1
00:00:05,240 --> 00:00:08,000
Speaker 1: Welcome to the Business of Tech powered by two Degrees Business,

2
00:00:08,080 --> 00:00:13,360
Speaker 1: the podcast exploring the innovators shaping our digital future. Today,

3
00:00:13,400 --> 00:00:17,840
Speaker 1: we're joined by Axton Pitt, co founder of Lipmaps, a

4
00:00:17,880 --> 00:00:23,720
Speaker 1: Wellington based startup that's transforming how scientific research is discovered, visualized,

5
00:00:23,880 --> 00:00:28,120
Speaker 1: and connected. With the background in biomedical science and computer science,

6
00:00:28,560 --> 00:00:31,840
Speaker 1: Acton's journey began with a personal mission to make an

7
00:00:31,880 --> 00:00:37,080
Speaker 1: impact in health and technology. Lipmaps recently made headlines with

8
00:00:37,159 --> 00:00:41,440
Speaker 1: its acquisition of US competitor research Rabbit and a one

9
00:00:41,479 --> 00:00:45,919
Speaker 1: million dollar funding round, bringing its powerful AI driven research

10
00:00:46,000 --> 00:00:50,760
Speaker 1: platform to over two million users worldwide. In this episode,

11
00:00:50,800 --> 00:00:54,880
Speaker 1: we'll dive into how Lipmaps is using artificial intelligence to

12
00:00:54,960 --> 00:00:59,600
Speaker 1: accelerate impactful science, the challenges and opportunities of building a

13
00:00:59,640 --> 00:01:02,880
Speaker 1: global SaaS business from New Zealand, and what the future

14
00:01:02,920 --> 00:01:06,840
Speaker 1: holds for research in the age of AI. So here's

15
00:01:06,920 --> 00:01:10,560
Speaker 1: Axton Pitt on episode ninety nine of the Business of

16
00:01:10,600 --> 00:01:20,840
Speaker 1: Tech acctent Pitt. Welcome to the Business of Tech. How

17
00:01:20,840 --> 00:01:23,839
Speaker 1: are you doing good? Thanks thanks for having me. Well, congratulations,

18
00:01:23,840 --> 00:01:26,160
Speaker 1: it's been a big couple of weeks for you. You've

19
00:01:26,200 --> 00:01:31,720
Speaker 1: done a big acquisition a competitor research Rabbit that's pretty cool.

20
00:01:31,840 --> 00:01:36,000
Speaker 1: You've raised a million dollars to fund your expansion and

21
00:01:36,120 --> 00:01:39,319
Speaker 1: that acquisition will give you what two million users researchers

22
00:01:39,360 --> 00:01:44,400
Speaker 1: around the world using lit maps, So that's very impressive.

23
00:01:44,400 --> 00:01:45,920
Speaker 1: So we're going to get to all of that and

24
00:01:46,040 --> 00:01:49,480
Speaker 1: talk about how you're using artificial intelligence as part of litmaps.

25
00:01:50,200 --> 00:01:52,880
Speaker 1: Just start by telling us a bit about your own background.

26
00:01:53,240 --> 00:01:57,160
Speaker 1: From what I can tell, computer science, biomedical research, that's

27
00:01:57,240 --> 00:01:58,960
Speaker 1: really been the sweet spot for you in terms of

28
00:01:59,040 --> 00:01:59,680
Speaker 1: your studies.

29
00:02:00,120 --> 00:02:03,360
Speaker 2: Yeah, so I studied undergraduate degree in biomedical science and

30
00:02:03,400 --> 00:02:05,720
Speaker 2: that was sort of getting one of the journey of

31
00:02:05,800 --> 00:02:08,680
Speaker 2: what's possible about the human body and trying to make

32
00:02:08,680 --> 00:02:11,200
Speaker 2: an impact in some way. And then I fell into

33
00:02:11,280 --> 00:02:13,280
Speaker 2: really the computer science realm, where I was programming as

34
00:02:13,280 --> 00:02:15,160
Speaker 2: a hobby and sort of tried to combine that with

35
00:02:15,720 --> 00:02:19,920
Speaker 2: that interest that I had based on a previous health incident.

36
00:02:19,919 --> 00:02:22,320
Speaker 2: Basically say, look, I had severe asthma as a kid.

37
00:02:22,360 --> 00:02:25,160
Speaker 2: Can I do something contribute in the medical space. So

38
00:02:25,440 --> 00:02:27,679
Speaker 2: that's where that's where those two interests sort of came from.

39
00:02:27,800 --> 00:02:30,160
Speaker 1: Okay, so is that when you're at University of Auckland,

40
00:02:30,200 --> 00:02:33,040
Speaker 1: is that what your area of research particularly looked at

41
00:02:33,080 --> 00:02:35,240
Speaker 1: some sort of answer to asthma.

42
00:02:35,919 --> 00:02:38,520
Speaker 2: No, just in terms of my personal backstory at a

43
00:02:38,520 --> 00:02:41,520
Speaker 2: bachelor level and taking those interests trying to translate it

44
00:02:41,560 --> 00:02:45,120
Speaker 2: into maybe a PhD pathway. Wasn't quite getting over the

45
00:02:45,120 --> 00:02:47,519
Speaker 2: limit there to do it, and then found computer science

46
00:02:47,560 --> 00:02:49,480
Speaker 2: and was lucky enough to go to San Francisco and

47
00:02:49,480 --> 00:02:52,600
Speaker 2: attend Apple's Developers Conference, which really kicked off this sort

48
00:02:52,600 --> 00:02:54,440
Speaker 2: of software and tech realm for me.

49
00:02:54,680 --> 00:02:57,480
Speaker 1: Yeah, I've been to a few of those, the WWDC

50
00:02:57,600 --> 00:03:02,640
Speaker 1: I think they call it epic conference and big health

51
00:03:02,680 --> 00:03:04,800
Speaker 1: tech focus as well. What you can do now with

52
00:03:04,960 --> 00:03:07,320
Speaker 1: Apple devices is pretty impressive.

53
00:03:07,800 --> 00:03:10,120
Speaker 3: Yeah. For me, that was the quantified health sort the era.

54
00:03:10,240 --> 00:03:12,200
Speaker 2: You know, things like you know, tracking your steps, your

55
00:03:12,200 --> 00:03:13,400
Speaker 2: heart rate, these sorts of things.

56
00:03:13,639 --> 00:03:14,359
Speaker 3: But I sort of found it.

57
00:03:14,440 --> 00:03:16,880
Speaker 2: You know, all those metrics can be a bit overwhelming

58
00:03:16,919 --> 00:03:18,919
Speaker 2: as a human and you're sort of like, what steps

59
00:03:18,919 --> 00:03:21,520
Speaker 2: should I be taking based on my particular case. And

60
00:03:21,560 --> 00:03:23,320
Speaker 2: so that's some of why I got into it, trying

61
00:03:23,320 --> 00:03:25,960
Speaker 2: to figure out, you know, what are the limits for yourself?

62
00:03:26,280 --> 00:03:28,160
Speaker 3: Is it ten thousand steps or is actually eight thousand

63
00:03:28,200 --> 00:03:29,880
Speaker 3: depending on who you are and what you're doing.

64
00:03:30,400 --> 00:03:35,760
Speaker 1: So there's that area of research that is particularly interesting

65
00:03:36,000 --> 00:03:38,480
Speaker 1: what you can do with that sort of data. But

66
00:03:38,560 --> 00:03:41,800
Speaker 1: in terms of lit maps, give us the genesis story

67
00:03:41,840 --> 00:03:43,840
Speaker 1: of that. You were at the University of Auckland and

68
00:03:43,880 --> 00:03:47,360
Speaker 1: what you identified some sort of problem around how people

69
00:03:47,400 --> 00:03:50,800
Speaker 1: were trying to find research papers in the course of

70
00:03:50,840 --> 00:03:52,600
Speaker 1: doing their own scientific research.

71
00:03:52,760 --> 00:03:53,400
Speaker 3: Yeah, that's right.

72
00:03:53,440 --> 00:03:56,080
Speaker 2: So me and my co founder Kyle Webster, we basically

73
00:03:56,720 --> 00:03:59,600
Speaker 2: came together and said at a student group actually called Kaias,

74
00:04:00,080 --> 00:04:02,360
Speaker 2: really looked at you know, he was struggling with his

75
00:04:02,440 --> 00:04:05,280
Speaker 2: PhD literature review process. So a lot of graduate students

76
00:04:05,320 --> 00:04:08,200
Speaker 2: struggled with this early first year of their PhD. What's

77
00:04:08,240 --> 00:04:10,480
Speaker 2: already been done? And you know, can you make a

78
00:04:10,600 --> 00:04:13,160
Speaker 2: novel contribution? And so we just found the tools that

79
00:04:13,400 --> 00:04:15,480
Speaker 2: you had to use very clunky, you know, these were

80
00:04:15,520 --> 00:04:18,080
Speaker 2: sort of nineties era web tools where you could very

81
00:04:19,000 --> 00:04:22,440
Speaker 2: old visualizations. We really wanted a nice user interface and

82
00:04:22,480 --> 00:04:25,880
Speaker 2: experience for people. So that was the real first insight,

83
00:04:25,960 --> 00:04:27,599
Speaker 2: and then we had a long and winding path to

84
00:04:27,600 --> 00:04:30,200
Speaker 2: get to where we are to deliver a solution for that.

