Meeting Title: Brainforge Interview w- Sam Date: 2026-02-18 Meeting participants: Samuel Roberts, Ruixi Wen
WEBVTT
1 00:03:12.740 ⇒ 00:03:13.630 Samuel Roberts: Hello?
2 00:03:32.680 ⇒ 00:03:33.749 Samuel Roberts: Can you hear me?
3 00:03:39.150 ⇒ 00:03:42.380 Samuel Roberts: If you’re talking, I can’t hear anything.
4 00:04:05.990 ⇒ 00:04:08.260 Ruixi Wen: Oh, hi, Sammy, you can’t hear me now.
5 00:04:08.260 ⇒ 00:04:10.329 Samuel Roberts: Yes, now I can. There you go, perfect.
6 00:04:10.330 ⇒ 00:04:16.999 Ruixi Wen: Okay, okay, perfect. Yeah, sorry, like, my computer recently has some, like, audio connection issue. It takes, like, really long.
7 00:04:17.000 ⇒ 00:04:24.000 Samuel Roberts: Oh, no. Yeah, I have… I’ve been having all kinds of trouble with my setup and my camera, so I had to get a new mic and everything, so I understand that.
8 00:04:24.840 ⇒ 00:04:30.429 Ruixi Wen: It’s annoying. Yeah, yeah. Thank you for taking the time to have this call with me today.
9 00:04:30.430 ⇒ 00:04:41.720 Samuel Roberts: Of course, of course, yeah. So, yeah, welcome, I’m Sam, so I, I work at Brainforge, in the, kind of, AI, kind of the AI tech lead,
10 00:04:42.080 ⇒ 00:04:46.080 Samuel Roberts: So… yeah, I think this is their… is this your first…
11 00:04:46.230 ⇒ 00:04:49.169 Samuel Roberts: conversation with someone? Or have you talked to people…
12 00:04:49.510 ⇒ 00:04:58.889 Ruixi Wen: Yeah, I actually, like, talked to both, like, Roberts and OTEM. Yeah, OTEM shared a lot of… about, like, how does, like, the company works. Okay, great.
13 00:04:59.050 ⇒ 00:05:07.429 Ruixi Wen: do, like, internal products, and both of them, I think, like, are probably, like, fit into, like, a better role in, like, like, AIPM, like, to help with.
14 00:05:07.430 ⇒ 00:05:08.260 Samuel Roberts: Yeah.
15 00:05:08.260 ⇒ 00:05:09.370 Ruixi Wen: workflows in.
16 00:05:09.370 ⇒ 00:05:12.960 Samuel Roberts: Okay, great, yeah. Okay, cool, I just didn’t have the notes here, so I wanted to make sure.
17 00:05:13.310 ⇒ 00:05:24.190 Ruixi Wen: Yeah, but he didn’t, like, went on to, like, too much details on, like, what kind of projects that are leading and all that, but… so, like, that’s why he referred me to you. So, yeah, perfect.
18 00:05:24.190 ⇒ 00:05:35.309 Samuel Roberts: Okay, great, yeah, so I think, just to start, I’d like to get to know your background a little bit, so if you could… I saw your LinkedIn, but if you could just give me, like, a quick, you know…
19 00:05:36.130 ⇒ 00:05:39.390 Samuel Roberts: elevator pitch of yourself, I guess, background, that’d be great.
20 00:05:39.940 ⇒ 00:05:55.430 Ruixi Wen: Yes, for sure, for sure. Yeah, so I graduated from USC, last year, and I actually, like, went to the same program as Roberts, so that’s how I got, like, to connect and know what he was building. And before that, I was, like, mainly, like, doing, like, consulting and finance internships, like, every business… Okay.
21 00:05:55.520 ⇒ 00:06:05.049 Ruixi Wen: But, like, it really occurred to my fourth year, where, like, I got on, like, a lot of… from my internship, I got on deals with, like, companies in data, where, like, they’re doing, like, data.
22 00:06:05.050 ⇒ 00:06:05.730 Samuel Roberts: Okay.
23 00:06:05.730 ⇒ 00:06:12.079 Ruixi Wen: data integration, and I found, like, wait, this is, like, so interesting, like, because in consulting, like, you have to, like.
24 00:06:12.350 ⇒ 00:06:25.890 Ruixi Wen: like, see all different kind of industry… projects from different industries, you don’t get to, like, solely work on tech industry and data ones. So I started recruiting, and I got into TensorStacks, which is a company that AI agent tool for data engineering.
25 00:06:25.890 ⇒ 00:06:36.480 Ruixi Wen: For… it’s very, like, it boosts, like, the data engineering team efficiency and productivity, so it can autonomously build models… data models and maintain data pipelines.
26 00:06:36.480 ⇒ 00:06:40.739 Ruixi Wen: And it works, like, particularly well on top of, like, dbt and the airflow.
