0:12:09.920 --> 0:12:11.040
<v Speaker 2>beneficial conversation.

0:12:11.360 --> 0:12:15.000
<v Speaker 1>And how are you prompting the AI in that kind

0:12:15.040 --> 0:12:18.760
<v Speaker 1>of an instance, Like how much or how little information

0:12:18.960 --> 0:12:22.679
<v Speaker 1>are you feeding at the start of that conversation if

0:12:22.679 --> 0:12:24.079
<v Speaker 1>you like, in the car, I think.

0:12:23.880 --> 0:12:24.640
<v Speaker 3>The more the better.

0:12:24.800 --> 0:12:28.400
<v Speaker 2>So it's explaining who the person is on the other side,

0:12:28.520 --> 0:12:32.520
<v Speaker 2>what role they're in their history, either working with us

0:12:32.559 --> 0:12:34.959
<v Speaker 2>or it might be a customer for example. You know,

0:12:35.040 --> 0:12:37.360
<v Speaker 2>what are the key challenges or the opportunities that we're

0:12:37.360 --> 0:12:40.120
<v Speaker 2>needing to work through being as specific as you possibly can,

0:12:40.679 --> 0:12:43.440
<v Speaker 2>so that the AI has full context of the environment

0:12:43.760 --> 0:12:45.280
<v Speaker 2>of what we're chatting about.

0:12:45.360 --> 0:12:50.079
<v Speaker 1>Now. Obviously, as a leader, like you're getting so much

0:12:50.400 --> 0:12:54.240
<v Speaker 1>input from so many sources, you're getting I would imagine

0:12:54.400 --> 0:12:58.600
<v Speaker 1>weekly reports from different teams, different functions. Can you tell

0:12:58.640 --> 0:13:03.040
<v Speaker 1>me how you're using AI to synthesize all that information

0:13:03.200 --> 0:13:05.080
<v Speaker 1>that must be incoming in your world.

0:13:05.240 --> 0:13:07.760
<v Speaker 2>That's probably been the biggest game changer for me, and

0:13:07.800 --> 0:13:10.720
<v Speaker 2>particularly in the last few months. I'm sure many of

0:13:10.720 --> 0:13:13.400
<v Speaker 2>your listeners and most leaders would experience this. You find

0:13:13.400 --> 0:13:18.040
<v Speaker 2>yourself getting different reports in different formats from five or

0:13:18.080 --> 0:13:22.840
<v Speaker 2>six different functions, and each presenting their own complex data sets.

0:13:22.920 --> 0:13:25.960
<v Speaker 2>They might have their own perspective, and in many cases

0:13:26.040 --> 0:13:28.760
<v Speaker 2>there's not really any single person that's holding the full picture.

0:13:29.679 --> 0:13:32.920
<v Speaker 2>So what I've experienced particularly lately is that now I

0:13:32.960 --> 0:13:35.680
<v Speaker 2>can ingest the range of different information I can pull.

0:13:35.760 --> 0:13:38.840
<v Speaker 2>Sales data in might be marketing metrics, it might be

0:13:39.400 --> 0:13:44.640
<v Speaker 2>customer insights, it might be performance cross border with different currencies.

0:13:44.720 --> 0:13:47.520
<v Speaker 2>It might be product and engineering updates as well. So

0:13:47.559 --> 0:13:51.199
<v Speaker 2>I can ingest a lot of different formats and data sets,

0:13:51.240 --> 0:13:53.480
<v Speaker 2>and what I'm finding is that the AI can now

0:13:53.559 --> 0:13:57.480
<v Speaker 2>produce really cohesive and consistent and useful reporting, which I

0:13:57.520 --> 0:14:00.640
<v Speaker 2>think previously would have taken hours and hours and hours

0:14:00.640 --> 0:14:03.600
<v Speaker 2>to consolidate. And what I'm seeing is that the quality

0:14:03.640 --> 0:14:07.000
<v Speaker 2>is incredibly high because it's spotting patterns across the data

0:14:07.440 --> 0:14:09.160
<v Speaker 2>and things that I may have missed if I've just

0:14:09.200 --> 0:14:10.680
<v Speaker 2>read each report individually.

0:14:11.760 --> 0:14:14.360
<v Speaker 1>And how is that set up behind the scenes? Is

0:14:14.360 --> 0:14:18.079
<v Speaker 1>that an automated process now is that you're manually uploading

0:14:18.120 --> 0:14:20.840
<v Speaker 1>documents to Glean or to claude, Like, what does that

0:14:20.880 --> 0:14:21.960
<v Speaker 1>look like behind the scenes.

0:14:22.200 --> 0:14:24.840
<v Speaker 2>So initially it had been using claw to be able

0:14:24.880 --> 0:14:27.240
<v Speaker 2>to pull that together in prompting it, in defining a

0:14:27.240 --> 0:14:29.880
<v Speaker 2>well formatted set of materials that are structured with clear

0:14:29.920 --> 0:14:30.800
<v Speaker 2>recommendations for.

0:14:30.760 --> 0:14:31.800
<v Speaker 3>The week ahead.