85
00:04:30,600 --> 00:04:33,400
Speaker 3: So when was lipmaps actually founded.

86
00:04:33,720 --> 00:04:36,080
Speaker 2: Yeah, so it was twenty sixteen, and there's been a

87
00:04:36,160 --> 00:04:37,960
Speaker 2: nights and Weekends project for a long time, and then

88
00:04:38,279 --> 00:04:40,479
Speaker 2: twenty twenty twenty twenty one is where we went full time.

89
00:04:40,760 --> 00:04:43,320
Speaker 1: Yeah, so actually that's coming up on almost ten years ago,

90
00:04:43,800 --> 00:04:47,240
Speaker 1: so you know, Web of Science and these other databases

91
00:04:47,279 --> 00:04:52,039
Speaker 1: and platforms researchers would have been using heavily. Then things

92
00:04:52,040 --> 00:04:56,039
Speaker 1: have evolved a lot. Artificial intelligence and generative AI have

93
00:04:56,160 --> 00:04:59,080
Speaker 1: come into the picture. But at the heart of all

94
00:04:59,360 --> 00:05:03,080
Speaker 1: lip maps is this concept of connectedness. Maybe explain that

95
00:05:03,279 --> 00:05:07,839
Speaker 1: and getting a visual representation of what research exists in

96
00:05:07,880 --> 00:05:08,279
Speaker 1: the world.

97
00:05:08,680 --> 00:05:11,919
Speaker 2: Yeah, so obviously we really want to help people understand

98
00:05:12,240 --> 00:05:14,720
Speaker 2: what's already been done so that they can contribute to

99
00:05:14,720 --> 00:05:17,039
Speaker 2: something novel. And so we're sort of coming at this

100
00:05:17,080 --> 00:05:21,080
Speaker 2: from the way papers cite each other and reference each

101
00:05:21,120 --> 00:05:23,720
Speaker 2: other essentially is really important. So if you're referencing a

102
00:05:23,720 --> 00:05:25,680
Speaker 2: certain domain or a certain topic, that sort of is

103
00:05:25,720 --> 00:05:28,960
Speaker 2: telling a lot about what you're looking into, and so

104
00:05:29,040 --> 00:05:32,440
Speaker 2: we leverage those citation patterns, and not only direct citations,

105
00:05:32,560 --> 00:05:35,719
Speaker 2: is also the sort of who's also citing those types

106
00:05:35,760 --> 00:05:37,560
Speaker 2: of things, and then so we can sort.

107
00:05:37,360 --> 00:05:39,239
Speaker 3: Of create this connectedness idea.

108
00:05:39,279 --> 00:05:41,640
Speaker 2: Of these papers are very similar because they cite the

109
00:05:41,640 --> 00:05:43,760
Speaker 2: literature in a similar way to you, and so that's

110
00:05:43,800 --> 00:05:46,919
Speaker 2: the core of our system. Plus we've shown that in

111
00:05:46,920 --> 00:05:49,960
Speaker 2: a visual way. So there's a timeline of publications with

112
00:05:50,360 --> 00:05:52,480
Speaker 2: citation coal on the vertical axis, and you can sort

113
00:05:52,480 --> 00:05:55,000
Speaker 2: of get a sense of the citation tree, if you will,

114
00:05:55,400 --> 00:05:58,440
Speaker 2: of historical papers to modern day and how they connect.

115
00:05:59,160 --> 00:06:02,480
Speaker 1: So you obviously a problem there that this was very

116
00:06:02,480 --> 00:06:04,360
Speaker 1: clunky the systems out there.

117
00:06:05,160 --> 00:06:06,760
Speaker 3: But it's almost deeper than that.

118
00:06:07,200 --> 00:06:11,640
Speaker 1: Is that the way that researchers sort of look for patterns,

119
00:06:11,640 --> 00:06:13,440
Speaker 1: and that is sort of changing as well. It's a

120
00:06:13,440 --> 00:06:17,200
Speaker 1: lot more visual and I've seen the lit maps interface,

121
00:06:17,279 --> 00:06:18,880
Speaker 1: you know, and it is very much here's a pocket

122
00:06:18,960 --> 00:06:22,200
Speaker 1: of research on a very specific topic, and here are

123
00:06:22,200 --> 00:06:25,680
Speaker 1: some more over there. So is it also you know,

124
00:06:25,800 --> 00:06:30,400
Speaker 1: researchers how they approach doing scientific research and looking for

125
00:06:30,440 --> 00:06:32,400
Speaker 1: all the papers that are going to be useful to

126
00:06:32,480 --> 00:06:35,760
Speaker 1: that journey of discovery has changed as well and evolved.

127
00:06:36,360 --> 00:06:38,480
Speaker 3: Yeah, I think you know, some things have said the same.

128
00:06:38,520 --> 00:06:40,280
Speaker 2: So you still want to be able to tell the

129
00:06:40,320 --> 00:06:43,520
Speaker 2: story of I've covered this topic well and thoroughly, and

130
00:06:43,560 --> 00:06:46,200
Speaker 2: I understand what an expert would say are the key

131
00:06:46,279 --> 00:06:47,640
Speaker 2: papers that are in your field?

132
00:06:47,920 --> 00:06:49,080
Speaker 3: So that is very constant.

133
00:06:49,279 --> 00:06:51,320
Speaker 2: But obviously the way that people are traversing this and

134
00:06:51,400 --> 00:06:53,279
Speaker 2: making this job easier has come.

135
00:06:53,160 --> 00:06:56,000
Speaker 3: About with tools like ourselves, And I think we've also helped.

136
00:06:55,760 --> 00:06:58,520
Speaker 2: Graduate students a lot because they're starting out and they

137
00:06:58,880 --> 00:07:01,280
Speaker 2: don't really have that mental pith sure of who are

138
00:07:01,279 --> 00:07:02,720
Speaker 2: the important people.

139
00:07:02,440 --> 00:07:04,000
Speaker 3: And topics in their field.

140
00:07:04,080 --> 00:07:06,280
Speaker 2: So we're essentially giving them a bootstrap to get that

141
00:07:06,640 --> 00:07:10,000
Speaker 2: mental model and maybe a professional senior research already has

142
00:07:10,000 --> 00:07:10,960
Speaker 2: that and their brain already.

143
00:07:11,200 --> 00:07:19,800
Speaker 1: Yeah, So generative AI. I've been using it for deep research. Literally,

144
00:07:19,840 --> 00:07:22,600
Speaker 1: with the likes of Perplexity, you can get clawed to

145
00:07:22,680 --> 00:07:27,760
Speaker 1: do some incredible research on your behalf to identify what

146
00:07:27,800 --> 00:07:31,160
Speaker 1: the main papers out there are and then summarize them

147
00:07:31,960 --> 00:07:35,480
Speaker 1: do a lot of the work for you. So some

148
00:07:35,520 --> 00:07:38,920
Speaker 1: of your users going through a similar sort of process here,

149
00:07:38,920 --> 00:07:43,720
Speaker 1: they're sort of short circuiting that research process by making

150
00:07:43,760 --> 00:07:46,280
Speaker 1: it a lot easier just to find exactly what's out

151
00:07:46,320 --> 00:07:47,160
Speaker 1: there they can draw on.

152
00:07:47,840 --> 00:07:51,920
Speaker 2: Yeah, so definitely our tool helps to get from zero

153
00:07:51,960 --> 00:07:55,360
Speaker 2: to one very easily. I think we're not obviously doing

154
00:07:55,360 --> 00:07:57,160
Speaker 2: the full literature review for you. So some of those

155
00:07:57,160 --> 00:07:59,880
Speaker 2: other tools like deep research from open AI is sort

156
00:07:59,880 --> 00:08:02,320
Speaker 2: of you put in a few queries and get this

157
00:08:02,560 --> 00:08:04,960
Speaker 2: report that's sort of maybe at a PhD level back

158
00:08:05,000 --> 00:08:06,680
Speaker 2: to you, but I think we're sort of in that

159
00:08:07,120 --> 00:08:10,000
Speaker 2: helping to educate the graduate student to get to the

160
00:08:10,000 --> 00:08:13,080
Speaker 2: point where they could produce that report. So there's an

161
00:08:13,120 --> 00:08:16,360
Speaker 2: understanding that maybe the citations that are in there, you

162
00:08:16,400 --> 00:08:18,880
Speaker 2: want to check them out and understand what the quality

163
00:08:18,960 --> 00:08:19,560
Speaker 2: level that is.