27 00:06:40.850 ⇒ 00:06:46.389 Ruixi Wen: And when I joined the company, I was drawing as, like, a founding go-to-market here, but because, like…
28 00:06:46.390 ⇒ 00:07:01.120 Ruixi Wen: me is, still, like, like, was, like, shipping features very quickly, and I was also the first contact point, for the product engineering team and the client. So I somehow just, like, put into this situation where, like, a lot of, like, the
29 00:07:01.120 ⇒ 00:07:07.579 Ruixi Wen: shipping features, I need to tell, like, the product engineering team weekly, like, oh, what we need to prioritize. Okay.
30 00:07:07.580 ⇒ 00:07:22.800 Ruixi Wen: the roadmap looking like, especially… we also, like, helping with, like, B2B customers, so… Okay. A lot of… we go through a POC stage, and clients were just, like, kept on… we kept on talking with clients to say, like, their testing results, and what these… what features they…
31 00:07:22.880 ⇒ 00:07:42.039 Ruixi Wen: what they’re… they look for, and what they think that’s particularly helpful for them. So, I would say, like, but also, like, I think just, like, part of the sales process. And I think I personally, like, really much enjoy this part of the process, for the process, which I didn’t get exposure to when I was, like, in college doing internships. Sure.
32 00:07:42.170 ⇒ 00:07:56.810 Ruixi Wen: the company, like, grew very fast, and I helped the company to create, like, this, like, first 2 million ARR as well. And the company decided, to take an acquisition offer from Snowflake at the end of the last year. Oh, nice.
33 00:07:57.470 ⇒ 00:08:03.789 Ruixi Wen: Yeah, so, like, they got acquired by Snowflake right now, and they’re, like, building, like, the Corda’s AI scene for Snowflake.
34 00:08:03.790 ⇒ 00:08:05.050 Samuel Roberts: Oh, yeah, yeah, okay.
35 00:08:05.470 ⇒ 00:08:11.530 Ruixi Wen: Yeah, yeah, yeah, yeah, yeah. Yeah, I think they just, like, announced, like, earlier this month, yeah. Sweet.
36 00:08:11.960 ⇒ 00:08:26.079 Ruixi Wen: Yeah, but for me, like, I think I’m more so interested in, like, the startup board, and after the acquisition, for me, like, they wanted me to put, like, a snowflake, sales team in general, but I think, like, that’s not where I’m seeing myself, going long.
37 00:08:26.080 ⇒ 00:08:26.900 Samuel Roberts: Right.
38 00:08:27.280 ⇒ 00:08:28.180 Samuel Roberts: Okay
39 00:08:28.340 ⇒ 00:08:35.309 Samuel Roberts: Great, okay, thank you. Yeah, perfect. That definitely helps me understand a little bit more of your background. Great. Alright, so,
40 00:08:35.610 ⇒ 00:08:41.779 Samuel Roberts: So you mentioned a little bit of dealing with, like, product team and stuff, like, bringing things… I guess I’m wondering, like, what…
41 00:08:42.780 ⇒ 00:08:48.439 Samuel Roberts: I guess, what types of projects have you worked on? Like, more specifics, I guess.
42 00:08:48.780 ⇒ 00:08:55.989 Samuel Roberts: in that sense, like, you gave me a little bit of a sense of, with the company, but I’m curious, like, what more specifics there?
43 00:08:55.990 ⇒ 00:09:15.949 Ruixi Wen: Yeah, yeah, yeah. I think, like, it really depends. So, like, it comes down to, like, to help us to better the product, because we already have, like, a product ready to be shipped and deployed. Okay. It’s more about, like, really bettering, like, its features and bettering, like, we have our own research team, so it’s also, like, helping, to communicate.
44 00:09:15.950 ⇒ 00:09:20.969 Ruixi Wen: the models we’re building. But I would say, like, I’m more so on the
45 00:09:20.970 ⇒ 00:09:33.349 Ruixi Wen: feature side, because the model we released, like, at the very end, like, right, kind of, like, right before the acquisition, so, I could just, like, get the client to know about, like, what we’re doing here.
46 00:09:33.350 ⇒ 00:09:37.359 Ruixi Wen: So, for example, some features, I would say, like,
47 00:09:37.360 ⇒ 00:09:55.379 Ruixi Wen: for example, like, a lot of data engineering managers I talk to, like, they’re pretty, frustrated with, like, the lineage, thing they see. Like, a lot of times, like, you don’t… when they pull out, like, things, like, you… like, you don’t get to see the lineage, so it’s, like, really hard for you to debug.
48 00:09:55.380 ⇒ 00:09:55.990 Samuel Roberts: Right.
49 00:09:55.990 ⇒ 00:10:08.389 Ruixi Wen: Like, for one feature that helped was, like, we have, like, a lineage that not only show, like, within the product. So, like, when you see a bug, you can go back to lineage and really track down to, like, where it’s coming from.