0:14:32.320 --> 0:14:34.400
<v Speaker 2>And now we're in the process of being able to

0:14:34.400 --> 0:14:38.480
<v Speaker 2>automate that utilizing Glean with an agent, which means that

0:14:38.520 --> 0:14:40.840
<v Speaker 2>on a weekly basis, I've got all of the inputs

0:14:40.880 --> 0:14:43.840
<v Speaker 2>of the reports coming from different teams and functions, and

0:14:43.960 --> 0:14:47.080
<v Speaker 2>then producing a format and an output that I've landed

0:14:47.120 --> 0:14:51.760
<v Speaker 2>on as being the format that's relevant, that provides cohesion

0:14:51.760 --> 0:14:55.120
<v Speaker 2>across the metrics, but then more importantly, recommendations and insights

0:14:55.160 --> 0:14:56.200
<v Speaker 2>for me to act on that.

0:14:56.240 --> 0:15:00.480
<v Speaker 1>Sounds absolutely amazing and such a timesaver as well, to know,

0:15:01.320 --> 0:15:03.960
<v Speaker 1>Joe clevisely like a large part of a leader's role

0:15:04.280 --> 0:15:07.480
<v Speaker 1>is strategic thinking. And you know, I think in the

0:15:07.520 --> 0:15:11.880
<v Speaker 1>AOL conversation we hear so much around all the productivity

0:15:12.080 --> 0:15:15.360
<v Speaker 1>benefits and the time savings, but I feel like people

0:15:15.400 --> 0:15:19.080
<v Speaker 1>talk less around how it is augmenting the quality of

0:15:19.160 --> 0:15:22.320
<v Speaker 1>their thinking. Like I'd love to know, Joe, you know

0:15:22.360 --> 0:15:26.080
<v Speaker 1>some other ways that you use AI to stress test

0:15:26.120 --> 0:15:29.560
<v Speaker 1>your thinking, to bounce ideas around, you know, anything that's

0:15:29.600 --> 0:15:31.720
<v Speaker 1>improving the quality of your thinking. How are you using

0:15:31.760 --> 0:15:32.360
<v Speaker 1>it in that way?

0:15:32.560 --> 0:15:33.320
<v Speaker 3>Yeah? Pretty much.

0:15:33.320 --> 0:15:38.200
<v Speaker 2>Any input that I'm getting where my instinct to respond

0:15:38.240 --> 0:15:40.600
<v Speaker 2>to it may not be as quick and I need

0:15:40.640 --> 0:15:42.800
<v Speaker 2>some time to think about it. So it might be

0:15:42.880 --> 0:15:45.880
<v Speaker 2>an email, it might be a Slack message, it might

0:15:45.960 --> 0:15:48.320
<v Speaker 2>be you know, just the thought that's come to me

0:15:48.400 --> 0:15:52.000
<v Speaker 2>after a meeting. I'm very open and willing to just

0:15:52.600 --> 0:15:55.840
<v Speaker 2>feed that into a claud or whatever model that I'm using,

0:15:56.400 --> 0:15:59.840
<v Speaker 2>and then prompted to help me evaluate what the issue

0:15:59.880 --> 0:16:02.440
<v Speaker 2>with is, what the challenges are, how am I consider

0:16:02.440 --> 0:16:04.600
<v Speaker 2>it and give me two or three options about how

0:16:04.640 --> 0:16:07.840
<v Speaker 2>to move forward with it? So it's increasingly becoming my

0:16:08.000 --> 0:16:11.080
<v Speaker 2>system in pretty much everything that I'm doing, which is

0:16:11.080 --> 0:16:13.480
<v Speaker 2>allowing me to be far more productive and more effective

0:16:13.480 --> 0:16:15.880
<v Speaker 2>in the moment for what that issue is that's coming

0:16:15.960 --> 0:16:16.360
<v Speaker 2>my way.

0:16:17.040 --> 0:16:20.440
<v Speaker 1>Amazing, Jo, it has been so great having you on.

0:16:20.800 --> 0:16:22.520
<v Speaker 1>I feel like it's not every day you get to

0:16:22.920 --> 0:16:25.400
<v Speaker 1>speak to the head of a company just really transparently

0:16:25.440 --> 0:16:28.680
<v Speaker 1>around how they are using AI and how it's really transformed,

0:16:29.080 --> 0:16:31.600
<v Speaker 1>like really a lot of what you do in your

0:16:31.840 --> 0:16:34.520
<v Speaker 1>role as president of Elmo. So thank you so much

0:16:34.560 --> 0:16:37.480
<v Speaker 1>for sharing and for being so practical with all those

0:16:37.520 --> 0:16:38.880
<v Speaker 1>examples that you have given.

0:16:39.080 --> 0:16:40.840
<v Speaker 3>Thanks Amantha, it's been great talking with you.

0:16:41.960 --> 0:16:45.160
<v Speaker 1>If you want to explore how Elmo is building AI

0:16:45.400 --> 0:16:49.000
<v Speaker 1>into how Australian organizations manage and develop their people, head

0:16:49.000 --> 0:16:52.640
<v Speaker 1>to Elmosoftware dot com dot au and of course to

0:16:52.720 --> 0:16:55.640
<v Speaker 1>follow How I Work wherever you listen to your podcasts

0:16:55.880 --> 0:16:59.000
<v Speaker 1>so you don't miss what's coming up next. If you

0:16:59.200 --> 0:17:02.040
<v Speaker 1>like today's show, make sure you hit follow on your

0:17:02.080 --> 0:17:05.840
<v Speaker 1>podcast app to be alerted when new episodes drop. How

0:17:05.920 --> 0:17:08.280
<v Speaker 1>I Work was recorded on the traditional land of the

0:17:08.320 --> 0:17:10.400
<v Speaker 1>Warrangery people, part of the Coulan nation.