164
00:08:19,600 --> 00:08:21,080
Speaker 3: And the AI tools are very good these.

165
00:08:21,040 --> 00:08:24,640
Speaker 2: Days, and I think people have to reassess maybe monthly

166
00:08:24,680 --> 00:08:26,880
Speaker 2: how good these tools are because they're evolving very quickly.

167
00:08:27,320 --> 00:08:30,160
Speaker 2: But I think there's a combination of there's a generative

168
00:08:30,680 --> 00:08:33,040
Speaker 2: AI tool that can do the work or some part

169
00:08:33,080 --> 00:08:34,439
Speaker 2: of the work, but also we want to be able

170
00:08:34,440 --> 00:08:37,480
Speaker 2: to train people to produce that. So we see ourselves

171
00:08:37,480 --> 00:08:41,400
Speaker 2: as merging both areas where we provide these diagrams to

172
00:08:41,480 --> 00:08:45,280
Speaker 2: help people understand the literature landscape easily, but also we

173
00:08:45,320 --> 00:08:47,760
Speaker 2: feed into these sorts of tools with a open.

174
00:08:47,520 --> 00:08:50,360
Speaker 3: A plugin or maybe AI eventually uses a tool.

175
00:08:50,240 --> 00:08:53,559
Speaker 2: Like ours to navigate the citation landscape.

176
00:08:53,760 --> 00:08:56,679
Speaker 1: Right, So in terms of your approach to AI, you're

177
00:08:56,720 --> 00:08:59,520
Speaker 1: literally drawing on some of those big large language models

178
00:08:59,679 --> 00:09:02,040
Speaker 1: likes of open AI and claud.

179
00:09:02,440 --> 00:09:05,000
Speaker 2: Yeah, so that we were this interesting hybrid of there's

180
00:09:05,000 --> 00:09:09,199
Speaker 2: the metadata and metrics that are very tried and true,

181
00:09:09,240 --> 00:09:12,400
Speaker 2: you know, how papers are evaluated, and then there's a

182
00:09:12,559 --> 00:09:15,280
Speaker 2: new generative I tools that'd say, you know, maybe they

183
00:09:15,280 --> 00:09:18,640
Speaker 2: can represent what a paper means and some abstract sense.

184
00:09:18,920 --> 00:09:21,120
Speaker 2: And so we're using those two different approaches to really

185
00:09:21,559 --> 00:09:25,000
Speaker 2: hopefully provide the best in class search. And that means

186
00:09:25,360 --> 00:09:27,079
Speaker 2: not only you're getting good results, but you can see

187
00:09:27,080 --> 00:09:29,280
Speaker 2: how they connect to what you know, and that's really

188
00:09:29,400 --> 00:09:31,480
Speaker 2: where our map visual diagram helps a lot.

189
00:09:31,880 --> 00:09:35,480
Speaker 1: And I guess that allows a researcher to to use

190
00:09:35,480 --> 00:09:39,880
Speaker 1: more sort of conversational AI type searches and prompts like

191
00:09:39,920 --> 00:09:43,160
Speaker 1: we do every day with chatchipt or perplexity, where you

192
00:09:43,160 --> 00:09:45,800
Speaker 1: can literally say, hey, you know, we're all the best

193
00:09:45,800 --> 00:09:49,599
Speaker 1: papers on breast cancer, for instance, and it will it

194
00:09:49,640 --> 00:09:51,040
Speaker 1: will go out there and look for them and then

195
00:09:51,080 --> 00:09:52,640
Speaker 1: represent them visually for you.

196
00:09:53,040 --> 00:09:54,040
Speaker 3: Yeah. I think that's a good point.

197
00:09:54,080 --> 00:09:56,520
Speaker 2: So there's a you know, a way that we can

198
00:09:57,040 --> 00:09:58,880
Speaker 2: tie you know, maybe you want to say, what are

199
00:09:58,920 --> 00:10:00,960
Speaker 2: the key themes in these papers that I be recommended

200
00:10:00,960 --> 00:10:02,840
Speaker 2: and the generati way I tool will help pull that

201
00:10:02,880 --> 00:10:05,080
Speaker 2: out for you. An example of this as well is

202
00:10:05,120 --> 00:10:07,400
Speaker 2: just the time speed up. So We've got all the

203
00:10:07,480 --> 00:10:10,880
Speaker 2: pedic surgeon in Spain and he looks at, you know,

204
00:10:10,920 --> 00:10:13,040
Speaker 2: one of the latest techniques to heal this broken bone,

205
00:10:13,120 --> 00:10:16,080
Speaker 2: or the surgery technique, and so sometimes on the day

206
00:10:16,080 --> 00:10:18,080
Speaker 2: of surgery he's looking that up before he goes into

207
00:10:18,080 --> 00:10:20,320
Speaker 2: the theater because it's a lot easier to run that

208
00:10:20,320 --> 00:10:21,760
Speaker 2: search than it used to be.

209
00:10:21,960 --> 00:10:23,640
Speaker 3: So that's some of the impact we're having.

210
00:10:23,960 --> 00:10:27,800
Speaker 1: So not only at the early stage research where someone's

211
00:10:27,920 --> 00:10:30,760
Speaker 1: formulating a hypothesis is going to set out on three

212
00:10:30,840 --> 00:10:34,640
Speaker 1: years and more of research to come up with new knowledge.

213
00:10:34,640 --> 00:10:38,560
Speaker 1: You're actually getting surgeons, people on the front line who

214
00:10:38,559 --> 00:10:41,320
Speaker 1: are using this as a learning tool before they potentially

215
00:10:41,360 --> 00:10:42,160
Speaker 1: go into surgery.

216
00:10:42,480 --> 00:10:44,600
Speaker 2: Yeah, and you know, that's a really great impact, and

217
00:10:44,640 --> 00:10:47,760
Speaker 2: that's one of our mission values is to accelerate impactful science.

218
00:10:47,800 --> 00:10:50,040
Speaker 2: We want to see you know, those outcomes, and that's

219
00:10:50,120 --> 00:10:52,520
Speaker 2: kind of what gets us out of bed most days.

220
00:10:52,760 --> 00:10:54,640
Speaker 2: But then there's also the other there's tons of other

221
00:10:54,720 --> 00:10:58,280
Speaker 2: use cases. We see people, you know, making investment decisions,

222
00:10:58,320 --> 00:11:01,560
Speaker 2: so they're looking at evaluating pharmaceuticals and saying, you know,

223
00:11:01,800 --> 00:11:04,280
Speaker 2: what colinical trials have been run and what's out there.

224
00:11:04,360 --> 00:11:07,520
Speaker 2: So we're sort of having impacts across many different demands.

225
00:11:08,240 --> 00:11:12,040
Speaker 1: Yeah, so I guess part of the scientific process is

226
00:11:12,120 --> 00:11:16,040
Speaker 1: knowing what's there, what people have done, and it's all

227
00:11:16,080 --> 00:11:20,640
Speaker 1: about building on the shoulders of giants, the people who

228
00:11:20,679 --> 00:11:22,840
Speaker 1: came before and did some of those seminal papers that

229
00:11:22,880 --> 00:11:25,760
Speaker 1: are highly cited. But it's also really about the gaps

230
00:11:25,760 --> 00:11:28,840
Speaker 1: as well. What's the novel bit where I can really

231
00:11:28,960 --> 00:11:32,000
Speaker 1: add value and create unique knowledge?

232
00:11:32,040 --> 00:11:36,199
Speaker 2: So is it good for identifying those gaps? Yeah, that's

233
00:11:36,240 --> 00:11:39,640
Speaker 2: a really hard topic. Obviously, we're not saying we solve

234
00:11:39,720 --> 00:11:41,440
Speaker 2: that problem with research gaps. I think we help a

235
00:11:41,440 --> 00:11:43,679
Speaker 2: lot to get to that point where you can at

236
00:11:43,720 --> 00:11:46,360
Speaker 2: least know what other papers are out there that on

237
00:11:46,440 --> 00:11:48,680
Speaker 2: topic for you. But obviously, yeah, there's a lot of reading,

238
00:11:48,720 --> 00:11:50,760
Speaker 2: a lot of thinking, and it is really hard to

239
00:11:50,840 --> 00:11:55,120
Speaker 2: identify and contribute to a gap. Sometimes there's so much work, Alrea,

240
00:11:55,160 --> 00:11:56,560
Speaker 2: you've been done in the field that it's hard to

241
00:11:56,600 --> 00:12:00,319
Speaker 2: find a novel PC to contribute to. So yes, definitely

242
00:12:00,360 --> 00:12:02,760
Speaker 2: help and have people tell us that we've helped with that,

243
00:12:02,880 --> 00:12:05,120
Speaker 2: but I know it's a really difficult thing to work on.