50 00:10:08.390 ⇒ 00:10:17.290 Ruixi Wen: But others, sometimes they also have, like, cross-organization, cross-project ones, and, like, the feature we ship, also able to, like, track down to, like, a completely
51 00:10:17.290 ⇒ 00:10:31.690 Ruixi Wen: different product, or across organizations. So that really helps them to, like, save a lot of time to know where the bug is coming, and especially, like, our tool, like, helped with, like, just generate the DBD code, like, with, like, some natural language input.
52 00:10:32.310 ⇒ 00:10:35.739 Ruixi Wen: And obviously, like, sometimes AI is, like, not doing its…
53 00:10:35.890 ⇒ 00:10:59.609 Ruixi Wen: It’s, like, although our product’s, like, very… it’s already, like, very vertical, like, it’s meant to, like, succeed on that, but we have, like, this loop verification, and whenever, like, you see something you don’t understand, so you can also go back to lineage to track it down to see, like, oh, where… where was that coming from? And to see if it’s, like, actually mirrored, what the employee has already typed, for the contacts.
54 00:10:59.610 ⇒ 00:11:00.469 Ruixi Wen: Of the data.
55 00:11:00.470 ⇒ 00:11:01.050 Samuel Roberts: Cool.
56 00:11:01.050 ⇒ 00:11:06.520 Ruixi Wen: So, yeah, that’s, like, one feature, like, is I would say, like, shipped very, very effectively. We see, like, some.
57 00:11:06.520 ⇒ 00:11:07.180 Samuel Roberts: Okay.
58 00:11:07.340 ⇒ 00:11:08.460 Ruixi Wen: Oh, well, yeah.
59 00:11:08.990 ⇒ 00:11:11.170 Samuel Roberts: Great, great. So, I guess,
60 00:11:11.330 ⇒ 00:11:19.730 Samuel Roberts: In terms of, like, communication and shepherding something like that through, how would you explain, like, project risks to people that are maybe less…
61 00:11:20.000 ⇒ 00:11:28.869 Samuel Roberts: like, non-technical clients, I guess, in this case, you know? So, like, for us, you know, we have to communicate a lot of that stuff, and I’m curious, how would you tend to…
62 00:11:29.050 ⇒ 00:11:31.800 Samuel Roberts: Think about communicating that kind of project risk.
63 00:11:32.420 ⇒ 00:11:39.740 Ruixi Wen: I see. Actually, I have to say, like, just because of, like, the nature of our product, like, our ICPS is, like, fully sold into, like, VKO,
64 00:11:40.040 ⇒ 00:11:41.159 Ruixi Wen: nearing CDU.
65 00:11:41.160 ⇒ 00:11:42.340 Samuel Roberts: Yeah, yeah.
66 00:11:42.510 ⇒ 00:11:50.689 Ruixi Wen: head of data, and so, like, actually, like, everybody I talk to, like, are pretty technical. Even for the local lines would be, like, a senior data engineer.
67 00:11:50.690 ⇒ 00:11:51.290 Samuel Roberts: Sure.
68 00:11:51.290 ⇒ 00:11:57.340 Ruixi Wen: This problem didn’t really occur just because, like, we… for… I think it’s very different from Brainforge, like, it’s more.
69 00:11:57.340 ⇒ 00:11:58.340 Samuel Roberts: Yeah.
70 00:11:58.340 ⇒ 00:11:58.850 Ruixi Wen: starting.
71 00:11:58.850 ⇒ 00:11:59.520 Samuel Roberts: Okay.
72 00:11:59.690 ⇒ 00:12:00.639 Ruixi Wen: Fine, but for us, like.
73 00:12:00.640 ⇒ 00:12:01.530 Samuel Roberts: Okay.
74 00:12:01.530 ⇒ 00:12:02.609 Ruixi Wen: Hopefully, yeah.
75 00:12:02.610 ⇒ 00:12:09.700 Samuel Roberts: Sure, sure, okay, so, like, I guess just talk me through, like, how you… if you were put in that position, then, what would you… what would be, like, the things you’d wanna…
76 00:12:09.880 ⇒ 00:12:16.150 Samuel Roberts: Think about when you’re trying to communicate project risk to people that might be less technical than the people you are used to talking to?
77 00:12:16.530 ⇒ 00:12:31.990 Ruixi Wen: Mmm, I’m thinking about that. So, I would say, like, even for talking to technical people, there are a lot of, like, risks, and I think, like, it’s kind of translatable, transferable, even for non-technical people. Sure. So, for example, when risk involved is, like.
78 00:12:31.990 ⇒ 00:12:54.470 Ruixi Wen: companies, we work in healthcare, or even about government-related jobs, so they really care about, like, their data sensitivity. They don’t want us to, like, have access to the metadata, and also they, want to be deployed as safe as possible. They just have, like, so much thought about that, and also, like, during the sales process, the infosec for them is always the longest, like, I think they’re…
79 00:12:54.470 ⇒ 00:13:16.740 Ruixi Wen: was, like, once I take 3-4 months, just, like, every week we have, like, documents to review and sign and communicate through to make sure, like, everything was, like, super compliant to gain the trust of the customer. So I think, like, that’s the part during the AI transformation overall, like, we’re like, oh, is this reliable, like, compared to, like, manually coded things? And when we share, like.