244
00:12:05,320 --> 00:12:08,959
Speaker 1: Yeah, and you've got two million users now, some of

245
00:12:09,000 --> 00:12:12,400
Speaker 1: them at Harvard, Stanford, Cambridge, all the big universities, you've

246
00:12:12,440 --> 00:12:17,280
Speaker 1: got researchers using them. At what point did it really

247
00:12:17,280 --> 00:12:19,760
Speaker 1: start to take off where you started to get hundreds

248
00:12:19,760 --> 00:12:22,760
Speaker 1: of thousands of people actually using lip maps And what

249
00:12:23,320 --> 00:12:27,360
Speaker 1: was the spark that really triggered that early success.

250
00:12:27,840 --> 00:12:30,440
Speaker 2: Yeah, so a lot of that came from feedback iteration.

251
00:12:30,600 --> 00:12:33,439
Speaker 2: So we shared the tool with our friends at Faculty

252
00:12:33,480 --> 00:12:36,240
Speaker 2: of Medical Health Science at University Walkan for example, got

253
00:12:36,559 --> 00:12:39,040
Speaker 2: mixed feedback and considerate from there. But then you know

254
00:12:39,440 --> 00:12:44,120
Speaker 2: online forums where people post preprints, for example, so archive

255
00:12:44,160 --> 00:12:46,839
Speaker 2: dot org we have a lot of the leading AI

256
00:12:46,960 --> 00:12:50,520
Speaker 2: papers or physics papers are published, so there's an integration

257
00:12:50,600 --> 00:12:53,080
Speaker 2: there that's really helped spread the word about lipmaps.

258
00:12:53,080 --> 00:12:54,520
Speaker 3: So it was sort of on a paper.

259
00:12:54,559 --> 00:12:56,520
Speaker 2: You can create a lipmap from this paper, and that's

260
00:12:56,840 --> 00:12:59,400
Speaker 2: really helped with growth. And then obviously I think just

261
00:13:00,280 --> 00:13:03,200
Speaker 2: having universities recommend us. So you know there's a website

262
00:13:03,360 --> 00:13:06,760
Speaker 2: on most libraries that have you know, how to search literature,

263
00:13:06,760 --> 00:13:09,400
Speaker 2: and there's a section about litmaps often which has really helped.

264
00:13:09,679 --> 00:13:14,160
Speaker 1: Yeah, And as I find with AI on a regular basis,

265
00:13:14,160 --> 00:13:18,640
Speaker 1: it hallucinates and it gives you weird results, I guess

266
00:13:18,640 --> 00:13:24,480
Speaker 1: because this is based on it's drawing on verified information sources,

267
00:13:24,520 --> 00:13:28,240
Speaker 1: including pre prints some of these databases where it hasn't

268
00:13:28,280 --> 00:13:31,200
Speaker 1: been through the full peer of view process, but you

269
00:13:31,240 --> 00:13:33,720
Speaker 1: can actually see the entire paper before it goes into

270
00:13:33,720 --> 00:13:37,480
Speaker 1: a big scientific journal. So I guess you're sort of

271
00:13:37,520 --> 00:13:41,640
Speaker 1: eliminating or listening the chance of that through drawing on

272
00:13:42,120 --> 00:13:45,600
Speaker 1: all of these established databases of verified information.

273
00:13:45,880 --> 00:13:47,440
Speaker 3: Yeah, that's been an active approach from us.

274
00:13:47,440 --> 00:13:51,280
Speaker 2: So we taken metadata from papers and that's what we

275
00:13:51,360 --> 00:13:53,720
Speaker 2: really recommend from. So there's not really risk of us

276
00:13:53,720 --> 00:13:56,760
Speaker 2: saying we're making up a citation, and we do validate

277
00:13:56,800 --> 00:13:58,800
Speaker 2: results where we work with large language models or the

278
00:13:58,800 --> 00:14:01,760
Speaker 2: way we work as well as really important to make

279
00:14:01,800 --> 00:14:04,480
Speaker 2: sure that things are not just made up. There's actually

280
00:14:04,480 --> 00:14:07,000
Speaker 2: a tie back to how we've recommended it, and you

281
00:14:07,000 --> 00:14:10,439
Speaker 2: can see what that means. Plus you can also filter

282
00:14:10,480 --> 00:14:12,760
Speaker 2: out things like preprints if you're interested in just the

283
00:14:12,840 --> 00:14:14,400
Speaker 2: published peer review papers as well.

284
00:14:14,520 --> 00:14:15,439
Speaker 3: Yeah.

285
00:14:15,640 --> 00:14:20,480
Speaker 1: Research Rabbit cool name that like task rabbit, but for research?

286
00:14:21,600 --> 00:14:24,560
Speaker 1: Were they on your radar and how did this acquisition

287
00:14:24,640 --> 00:14:25,280
Speaker 1: come about?

288
00:14:25,840 --> 00:14:28,760
Speaker 2: Yeah, so we obviously have known them in the industry.

289
00:14:28,800 --> 00:14:31,480
Speaker 2: I guess it's a small world, and you know, of

290
00:14:31,480 --> 00:14:33,720
Speaker 2: people in their network. I think an interesting opportunity came

291
00:14:33,800 --> 00:14:36,160
Speaker 2: up when they were looking to sell, and so we

292
00:14:36,280 --> 00:14:38,320
Speaker 2: basically said, you know, this is a great way for

293
00:14:38,640 --> 00:14:42,360
Speaker 2: two very similar tools to come together and merge and

294
00:14:42,440 --> 00:14:44,880
Speaker 2: serve very common user base. So it's basically a story

295
00:14:44,880 --> 00:14:48,760
Speaker 2: of two independently successful tools coming together and serving a

296
00:14:48,840 --> 00:14:49,600
Speaker 2: very similar market.

297
00:14:49,640 --> 00:14:52,160
Speaker 3: So we're quite excited about providing a really good.

298
00:14:52,000 --> 00:14:55,520
Speaker 2: Service for both both user bases and over time creating

299
00:14:55,520 --> 00:14:58,080
Speaker 2: a tool that serves both people, both groups of people.

300
00:14:58,520 --> 00:15:02,360
Speaker 1: So will you carry on research rabbits platform and do

301
00:15:02,400 --> 00:15:03,720
Speaker 1: you plan to merge them at some point?

302
00:15:04,080 --> 00:15:05,640
Speaker 3: Yeah, so we're still working through those plans.

303
00:15:05,800 --> 00:15:07,640
Speaker 2: Essentially, what we ended up want to end up with

304
00:15:07,760 --> 00:15:10,680
Speaker 2: is a tool that serves both user groups really well.

305
00:15:11,080 --> 00:15:15,360
Speaker 2: And that's the graduate student today. So whether that's a

306
00:15:15,440 --> 00:15:19,360
Speaker 2: research a product or that's a mass product or some

307
00:15:19,400 --> 00:15:21,960
Speaker 2: sort of merged brand that we'll see, but the end

308
00:15:21,960 --> 00:15:24,880
Speaker 2: of the day, it will be the same value offering

309
00:15:24,920 --> 00:15:26,280
Speaker 2: that is on the market today.

310
00:15:26,400 --> 00:15:32,040
Speaker 1: Great, and a million dollars obviously that's a decent amount

311
00:15:32,040 --> 00:15:34,720
Speaker 1: of money. Is that sort of I guess seed funding

312
00:15:34,760 --> 00:15:36,200
Speaker 1: you'd described that as.

313
00:15:36,320 --> 00:15:39,320
Speaker 3: Yeah, So we've had various levels of funding.

314
00:15:39,320 --> 00:15:41,280
Speaker 2: I think you know, classing these rounds, it's yeah, it's

315
00:15:41,280 --> 00:15:44,080
Speaker 2: basically another seed round for us to get us to

316
00:15:44,160 --> 00:15:46,160
Speaker 2: the next level of growth and get into a Series

317
00:15:46,200 --> 00:15:46,880
Speaker 2: A or something like this.

318
00:15:46,960 --> 00:15:51,360
Speaker 1: Yeah, and obviously the real opportunity is the international research community.

319
00:15:51,360 --> 00:15:54,360
Speaker 1: I guess that's when ninety nine percent of your research

320
00:15:54,400 --> 00:15:56,240
Speaker 1: bases do you have local users?

321
00:15:56,600 --> 00:15:59,160
Speaker 2: Yeah, so about eight percent of our paid user bases

322
00:15:59,160 --> 00:16:02,480
Speaker 2: and New Zealand, so small contingent, but it's good to have.

323
00:16:02,640 --> 00:16:05,480
Speaker 2: So obviously we're now based in Victoria University of Wellington

324
00:16:05,480 --> 00:16:09,240
Speaker 2: and being around that is awesome to have hands on you.

325
00:16:09,360 --> 00:16:11,000
Speaker 3: Face to face time with users.