80 00:13:16.740 ⇒ 00:13:22.239 Ruixi Wen: literally our data, which is, like, the most valuable thing in today’s world. Right.
81 00:13:22.240 ⇒ 00:13:30.890 Ruixi Wen: Yeah, and I think, like, for this kind of risk is, like, we really try to, like, I think, like, elite… really try to, like,
82 00:13:31.050 ⇒ 00:13:37.610 Ruixi Wen: like… gain a trust, like, through three ways, three ways. So, one thing’s, like, something’s, like.
83 00:13:37.720 ⇒ 00:13:51.370 Ruixi Wen: we build our product, like, we don’t get access… by default, we don’t get access to… get to see, like, the metadata part. Although, like, we can, but, like, by default, we don’t… we don’t, and our NDA, like, ban us from doing that.
84 00:13:51.370 ⇒ 00:14:00.989 Ruixi Wen: Second thing is, like, we have, like, this human loops, like, you are always there, able to track, like, what’s happening, like, it’s not just, like, AI wrong, run everything, like, you can…
85 00:14:00.990 ⇒ 00:14:08.379 Ruixi Wen: you can set the mode yourself. And you don’t allow AI to do something it wouldn’t do for you, but if you want them to run, like, they will help you.
86 00:14:08.380 ⇒ 00:14:21.859 Ruixi Wen: So you can just, like, really be in the loop, updated from that. And I think, like, for brain projects, the same thing, that’s why it’s, like, a more service consulting, like, humans are always involved to make sure everything works properly when it’s an integrated system.
87 00:14:21.860 ⇒ 00:14:28.610 Ruixi Wen: And the third thing, I think it’s more about, like, I really need to talk with the engineering team to make sure, like, the deployment is, like, what they want.
88 00:14:28.610 ⇒ 00:14:42.440 Ruixi Wen: So, like, we always, like, put their data into, like, containers and ship to them, or if necessary, we will make it, like, a SaaS format, whichever, like, format they’re most comfortable with. And I think that part, like, it’s really, like, a service part, especially talking to, like.
89 00:14:42.480 ⇒ 00:14:46.949 Ruixi Wen: it’s a B2B business, so whatever customer is, like, most comfortable with.
90 00:14:47.360 ⇒ 00:14:54.579 Samuel Roberts: Sure, sure. Okay, great. Okay, I guess, tell me a little more about yourself. Like, what are your career goals, for example?
91 00:14:54.980 ⇒ 00:15:13.139 Ruixi Wen: Yes, for me, I think I… I really… I think, like, I really enjoy, like, working in, like, an AI world. I think it’s, like, just, like, I think for me, what really makes me happy was, like, this is such a, like, you’re literally in the cutting edge of the forefront technology, and it’s just, like, a tool that has, like, so much…
92 00:15:13.140 ⇒ 00:15:28.969 Ruixi Wen: potential hasn’t been released, and when you’re talking to a customer, by fundamentally, you’re, like, painting the blueprint. You’re educating them, teaching them, like, something, they didn’t know about, and they didn’t know, like, it could work like that. It could help them to save so much time and money.
93 00:15:28.970 ⇒ 00:15:29.510 Samuel Roberts: Nape.
94 00:15:29.510 ⇒ 00:15:48.429 Ruixi Wen: And I definitely want to, like, like, like, this is a higher level goal, like, this is, like, my core motivation to, to, like, continue on this. And secondly, I think, like, although I do enjoy salsa Lodge, which, like, I get to talk with customers, but I think, like, the product part is just, like, you get to see your impact,
95 00:15:48.570 ⇒ 00:15:52.290 Ruixi Wen: more… Hold up, is this, like, more…
96 00:15:52.390 ⇒ 00:16:09.159 Ruixi Wen: more tangible way, because, like, sales is, like, you’re based on, like, whatever service or product they already have, but I think that when you come to products, like, you can hold, like, really level up the experience for the customers. Like, you, you take, like, a more proactive approach, like…
97 00:16:09.160 ⇒ 00:16:19.669 Ruixi Wen: to manage this thing. So, I think, like, for me, right now, like, my career goes, like, more pivoting to, like, more exposure to tech, and get to, like, more properly, like, manage, like, the workflows.
98 00:16:20.270 ⇒ 00:16:21.329 Samuel Roberts: Cool, cool, that sounds
99 00:16:22.250 ⇒ 00:16:26.079 Samuel Roberts: Excuse me, sorry, I’m coming down with a little bit of a cold, so I apologize if I’m coughing.
100 00:16:26.080 ⇒ 00:16:27.460 Ruixi Wen: Oh, no.
101 00:16:27.520 ⇒ 00:16:30.290 Samuel Roberts: Yeah, so I used to, excuse me. Hope you get a better food.
102 00:16:30.400 ⇒ 00:16:34.950 Samuel Roberts: Thank you, thank you, I appreciate that. Yeah, I guess,
103 00:16:35.350 ⇒ 00:16:42.179 Samuel Roberts: What, let’s start with something… start off on a good note, and then I’ll jump to the other one, but, what’s something you’re really good at, professionally?
104 00:16:42.670 ⇒ 00:16:51.570 Ruixi Wen: I think, like, something I’m really good at is, like, I’m really good at, like, communicating with, like, multiple stakeholders, and I think, like, I’m, like.
105 00:16:51.640 ⇒ 00:17:11.519 Ruixi Wen: pretty much, like, like, trained on that. So for example, like, in my past role, it was, like, communicating, like, with my CEO, like, because our team was also, like, not too big, and communicating with product engineering team, and also, like, like, a majority of time was, like, given to, like, communicate with, like, clients, and, like.
106 00:17:11.520 ⇒ 00:17:20.740 Ruixi Wen: down the pipeline, I probably have to, like, communicate with more than 5 clients in one day. And I think I’m pretty good, like, managing that, and I was able to, like.
107 00:17:20.740 ⇒ 00:17:34.239 Ruixi Wen: really, translate, like, what they… what’s, like, the proper way to different people well. And I’m also, like, really, like… I would say, like, I have this natural alert to know, like, what is happening, and to save the.
108 00:17:34.240 ⇒ 00:17:36.130 Samuel Roberts: Time for both sides the most.
109 00:17:36.130 ⇒ 00:17:36.600 Ruixi Wen: Yeah.
110 00:17:36.600 ⇒ 00:17:51.240 Samuel Roberts: Cool, okay. So then on the flip side of that, what is, what are you not good at, or maybe not interested in doing professionally? Like, where do you, like, either draw that line where, like, okay, that’s the stuff I’m not interested in, or I just… I’m not good at that, like, I want to stay in my lane kind of thing? What’s something like that?
111 00:17:51.590 ⇒ 00:18:02.699 Ruixi Wen: Yes, yes, I think, like, one thing I’m, like, not as good at was, like, because for my previous role, like, I was taking a sales drop, so for us, like, we…
112 00:18:02.700 ⇒ 00:18:22.699 Ruixi Wen: have, like, both, like, all-bound and inbound motions, and, the inbound motion, like, me and, like, another, like, the sales engineer handle them together, but, for the all-bound motion, I basically crafted from the scratch, which, like, takes a long time to do. So what it does, like, it really requires me to, every day, wake up and, like.
113 00:18:22.890 ⇒ 00:18:27.049 Ruixi Wen: Like, send, like, send, like, 100 linking requests, find the right process.
114 00:18:27.050 ⇒ 00:18:27.670 Samuel Roberts: Oh, yeah.
115 00:18:27.670 ⇒ 00:18:41.679 Ruixi Wen: Yep. And calling, like, making, like, 50 phone calls a day. I mean, it’s… I mean, it’s, like, fundamental for the company to grow, I would say, but I would just say it’s, like, pretty repetitive, and it’s, like.
116 00:18:41.680 ⇒ 00:18:49.079 Ruixi Wen: somewhere, like, you… you’re not guaranteed you get results. Like, sometimes I call as well, because, like, this person happened to be
117 00:18:49.200 ⇒ 00:19:06.949 Ruixi Wen: like, the best fit, so, like, it actually had the best result, but sometimes I call, like, I call more or reach out to more people, it doesn’t mean, like, they have… they’re in the right spot, where they’re interested in our product at this stage. So, I think, like, those kind of, like, repetitive things was, like, a part of the reason that
118 00:19:06.950 ⇒ 00:19:20.130 Ruixi Wen: at least, like, I’m not, like, that passionate about, like, Alban sales motion right now. Like, I tried that, it was… it was cool, like, it’s necessary for the company, but it’s, like, repetitive nature is not, like, very exciting to me, and I don’t feel like…
119 00:19:20.150 ⇒ 00:19:24.190 Ruixi Wen: I feel like I was pretty far from product side, I was thinking.
120 00:19:24.190 ⇒ 00:19:26.480 Samuel Roberts: Yeah, no, I totally see that, that definitely…
121 00:19:26.480 ⇒ 00:19:32.849 Ruixi Wen: So, like, I said, those repatting, like, bores me a little bit, but just, like, how to push it through in my previous job. Yep.
122 00:19:32.850 ⇒ 00:19:48.460 Samuel Roberts: Yep, great, great. Okay, well, I want to make sure you have time to ask questions about, you know, Brainforge or anything, so I don’t want to just keep peppering you with questions, but I have some more, but if you have any questions, feel free, you know, what can I… what can I tell you, what can I answer for you right now?