326
00:16:11,160 --> 00:16:12,800
Speaker 2: But yeah, we obviously it's spending a lot of that

327
00:16:13,280 --> 00:16:16,960
Speaker 2: seed capital on growth and expanding and making sure we're

328
00:16:17,120 --> 00:16:20,040
Speaker 2: at the right conferences and in front of the right people.

329
00:16:20,600 --> 00:16:24,400
Speaker 3: So actually take on artificial intelligence and.

330
00:16:24,320 --> 00:16:29,560
Speaker 1: Its impact, particularly since we've had generative AI come along

331
00:16:30,080 --> 00:16:35,160
Speaker 1: on how research has done. Obviously, there are applications for

332
00:16:35,280 --> 00:16:41,280
Speaker 1: the science itself. Volpara Willingtson company using AI and other

333
00:16:41,360 --> 00:16:45,480
Speaker 1: tools to look at mammograms to try and identify, for instance,

334
00:16:45,520 --> 00:16:50,880
Speaker 1: tumors or strange looking growths more effectively than a human

335
00:16:50,920 --> 00:16:54,200
Speaker 1: can do. That manually that's been a very successful business.

336
00:16:54,400 --> 00:16:58,000
Speaker 1: There are lots of other applications of AI, material science

337
00:16:58,040 --> 00:17:02,760
Speaker 1: and the like, but in terms of how researchers actually

338
00:17:02,840 --> 00:17:06,639
Speaker 1: set out on that path and organize their thoughts and

339
00:17:06,760 --> 00:17:10,440
Speaker 1: come up with the papers that are going to inform

340
00:17:10,600 --> 00:17:12,280
Speaker 1: what they're potentially going to spend a lot of money

341
00:17:12,280 --> 00:17:14,679
Speaker 1: on doing lab trials and the like.

342
00:17:15,280 --> 00:17:17,800
Speaker 3: How is it changing, how's it evolving, and where is

343
00:17:17,880 --> 00:17:19,800
Speaker 3: AI really going to take that process in the next

344
00:17:19,840 --> 00:17:22,640
Speaker 3: few years. Yeah, I think there's a big spectrum impact.

345
00:17:22,640 --> 00:17:24,879
Speaker 2: So I think on the one hand, there's you know,

346
00:17:25,160 --> 00:17:28,160
Speaker 2: things like alpha fold three that have really helped move

347
00:17:28,200 --> 00:17:32,199
Speaker 2: a particular domain forward because it can understand how we

348
00:17:32,280 --> 00:17:35,200
Speaker 2: go from sequence to structure a lot better than any

349
00:17:35,240 --> 00:17:36,120
Speaker 2: other previous.

350
00:17:35,760 --> 00:17:36,320
Speaker 3: Approach to that.

351
00:17:36,359 --> 00:17:41,960
Speaker 1: That's like protein folding, Yeah, things hardcore basic level science.

352
00:17:41,800 --> 00:17:44,239
Speaker 2: Yeah, which is a very particular domain and they've had

353
00:17:44,280 --> 00:17:46,879
Speaker 2: really great benchmarks, so you know how well the tool's doing,

354
00:17:46,920 --> 00:17:48,720
Speaker 2: which I think is really important because otherwise it's hard

355
00:17:48,720 --> 00:17:49,960
Speaker 2: to evaluate is.

356
00:17:49,880 --> 00:17:51,400
Speaker 3: It solving things in a way.

357
00:17:51,880 --> 00:17:55,359
Speaker 2: But there's other approaches where you know, maybe if AI

358
00:17:55,520 --> 00:17:58,000
Speaker 2: solves all problems for you, the people who are in

359
00:17:58,040 --> 00:18:00,760
Speaker 2: training don't really have to push them and undergo their

360
00:18:00,840 --> 00:18:04,280
Speaker 2: education process where they test their ideas and battle tests

361
00:18:04,320 --> 00:18:06,800
Speaker 2: and learn from their supervisor and get that sort of

362
00:18:06,800 --> 00:18:10,400
Speaker 2: critical thinking element that might be taken away if AI

363
00:18:10,480 --> 00:18:13,240
Speaker 2: does all their job for them. So there's this interesting

364
00:18:13,240 --> 00:18:18,000
Speaker 2: balance of there's really great impact and tools that are

365
00:18:18,000 --> 00:18:20,520
Speaker 2: available now, but there's also that risk and that balance

366
00:18:20,560 --> 00:18:23,679
Speaker 2: of does this tool produce results that I can trust?

367
00:18:24,080 --> 00:18:27,240
Speaker 2: And also is it sort of shortcutting the process that

368
00:18:27,320 --> 00:18:30,879
Speaker 2: has maybe required to get that critical thinking and important

369
00:18:31,080 --> 00:18:32,000
Speaker 2: development happening.

370
00:18:32,440 --> 00:18:35,480
Speaker 1: It's a really good point because the whole idea of

371
00:18:36,560 --> 00:18:39,840
Speaker 1: being a good scientist a good researcher is that curiosity

372
00:18:40,880 --> 00:18:44,719
Speaker 1: and that ability to go out and discover stuff for

373
00:18:44,760 --> 00:18:47,320
Speaker 1: yourself and not all just to be handed to you

374
00:18:47,760 --> 00:18:48,200
Speaker 1: on a plate.

375
00:18:48,240 --> 00:18:51,359
Speaker 3: So how do you deal with that in the MAHAPS context.

376
00:18:51,800 --> 00:18:53,920
Speaker 2: Yeah, and I think for us, you know, we really

377
00:18:53,960 --> 00:18:56,879
Speaker 2: feel it's human plus AI. So empowering people with the

378
00:18:56,960 --> 00:18:59,480
Speaker 2: right tools is really our approach. And obviously you could

379
00:18:59,520 --> 00:19:01,320
Speaker 2: argue some of these other tools al really do that,

380
00:19:01,359 --> 00:19:03,719
Speaker 2: but I think our particular approach is really you know,

381
00:19:03,800 --> 00:19:06,560
Speaker 2: whether it's explaining results, so the diagram to help you

382
00:19:06,600 --> 00:19:09,159
Speaker 2: see our things put together or just our approach to

383
00:19:09,640 --> 00:19:12,160
Speaker 2: you know, making sure that we don't have open ended

384
00:19:12,240 --> 00:19:14,880
Speaker 2: sort of AI chat. It's something where we can validate

385
00:19:14,920 --> 00:19:17,639
Speaker 2: results and show you know, this is time back to

386
00:19:17,720 --> 00:19:19,840
Speaker 2: literature in this way that's really important for us. So

387
00:19:19,880 --> 00:19:21,920
Speaker 2: that I guess it's the way we're going about it

388
00:19:22,040 --> 00:19:24,359
Speaker 2: to empower people. And as you mentioned, you know, if

389
00:19:24,400 --> 00:19:27,119
Speaker 2: there's R and D teams at Volpara or other companies

390
00:19:27,720 --> 00:19:30,160
Speaker 2: using these types of tools in their processes, that's really

391
00:19:30,160 --> 00:19:32,840
Speaker 2: where we want to help have an impact as well

392
00:19:32,880 --> 00:19:34,119
Speaker 2: beyond academia.

393
00:19:34,359 --> 00:19:36,520
Speaker 1: Yeah, and it's just got to be huge efficiency across

394
00:19:36,560 --> 00:19:42,360
Speaker 1: the board. Everything from writing research proposals and funding applications

395
00:19:42,400 --> 00:19:43,080
Speaker 1: that takes up.

396
00:19:42,960 --> 00:19:44,000
Speaker 3: A huge amount of time.

397
00:19:44,040 --> 00:19:46,760
Speaker 1: If you can use AI and that process through to

398
00:19:46,800 --> 00:19:49,040
Speaker 1: writing up the results, you know, it's a months long,

399
00:19:49,800 --> 00:19:53,600
Speaker 1: years long process. Sometimes a few submittings something to nature

400
00:19:53,680 --> 00:19:54,119
Speaker 1: or science.

401
00:19:54,160 --> 00:19:58,200
Speaker 3: You have big journals. There's a lot of work required here.

402
00:19:58,240 --> 00:20:00,560
Speaker 1: So I guess you know, when the politician say to us,

403
00:20:00,720 --> 00:20:04,679
Speaker 1: we're funding AI because fundamentally it's going to help on

404
00:20:04,760 --> 00:20:10,680
Speaker 1: things like healthcare, sustainability and the climate. They're looking at

405
00:20:10,720 --> 00:20:14,680
Speaker 1: all of those efficiencies that are gained that shorten the

406
00:20:14,720 --> 00:20:16,320
Speaker 1: amount of time it's going to take.

407
00:20:16,160 --> 00:20:18,920
Speaker 3: For a research team to get something to fruition.