123 00:19:48.760 ⇒ 00:20:07.470 Ruixi Wen: Okay, amazing, thank you so much. Like, I really appreciate, like, the questions you asked about me, things are super, super valid questions, and I appreciate the chance, like, I can talk more about, like, myself. Yeah, maybe I will, like, ask a… like, a question, like, I’m most curious about, like, for example, for Q1 or for this year, like,
124 00:20:07.470 ⇒ 00:20:16.389 Ruixi Wen: For… could you talk more, like, more about, like, the details, like, your project? Like, if you’re… you can share them, like, some projects you’re leading, or some prioritization.
125 00:20:16.390 ⇒ 00:20:17.230 Samuel Roberts: Yeah.
126 00:20:17.230 ⇒ 00:20:18.589 Ruixi Wen: Eating a rhino, internally.
127 00:20:18.590 ⇒ 00:20:25.460 Samuel Roberts: Yeah, so I won’t get into specifics, obviously, but, yeah, we have a couple, so, you know, Brainforge.
128 00:20:25.590 ⇒ 00:20:35.420 Samuel Roberts: does a lot of data work for clients, and the AI is kind of a newer thing that we’ve started doing. And so we have a couple clients that we’re working with on, kind of.
129 00:20:35.740 ⇒ 00:20:50.169 Samuel Roberts: AI enablement in certain ways, and so, for example, one is a, you know, a chatbot, which is like, you know, classic AI interaction stuff, but, it’s chatting over a whole bunch of documents and a whole bunch of,
130 00:20:50.170 ⇒ 00:21:06.570 Samuel Roberts: basically assignments for different people, so customer service reps can more easily… you know, the company we’re working with, all of their customer service is done by people. They’re not interested in doing AI customer service, but we’re helping their customer service agents, you know, utilize an AI agent that can look up
131 00:21:06.570 ⇒ 00:21:26.310 Samuel Roberts: the relevant pieces of the docs for, you know, scheduling things, or refunds, or whatever it is. So we’re helping them, you know, enabling the people to do, you know, better work and help them that way. The other side of it is building a little bit of a… what would probably be something like a SaaS platform, but it’s just for a client.
132 00:21:27.100 ⇒ 00:21:47.549 Samuel Roberts: And so they’re looking… they have a bunch of other… they have clients that they, work with, and so there’s a lot of data sources there that we’re helping them bring in, but then being able to chat over that data, being able to, model things there, send reports, they have a whole bunch of employees that need to get into that system and do things for a number of different, brands and companies that they work with.
133 00:21:47.620 ⇒ 00:21:56.929 Samuel Roberts: And so we’re helping them, you know, kind of take what they’ve been doing a little bit with, you know, Claude or ChatGPT, and…
134 00:21:57.320 ⇒ 00:22:05.380 Samuel Roberts: customizing it a bit more. So, one thing that we’ve done a decent amount of is people are using these AI tools, but they’re not
135 00:22:05.460 ⇒ 00:22:21.900 Samuel Roberts: quite crafted for exactly their use case, or they’re doing something over and over and over again, and the AI can do it, but when you’re just chatting, it’s not necessarily the most efficient way to have it run. So building agents that can handle that a little more agentically, a little more on their own with a human in the loop, there’s a lot of that happening.
136 00:22:22.610 ⇒ 00:22:24.080 Samuel Roberts: Excuse me, sorry.
137 00:22:24.260 ⇒ 00:22:38.379 Samuel Roberts: Yeah, but so we’re, we’re, you know, kind of… we have, you know, not the biggest AI team right now compared to, other parts of the company, necessarily, but, number of engineers that, are working on a few of those different projects, as well as a couple other parts of different client work.
138 00:22:38.480 ⇒ 00:22:44.700 Samuel Roberts: So we’re… we’re spread out a little bit, we have our hands in lots of different things throughout different clients, but yeah, it’s,
139 00:22:45.450 ⇒ 00:23:01.669 Samuel Roberts: it’s really, you know, it’s interesting because, you know, I come from a startup, more product background, and so I’m used to working on, you know, one product with lots of different features and stuff, but selling one thing, building one thing, and so being able to work on lots of different things, I find, is really interesting for me, so…
140 00:23:02.310 ⇒ 00:23:09.300 Ruixi Wen: Yeah, yeah, yeah, I definitely see, like, the exposure will be, like, like, being exponential, like, being exponentially increased from there, yeah.
141 00:23:09.300 ⇒ 00:23:11.170 Samuel Roberts: Oh, yeah, yeah, definitely, definitely.
142 00:23:11.170 ⇒ 00:23:19.350 Ruixi Wen: I’m curious, like, could you, could you talk, like, a little bit more about, like, I’m curious, like, you mentioned twice for both of these projects, like, chat over data was, like.
143 00:23:19.350 ⇒ 00:23:20.260 Samuel Roberts: Yeah.
144 00:23:20.260 ⇒ 00:23:29.799 Ruixi Wen: Could you tell me, like, what does that mean? Is it, like, you guys have, like, a U by UX to, like, help to locate the data, like, you’re looking for? What does it… does it, like, look like?