408
00:20:19,840 --> 00:20:21,880
Speaker 2: Yeah, I think there's definitely and you can't deny there's

409
00:20:21,880 --> 00:20:24,640
Speaker 2: an impact there of saving time in lots of ways,

410
00:20:24,640 --> 00:20:26,520
Speaker 2: and we think we are contributing as well.

411
00:20:27,040 --> 00:20:27,200
Speaker 3: Yeah.

412
00:20:27,240 --> 00:20:29,760
Speaker 2: I guess it's balancing that with you know, making sure

413
00:20:29,800 --> 00:20:32,560
Speaker 2: that you are still investing time and training up the

414
00:20:32,560 --> 00:20:37,359
Speaker 2: next generation and having enough funding to have the programs

415
00:20:37,680 --> 00:20:38,040
Speaker 2: be run.

416
00:20:38,560 --> 00:20:42,399
Speaker 1: How well received is AI buy, you know, the Stanford

417
00:20:42,480 --> 00:20:46,160
Speaker 1: and Harvard and MIT scientists, I'm sure by that AI

418
00:20:46,240 --> 00:20:48,040
Speaker 1: scientists themselves they love it.

419
00:20:48,080 --> 00:20:48,920
Speaker 3: But is there a little bit.

420
00:20:48,880 --> 00:20:52,119
Speaker 1: Of pushback in skepticism from scientists around AI?

421
00:20:52,480 --> 00:20:55,159
Speaker 3: Yeah. I think again there's different levels.

422
00:20:55,160 --> 00:20:58,080
Speaker 2: So people in the teaching realm, I think there's maybe

423
00:20:58,119 --> 00:21:01,120
Speaker 2: an epidemic of cheating, and how do you handle every

424
00:21:01,119 --> 00:21:03,680
Speaker 2: assignment being submitted and generated.

425
00:21:03,280 --> 00:21:04,920
Speaker 3: By CHATGBT maybe, and so.

426
00:21:05,000 --> 00:21:07,720
Speaker 2: Evaluating and dealing with that is maybe a really difficult

427
00:21:07,720 --> 00:21:08,560
Speaker 2: battle that's going on.

428
00:21:08,920 --> 00:21:11,360
Speaker 3: And then there's the you know, publish or perish.

429
00:21:11,000 --> 00:21:13,760
Speaker 2: Type approach where you know you're trying to produce these

430
00:21:13,800 --> 00:21:15,919
Speaker 2: results and get in really high quality journals. And so

431
00:21:15,960 --> 00:21:20,679
Speaker 2: maybe the AI helping write has increased the temperature of

432
00:21:20,960 --> 00:21:26,160
Speaker 2: competition and pace of publishing there. So I think it's

433
00:21:26,560 --> 00:21:29,760
Speaker 2: really unlocking some new problems. And you know, as I

434
00:21:29,800 --> 00:21:32,240
Speaker 2: mentioned in protein folding and other domains, it's really allowing

435
00:21:32,240 --> 00:21:35,080
Speaker 2: people to go and solve problems they wanted to solve before.

436
00:21:35,640 --> 00:21:38,680
Speaker 2: This was a roadblock in terms of solving a sequence

437
00:21:38,960 --> 00:21:42,720
Speaker 2: to structure. So I think it's unlocking really great opportunities

438
00:21:42,760 --> 00:21:46,480
Speaker 2: but also causing headaches and other areas that yeah, maybe

439
00:21:46,480 --> 00:21:49,719
Speaker 2: we should assess students differently and avoid some of these issues.

440
00:21:49,720 --> 00:21:52,560
Speaker 2: But yeah, it's definitely causing headaches I think as well, and.

441
00:21:52,520 --> 00:21:55,400
Speaker 1: I guess forget the next phase and probably some researchers

442
00:21:55,400 --> 00:21:58,240
Speaker 1: are already doing it to some extent. Is you go

443
00:21:58,320 --> 00:22:01,119
Speaker 1: to lip maps, you get to identify, we're all the

444
00:22:01,240 --> 00:22:04,840
Speaker 1: really good research papers, the ones you hadn't necessarily thought off.

445
00:22:05,200 --> 00:22:07,960
Speaker 1: You download all of those, then you put them into

446
00:22:08,280 --> 00:22:12,159
Speaker 1: a large language model and create a literature review or

447
00:22:12,560 --> 00:22:16,159
Speaker 1: at least a first cut off a paper, and in

448
00:22:16,240 --> 00:22:18,399
Speaker 1: stat refining it from there. So that's another way that

449
00:22:18,440 --> 00:22:21,720
Speaker 1: it could speed up massively the process of doing science.

450
00:22:22,240 --> 00:22:24,600
Speaker 3: Yeah, so there's definitely that knowledge work.

451
00:22:24,880 --> 00:22:27,720
Speaker 2: We want people to be probably running those experiments rather

452
00:22:27,760 --> 00:22:31,440
Speaker 2: than hopefully not wasting time with all the sort of

453
00:22:31,520 --> 00:22:33,560
Speaker 2: admin and behind the scenes stuff. So yeah, I definitely

454
00:22:33,560 --> 00:22:36,320
Speaker 2: think if we can shortcut some of that then that's

455
00:22:36,400 --> 00:22:38,840
Speaker 2: going to help with pushing science forward.

456
00:22:39,320 --> 00:22:40,960
Speaker 1: So what sort of operation do you have we're here

457
00:22:40,960 --> 00:22:43,640
Speaker 1: in Wellington at Victoria University.

458
00:22:44,320 --> 00:22:46,199
Speaker 3: Do you have a team of coders here? Have you

459
00:22:46,280 --> 00:22:48,040
Speaker 3: done that in house? Yeah, that's right.

460
00:22:48,080 --> 00:22:50,280
Speaker 2: So we've taken a really full stack approach, so we've

461
00:22:50,280 --> 00:22:54,200
Speaker 2: done everything from the software development piece, the design piece,

462
00:22:54,359 --> 00:22:56,560
Speaker 2: and even the marketing and getting the word out.

463
00:22:56,600 --> 00:22:58,520
Speaker 3: So we're very full stack in that way.

464
00:22:58,720 --> 00:23:00,679
Speaker 2: And so yeah, we're great to be based in the

465
00:23:00,800 --> 00:23:04,320
Speaker 2: university because we can also just you know, get students

466
00:23:04,320 --> 00:23:05,600
Speaker 2: and pay them to do user testing.

467
00:23:05,640 --> 00:23:06,880
Speaker 3: It's really awesome to.

468
00:23:06,840 --> 00:23:10,880
Speaker 2: Have that direct access to who we are serving. So yeah,

469
00:23:10,960 --> 00:23:13,359
Speaker 2: really proud to build the product here and export it.

470
00:23:13,400 --> 00:23:16,040
Speaker 2: I think that's something also proud of is you know,

471
00:23:16,320 --> 00:23:20,639
Speaker 2: we're not just doing some of this and exploring some

472
00:23:20,680 --> 00:23:21,160
Speaker 2: of the jobs.

473
00:23:21,200 --> 00:23:22,560
Speaker 3: Maybe I'm also proud of.

474
00:23:22,480 --> 00:23:25,280
Speaker 1: That, and you're probably like a number of New Zealand

475
00:23:25,280 --> 00:23:31,240
Speaker 1: software startups that had a really interesting novel, Idea got

476
00:23:31,280 --> 00:23:34,920
Speaker 1: a bit of traction internationally. Then the AI wave came

477
00:23:35,040 --> 00:23:36,399
Speaker 1: and then you had to figure out how do we

478
00:23:36,440 --> 00:23:40,600
Speaker 1: integrate this into our core platform. What's that been like

479
00:23:40,680 --> 00:23:43,600
Speaker 1: getting your head around these large language models, how the

480
00:23:43,720 --> 00:23:47,040
Speaker 1: interface between your business and the likes to open AI

481
00:23:48,400 --> 00:23:53,040
Speaker 1: and these other companies works doing stuff in the cloud,

482
00:23:53,080 --> 00:23:54,479
Speaker 1: you know, where at least you're not, you know, not

483
00:23:54,880 --> 00:23:56,879
Speaker 1: having to build a large language model, which would be

484
00:23:57,040 --> 00:24:01,840
Speaker 1: vastly expensive, but there are costs, added costs and licensing costs.

485
00:24:01,840 --> 00:24:03,919
Speaker 1: What's it been like getting your head around that and

486
00:24:04,000 --> 00:24:08,119
Speaker 1: even having to at a fundamental code and develop for

487
00:24:08,359 --> 00:24:10,600
Speaker 1: an AI centric product.

488
00:24:10,800 --> 00:24:12,840
Speaker 2: Yeah, I think a lot of that's come from even

489
00:24:12,880 --> 00:24:15,399
Speaker 2: though we were a bit before AI was mainstream. I

490
00:24:15,400 --> 00:24:18,240
Speaker 2: think there was GPT two when we're starting and some

491
00:24:18,320 --> 00:24:20,240
Speaker 2: of the early versions, so we're really aware of what

492
00:24:20,400 --> 00:24:20,840
Speaker 2: was going on.