145 00:23:29.800 ⇒ 00:23:33.629 Samuel Roberts: Yeah, so it’s… the idea, so for…
146 00:23:33.820 ⇒ 00:23:46.380 Samuel Roberts: for one of them, it’s more documents, it’s more, policies and things like that. So it’s not necessarily, you know, data where you need to aggregate things or calculate things, but it’s just…
147 00:23:46.880 ⇒ 00:23:56.269 Samuel Roberts: You know, the customer that the customer service rep is talking to wants to do something, but it’s a specific department in this other company, and so there’s different policies, and…
148 00:23:56.270 ⇒ 00:24:08.130 Samuel Roberts: you know, we built basically, like, a RAG pipeline, which, you know, retrieval augmented generation, so I can look that stuff up, pass that to the AI, the AI can explain that to the CSRs, and then they can tell that to the customer.
149 00:24:08.130 ⇒ 00:24:21.929 Samuel Roberts: The other side of it is a little more, you know, there’s data coming in from various sources, that, you know, normally you’d want to build a pipeline and aggregate and build some reports and maybe dashboards, but, they’re looking to actually
150 00:24:22.340 ⇒ 00:24:26.149 Samuel Roberts: use, you know, an AI chat, basically, to
151 00:24:26.150 ⇒ 00:24:44.259 Samuel Roberts: process a little bit of that, get some suggestions. They’re doing a little more creative work with it. So it’s not just, like, where’s the data, but, you know, what was the sales for the last week for this brand, and how did the emails work for that, and, you know, what other…
152 00:24:44.760 ⇒ 00:24:54.549 Samuel Roberts: You know, processing a little more of, maybe, the creative side of it, along with the data source, for the metrics.
153 00:24:54.550 ⇒ 00:25:09.769 Samuel Roberts: being able to kind of combine that into kind of one UI, not have to jump from different dashboards for different data providers, but then also bring all that together to put it into Claude or ChatGPT, and then, you know, work with that. So kind of centralizing all that into one
154 00:25:09.860 ⇒ 00:25:13.079 Samuel Roberts: One dashboard, really. One… not dashboard, but
155 00:25:13.440 ⇒ 00:25:15.840 Samuel Roberts: Interface that has a number of different things that they’re looking for.
156 00:25:15.840 ⇒ 00:25:17.589 Ruixi Wen: I interpret visualization for that.
157 00:25:17.590 ⇒ 00:25:18.800 Samuel Roberts: Yeah, exactly.
158 00:25:19.500 ⇒ 00:25:34.480 Ruixi Wen: I see. Yeah, that makes total sense, yeah. Yeah, especially, like, the first one, I just think, like, a few, like, frustration… frustrating experience I had, like, where they ship me products, and I look for customer service, and they’re like, oh, let me transfer you to someone, wait, let me go talk to someone, and I was, like, on phone for hours, I think, like.
159 00:25:34.850 ⇒ 00:25:39.360 Ruixi Wen: This is, like, a very, very common use case, that can help the customer.
160 00:25:39.360 ⇒ 00:25:48.690 Samuel Roberts: Yeah, and it’s really, yeah, it’s about, like, you know, enabling the people to get their work done that they need, rather than just, like, replacing the customer service rep or something like that.
161 00:25:49.230 ⇒ 00:25:55.940 Samuel Roberts: helping them access the tools, rather than… you know, because there’s risk to just putting the AI chatbots out there. I’m sure you’ve seen stuff like that, where people can…
162 00:25:55.940 ⇒ 00:25:56.760 Ruixi Wen: Yes.
163 00:25:56.760 ⇒ 00:26:00.470 Samuel Roberts: prompt, inject, and do all kinds of other things, so… Yeah.
164 00:26:01.160 ⇒ 00:26:03.919 Samuel Roberts: Other questions I can help?
165 00:26:04.210 ⇒ 00:26:12.460 Ruixi Wen: Yeah, I think another question I have is, like, for this role specifically, what do you think, like, are important, like, I don’t know, like.
166 00:26:12.460 ⇒ 00:26:19.350 Samuel Roberts: important things you look for that can be, like, really competent in this role, like, in terms of, like, skills, or… Yeah.
167 00:26:19.350 ⇒ 00:26:21.550 Ruixi Wen: Daily work that you need to be really good at.
168 00:26:22.340 ⇒ 00:26:31.330 Samuel Roberts: I mean, I think communication is one of the biggest things, obviously. You know, working on a team in general, that’s huge. But especially as a more…
169 00:26:32.240 ⇒ 00:26:44.159 Samuel Roberts: you know, product manager, project manager, kind of those sort of roles that are maybe not just at the keyboard. You know, engineers still need to be able to communicate, but, you know, we can kind of sometimes shove that off to someone else, and that’s kind of, you know.
170 00:26:44.370 ⇒ 00:27:01.459 Samuel Roberts: being able to wrangle people together, make sure everyone’s on the same page, make sure that, requirements are clear, all that sort of stuff, communicating what’s possible and what’s, feasible, and, like, understanding
171 00:27:02.000 ⇒ 00:27:05.310 Samuel Roberts: What the technology can do is a huge part of it, because.
172 00:27:05.610 ⇒ 00:27:09.110 Ruixi Wen: Yeah. Excuse me, things are changing all the time, so it’s staying up to date with that.
173 00:27:09.380 ⇒ 00:27:21.570 Samuel Roberts: can be… can be challenging, even, you know, even when you’re very technical, you know. So, I think kind of having a good sense of the state of the art, and then also being able to then,
174 00:27:22.210 ⇒ 00:27:34.500 Samuel Roberts: explain that to the clients that might be less, you know, even less technical than the product manager, which is less technical, probably, than the engineer. So there’s a level of, like, who are you talking to at any given time, and making sure that that’s…
175 00:27:34.740 ⇒ 00:27:49.649 Samuel Roberts: you know, you’re explaining things the right way, which I think is important for all roles, really. You know, you need to be able to communicate to the people you’re talking to, but I think specifically for a role like this, you’re a bit of a… almost like a translator between different groups.
176 00:27:49.830 ⇒ 00:27:55.170 Samuel Roberts: Where… Yeah, I think that’s kind of the biggest thing, like, making sure that the right people understand
177 00:27:55.310 ⇒ 00:27:58.679 Samuel Roberts: And that people aren’t talking past each other in certain ways is big.
178 00:27:59.590 ⇒ 00:28:02.199 Ruixi Wen: Yes, yes, yes, totally, totally.
179 00:28:02.250 ⇒ 00:28:12.870 Ruixi Wen: Yeah, I… especially, like, the first point, like, to really understand, like, the… where the technology at? Like, I feel like in my previous role, I always got… because I’m very client-facing, so I always got challenged
180 00:28:12.870 ⇒ 00:28:26.409 Ruixi Wen: from clients. Oh, so how about, like, from this, like, report that you really newly released, how do you guys differentiate with them? How do you perform better than them? And I was like, well, what is happening? Like, and I have to, like, really look up and see, like, oh, what is going on? Like, you really.
181 00:28:26.410 ⇒ 00:28:27.150 Samuel Roberts: Yeah.
182 00:28:27.450 ⇒ 00:28:33.850 Ruixi Wen: stay very up-to-date. Otherwise, like, you will be the one who actually got challenged, like…
183 00:28:33.850 ⇒ 00:28:42.870 Samuel Roberts: Yeah, exactly, exactly. And there’s things… every day there’s a new model, or a new tool, or, you know, someone’s using something in a different way, or a new,
184 00:28:43.300 ⇒ 00:28:50.539 Samuel Roberts: strategy for doing stuff. Staying up to date with that is… is… is tough, because it’s fast, but it’s exciting, too, so… yeah.
185 00:28:50.540 ⇒ 00:28:51.770 Ruixi Wen: Yes, yes.
186 00:28:51.970 ⇒ 00:29:02.130 Samuel Roberts: Alright, well, I think we’re getting close to time here, so I just wanted to make sure we’re not going too over for respecting both our schedules, but, any other things I can help you with, question-wise?
187 00:29:02.130 ⇒ 00:29:19.369 Ruixi Wen: Yeah, yeah, I think that’s, like, the number, like, top two questions on, like, having online, and I think how you explained to me, like, perfectly, but yeah, like, I think there’s just, like, so much things I’m curious about, and definitely, like, hope, like, we have a chance, like, to, to, like, learn more about everything, like, down the line.
188 00:29:19.900 ⇒ 00:29:20.560 Samuel Roberts: Great.
189 00:29:21.190 ⇒ 00:29:23.540 Ruixi Wen: Yeah, thank you so much for the time, Samuel.
190 00:29:23.540 ⇒ 00:29:35.859 Samuel Roberts: Yeah, thank you as well for making time, and yeah, I think, I’ll take this back and, you know, reconvene with people, and you’ll hear back one way or another. I don’t know exactly, I’m actually…
191 00:29:35.900 ⇒ 00:29:48.560 Samuel Roberts: out of office at the end of this week, but I’m sure things will be moving. So look… look to hear back, probably sometime in the next week about something, one way or the other. Okay. Whether it’s the next step or what, but, yeah.
192 00:29:48.760 ⇒ 00:29:50.060 Samuel Roberts: Thank you for taking the time.
193 00:29:50.330 ⇒ 00:29:52.000 Ruixi Wen: Thank you so much, Samuel, yeah.
194 00:29:52.000 ⇒ 00:29:53.109 Samuel Roberts: Alright, yeah.
195 00:29:53.420 ⇒ 00:29:54.930 Ruixi Wen: Nice meeting you today.
196 00:29:54.930 ⇒ 00:29:56.589 Samuel Roberts: You as well. Yep, buh-bye.