493
00:24:20,880 --> 00:24:22,000
Speaker 3: So I guess because.

494
00:24:21,720 --> 00:24:24,439
Speaker 2: We're the full stack and do some of the deeper

495
00:24:24,480 --> 00:24:26,879
Speaker 2: tech work, it means that we can really integrate with

496
00:24:26,920 --> 00:24:29,280
Speaker 2: these other tools at a very low level and take

497
00:24:29,280 --> 00:24:32,200
Speaker 2: advantage of them from running our own models and having

498
00:24:32,240 --> 00:24:34,879
Speaker 2: GPUs that do inference and not the training part, but

499
00:24:34,920 --> 00:24:38,800
Speaker 2: actually run it at time of getting data out of it.

500
00:24:39,200 --> 00:24:41,119
Speaker 2: But that has changed the landscape a lot, so what

501
00:24:41,160 --> 00:24:44,080
Speaker 2: people expect as well has been raised, so users expect

502
00:24:44,240 --> 00:24:46,600
Speaker 2: a sort of a gener AI tool most of the time.

503
00:24:47,000 --> 00:24:48,600
Speaker 2: But I think you have to do something that's unique.

504
00:24:48,640 --> 00:24:51,880
Speaker 2: You can't just recreate chat GPT for science. I think

505
00:24:51,880 --> 00:24:54,480
Speaker 2: you have to contribute your own thing. And that's really

506
00:24:54,480 --> 00:24:57,520
Speaker 2: where the visual diagrams help a lot. And that's sort

507
00:24:57,520 --> 00:24:59,960
Speaker 2: of our point of view, is that human plus AI,

508
00:25:00,040 --> 00:25:02,399
Speaker 2: as I mentioned, to leverage some of the technology and

509
00:25:02,440 --> 00:25:05,080
Speaker 2: a great user interface and deliver from the customer.

510
00:25:05,960 --> 00:25:08,800
Speaker 1: The issue of bias is talked about a lot in

511
00:25:08,840 --> 00:25:12,399
Speaker 1: relation to AI. That's really why governments are reluctant to

512
00:25:12,480 --> 00:25:15,760
Speaker 1: use it for decision making. We've seen some examples of,

513
00:25:15,800 --> 00:25:19,160
Speaker 1: for instance, the courts in the US being biased against

514
00:25:20,840 --> 00:25:25,239
Speaker 1: black people in decision making in courts, so that was

515
00:25:25,520 --> 00:25:29,720
Speaker 1: a backlash quite rightly so against that. I guess because

516
00:25:29,760 --> 00:25:34,840
Speaker 1: you are you're citing established literature removes some of the

517
00:25:35,040 --> 00:25:36,959
Speaker 1: scope for bias here. But I guess there is bias

518
00:25:37,000 --> 00:25:43,280
Speaker 1: within the scientific process as well. Beyond the impact factor

519
00:25:43,320 --> 00:25:47,760
Speaker 1: of research, there are other things that, just as human bias,

520
00:25:47,840 --> 00:25:51,879
Speaker 1: we will gravitate towards certain types of information.

521
00:25:52,160 --> 00:25:54,359
Speaker 3: Is that something you have to deal with, Yeah, I

522
00:25:54,359 --> 00:25:55,440
Speaker 3: think to a lesser extent.

523
00:25:55,480 --> 00:25:58,479
Speaker 2: I think, yes, there's obviously the occurring of citations of

524
00:25:59,080 --> 00:26:01,199
Speaker 2: something that's highly started, it gets more citations, and so

525
00:26:01,240 --> 00:26:05,800
Speaker 2: there's a sort of a Matthew's effect there. But yeah,

526
00:26:05,800 --> 00:26:07,439
Speaker 2: I guess from our point of view, we want to

527
00:26:07,440 --> 00:26:09,960
Speaker 2: empower the user. So if they have a way of

528
00:26:10,000 --> 00:26:12,000
Speaker 2: looking at the literature that's important to them, they can

529
00:26:12,000 --> 00:26:14,639
Speaker 2: put in their topics and then we help triarge the

530
00:26:14,680 --> 00:26:17,320
Speaker 2: results by that. And obviously there might be bias and

531
00:26:17,400 --> 00:26:21,760
Speaker 2: the way that the model is then sorting those results

532
00:26:21,760 --> 00:26:25,480
Speaker 2: into those triarched topics for you. But I think it's

533
00:26:26,320 --> 00:26:30,480
Speaker 2: a really being transparent and exposing how we do things

534
00:26:30,560 --> 00:26:33,040
Speaker 2: is helpful so that you can see the bias at

535
00:26:33,119 --> 00:26:36,280
Speaker 2: least and have an understanding of what that means. I think, yeah,

536
00:26:36,320 --> 00:26:39,000
Speaker 2: inherently there's going to be issues like that to deal with,

537
00:26:39,080 --> 00:26:41,960
Speaker 2: but I think if we're transparent and we give user agency,

538
00:26:41,840 --> 00:26:43,840
Speaker 2: they can have impact there.

539
00:26:44,040 --> 00:26:47,200
Speaker 3: Yeah. So what's the next steps for maps.

540
00:26:47,200 --> 00:26:50,320
Speaker 1: You've obviously got this business that you'll be integrating research

541
00:26:50,400 --> 00:26:51,240
Speaker 1: raviity picking.

542
00:26:51,080 --> 00:26:52,159
Speaker 3: Up a team of people with that.

543
00:26:52,560 --> 00:26:55,359
Speaker 2: Yeah, so part of the capital was to help scale

544
00:26:55,400 --> 00:26:57,280
Speaker 2: the team, and obviously we've had our own growth to

545
00:26:57,320 --> 00:27:00,000
Speaker 2: one mile of rr A run rate, and that meant,

546
00:27:00,200 --> 00:27:02,480
Speaker 2: you know, we need more team members to deal with that.

547
00:27:02,640 --> 00:27:07,200
Speaker 2: So there's yeah, that effort has really enabled us to

548
00:27:07,240 --> 00:27:09,920
Speaker 2: scale up and address the opportunity ahead of us.

549
00:27:10,160 --> 00:27:13,919
Speaker 1: Yes, that's a million annual recurring revenue they call it

550
00:27:13,960 --> 00:27:16,720
Speaker 1: in the in the software startup world.

551
00:27:16,720 --> 00:27:19,000
Speaker 3: That's great. What's your business model?

552
00:27:19,040 --> 00:27:21,880
Speaker 1: How do people there's a premium tier and a free tier,

553
00:27:22,280 --> 00:27:23,240
Speaker 1: like a lot of these.

554
00:27:23,160 --> 00:27:23,840
Speaker 3: Sorts of services.

555
00:27:23,960 --> 00:27:26,719
Speaker 2: Yeah, that's right. So we do think that, you know,

556
00:27:26,920 --> 00:27:29,359
Speaker 2: we want to deliver a quality service. So essentially as

557
00:27:29,359 --> 00:27:30,879
Speaker 2: a freemium model, so you can get a lot of

558
00:27:30,920 --> 00:27:33,240
Speaker 2: value for free, you can throw in a few papers

559
00:27:33,240 --> 00:27:35,679
Speaker 2: that you know and quickly get recommendations. But once you

560
00:27:35,720 --> 00:27:39,280
Speaker 2: really dive into a specific topic ERAa you want you know,

561
00:27:39,400 --> 00:27:43,480
Speaker 2: maybe uploading thousands of papers and seeing recommendations from that,

562
00:27:43,480 --> 00:27:45,920
Speaker 2: that's when you subscribe and pay for a pro tier

563
00:27:46,240 --> 00:27:49,080
Speaker 2: of our tool. So it's that software as a service

564
00:27:49,160 --> 00:27:51,880
Speaker 2: model where there's a large free user base and then

565
00:27:52,280 --> 00:27:55,320
Speaker 2: a smaller percentage pain And it's interesting, you know that

566
00:27:55,720 --> 00:27:59,879
Speaker 2: in this sort of scientific literature space. We've had another

567
00:28:00,040 --> 00:28:03,480
Speaker 2: very successful in New Zealand, started at Publands, came again

568
00:28:03,840 --> 00:28:07,760
Speaker 2: out of Wellington out of Creative HQ. The accelerator here

569
00:28:08,000 --> 00:28:11,439
Speaker 2: in town also got to I think over a million

570
00:28:11,720 --> 00:28:15,080
Speaker 2: users before it was acquired, So it's interesting. It's too

571
00:28:15,520 --> 00:28:17,439
Speaker 2: in a similar field they were doing slightly different. That

572
00:28:17,480 --> 00:28:20,360
Speaker 2: was more about I think research reviews and getting visibility

573
00:28:20,680 --> 00:28:23,520
Speaker 2: into those. But what does that say do you think

574
00:28:23,560 --> 00:28:28,800
Speaker 2: about about how we come up with successful ideas in

575
00:28:29,160 --> 00:28:32,880
Speaker 2: this whole space? In New Zealand, Sir Paul Callahan famously said,

576
00:28:32,920 --> 00:28:37,600
Speaker 2: look for the lucrative niches that no one is doing

577
00:28:38,400 --> 00:28:40,959
Speaker 2: well around the world. Is that the case for you?

578
00:28:41,000 --> 00:28:44,040
Speaker 2: Do you think that you looked at what was happening

579
00:28:44,240 --> 00:28:48,320
Speaker 2: in the US and Europe, the big bastions of academic

580
00:28:48,360 --> 00:28:51,560
Speaker 2: literature and research, and you just weren't seeing anything like it. Yeah,

581
00:28:51,600 --> 00:28:54,440
Speaker 2: so we definitely thought we had a unique contribution to make,

582
00:28:54,520 --> 00:28:57,480
Speaker 2: and we were lucky enough to be connected with the

583
00:28:57,520 --> 00:29:00,280
Speaker 2: founders behind Publons and that really helped us as well.

584
00:29:00,320 --> 00:29:02,880
Speaker 2: I think they blazed a trail for us to see

585
00:29:02,920 --> 00:29:04,680
Speaker 2: that there is some value here, and you know, some

586
00:29:04,760 --> 00:29:07,440
Speaker 2: investors were on that journey so they could see, oh,

587
00:29:07,440 --> 00:29:10,280
Speaker 2: there's another possibility here. So that was really awesome. And

588
00:29:10,280 --> 00:29:13,240
Speaker 2: I think because this is a real particular niche. I

589
00:29:13,280 --> 00:29:16,640
Speaker 2: think having a domain expert essentially and the doorstep has

590
00:29:16,680 --> 00:29:20,760
Speaker 2: been really helpful. And obviously and as well, Andrew from

591
00:29:20,760 --> 00:29:23,719
Speaker 2: Pavlon's led this round with his angel group Great Up

592
00:29:23,720 --> 00:29:26,400
Speaker 2: in the UK, so there is almost a guess the

593
00:29:26,400 --> 00:29:31,480
Speaker 2: next generation of that sort of startup and that theme coming.

594
00:29:31,280 --> 00:29:33,240
Speaker 3: Through as well. So I think it's very personal.

595
00:29:33,280 --> 00:29:36,040
Speaker 2: There's a people element, but also I think there's a

596
00:29:36,040 --> 00:29:40,520
Speaker 2: lot of science and I guess expertise in Wellington. That's

597
00:29:40,560 --> 00:29:43,200
Speaker 2: meant you know, these types of things may happen in

598
00:29:43,200 --> 00:29:43,960
Speaker 2: the future as well.

599
00:29:45,080 --> 00:29:46,560
Speaker 3: But yeah, that's what my take on.

600
00:29:47,040 --> 00:29:50,920
Speaker 1: I'm glad to hear that about Andrew investing because that's

601
00:29:50,920 --> 00:29:54,640
Speaker 1: the model that is really sustaining and growing the impact

602
00:29:54,640 --> 00:29:59,560
Speaker 1: of our startup ecosystem is founders reinvest in some of

603
00:29:59,600 --> 00:30:03,600
Speaker 1: the proces into their sales, into the next generation of startups.

604
00:30:03,600 --> 00:30:06,240
Speaker 3: So that's and the expertise as well.

605
00:30:06,520 --> 00:30:09,360
Speaker 1: They know they've seen the opportunities and they're looking back

606
00:30:09,360 --> 00:30:11,840
Speaker 1: in New Zealand going there's great ideas here.

607
00:30:11,840 --> 00:30:13,680
Speaker 3: I want to help foster them and take them global

608
00:30:13,720 --> 00:30:14,080
Speaker 3: as well.

609
00:30:14,720 --> 00:30:17,720
Speaker 2: Yeah, and it's really great obviously to have some of

610
00:30:17,760 --> 00:30:21,080
Speaker 2: that experience and you know he's worked for weather Science

611
00:30:21,120 --> 00:30:24,840
Speaker 2: that claravate and run product there. So that's also a

612
00:30:24,840 --> 00:30:26,840
Speaker 2: massive thing to have someone who's not only how to

613
00:30:26,840 --> 00:30:29,840
Speaker 2: start up an exodent but also became part of that

614
00:30:29,880 --> 00:30:33,000
Speaker 2: big bastion tool that everyone knows and helping to feed

615
00:30:33,080 --> 00:30:35,960
Speaker 2: back into the next generation of companies as time goes on,

616
00:30:36,040 --> 00:30:38,000
Speaker 2: and he's got that experience behind him.

617
00:30:38,600 --> 00:30:39,040
Speaker 3: Excellent.

618
00:30:39,120 --> 00:30:41,880
Speaker 1: Well, Accident a great business and one it has a

619
00:30:41,880 --> 00:30:47,640
Speaker 1: great mission helping researchers accelerate the understanding of knowledge and

620
00:30:47,720 --> 00:30:50,640
Speaker 1: creation of new knowledge. So good luck as you bring

621
00:30:50,720 --> 00:30:54,400
Speaker 1: research Rabbit into the fold and begin to build your

622
00:30:54,440 --> 00:30:56,880
Speaker 1: revenue into your userbase worldwide.

623
00:30:57,080 --> 00:31:00,000
Speaker 3: Awesome, thanks so much. Thank you.

624
00:31:05,200 --> 00:31:09,560
Speaker 1: That was Accident co founder off Litmaps sharing the remarkable

625
00:31:09,600 --> 00:31:13,160
Speaker 1: story behind what I think is one of New Zealand's

626
00:31:13,400 --> 00:31:15,360
Speaker 1: most exciting emerging sort.

627
00:31:15,200 --> 00:31:16,080
Speaker 3: Of text stories.

628
00:31:16,400 --> 00:31:18,680
Speaker 1: It's really an example I think of the sort of

629
00:31:18,680 --> 00:31:22,680
Speaker 1: start up the late great physicist Sir Paul Callahan encouraged

630
00:31:22,760 --> 00:31:27,880
Speaker 1: New Zealand to produce one that focuses on lucrative international

631
00:31:28,000 --> 00:31:31,840
Speaker 1: niches that aren't really being well served by bigger markets.

632
00:31:32,240 --> 00:31:36,280
Speaker 1: Scientific literature, turns out, is one of them. Publons which

633
00:31:36,360 --> 00:31:40,440
Speaker 1: was founded by Andrew Preston in Wellington in twenty twelve,

634
00:31:41,120 --> 00:31:45,080
Speaker 1: allowed academics to track, verify, and showcase their peer review

635
00:31:45,160 --> 00:31:50,200
Speaker 1: and editorial contributions to academic journals. It was bought in

636
00:31:50,400 --> 00:31:54,600
Speaker 1: twenty seventeen by Claravate, the New York Stock Exchange listed

637
00:31:54,640 --> 00:31:57,880
Speaker 1: company that owns a Web of Science, a massive collection

638
00:31:58,480 --> 00:32:02,840
Speaker 1: of scientific literature atabases. That was a multimillion dollar deal

639
00:32:02,920 --> 00:32:07,080
Speaker 1: undisclosed exactly how much at the time. So great to

640
00:32:07,120 --> 00:32:10,800
Speaker 1: see Andrew involved in lip Maps two as an investor

641
00:32:10,920 --> 00:32:15,160
Speaker 1: and a mentor. So thanks to Axton for coming on.

642
00:32:15,560 --> 00:32:18,560
Speaker 1: Thanks to you for listening. Head over to Business Tesk

643
00:32:18,600 --> 00:32:21,560
Speaker 1: dot co dot nz to access the show notes in

644
00:32:21,640 --> 00:32:24,720
Speaker 1: the podcast section there, including my top ten list of

645
00:32:24,760 --> 00:32:27,560
Speaker 1: text stories to read this week from around the web.

646
00:32:28,160 --> 00:32:31,360
Speaker 1: Don't forget to subscribe to the podcast via your favorite

647
00:32:31,480 --> 00:32:36,240
Speaker 1: podcasting app or on iHeartRadio. Leave a review or rating two.

648
00:32:36,480 --> 00:32:38,680
Speaker 1: If you like the show and get in touch with

649
00:32:38,720 --> 00:32:42,040
Speaker 1: your feedback and guest suggestions, just email me on Peter

650
00:32:42,160 --> 00:32:46,200
Speaker 1: Atpetergriffin dot co dot nz or connect with me on LinkedIn.

651
00:32:46,560 --> 00:32:49,800
Speaker 1: I'll catch you next week for our one hundredth episode

652
00:32:49,880 --> 00:32:50,880
Speaker 1: of the Business of Tech.

653
00:32:51,200 --> 00:32:51,680
Speaker 3: See you then