Meeting Title: Brainforge Interview w- Sam Date: 2026-05-04 Meeting participants: Samuel Roberts, Cole Carter
WEBVTT
1 00:00:37.440 ⇒ 00:00:38.490 Samuel Roberts: Hello.
2 00:00:39.960 ⇒ 00:00:41.219 Samuel Roberts: You hear me alright?
3 00:00:41.560 ⇒ 00:00:43.709 Cole Carter: I can hear you. Hey, how’s it going?
4 00:00:43.980 ⇒ 00:00:45.380 Samuel Roberts: Good, good, how about yourself?
5 00:00:45.700 ⇒ 00:00:47.610 Cole Carter: I’m doing well… for a Monday.
6 00:00:47.610 ⇒ 00:00:48.400 Samuel Roberts: Yeah.
7 00:00:48.400 ⇒ 00:00:51.969 Cole Carter: I’m just coming off of a… Sinus infection, so…
8 00:00:51.970 ⇒ 00:00:53.479 Samuel Roberts: Oh, that’s never fun.
9 00:00:53.480 ⇒ 00:00:54.950 Cole Carter: Goes a little bit better than last week.
10 00:00:55.140 ⇒ 00:00:57.029 Samuel Roberts: Oh, yeah, was it, like, all last week?
11 00:00:57.320 ⇒ 00:00:59.430 Cole Carter: Yep, the whole time, so I was like…
12 00:01:00.010 ⇒ 00:01:07.029 Cole Carter: it was okay in the morning, and then by, like, 3PM, just the congestion and the headache kind of just, like, turned my brain off.
13 00:01:07.290 ⇒ 00:01:11.800 Cole Carter: So I worked through it, but it was, like, 6-hour days instead of…
14 00:01:11.800 ⇒ 00:01:12.430 Samuel Roberts: Yeah.
15 00:01:12.430 ⇒ 00:01:13.070 Cole Carter: Cool thing.
16 00:01:13.230 ⇒ 00:01:13.710 Cole Carter: Sounds good.
17 00:01:13.710 ⇒ 00:01:14.839 Samuel Roberts: I’m glad you’re feeling better.
18 00:01:15.460 ⇒ 00:01:16.020 Cole Carter: Thank you.
19 00:01:16.430 ⇒ 00:01:24.609 Samuel Roberts: Yeah, well, okay, so, the way this goes is a quick, you know, thing. I’ll introduce myself, ask you to introduce yourself,
20 00:01:24.990 ⇒ 00:01:37.259 Samuel Roberts: I’ve got some questions after that. I like to get about halfway, and then leave time for you to ask questions so we don’t get to the end and have no time for that. And then, yeah, we’ll just kind of chat from there.
21 00:01:37.470 ⇒ 00:01:38.400 Samuel Roberts: Does that sound good?
22 00:01:38.630 ⇒ 00:01:39.310 Cole Carter: Yep.
23 00:01:39.310 ⇒ 00:01:47.050 Samuel Roberts: Alright, cool. So, yeah, my name is Sam Roberts. I’m the AI tech lead here at Brainforge. Been here since July,
24 00:01:47.260 ⇒ 00:02:05.490 Samuel Roberts: my career thus far was a lot of startups, web tech, more recently, obviously, a lot of AI stuff. But, that’s kind of my, you know, I come from, like, the product world, so, that’s sort of where I’ve been. And then, yeah, I think that’s… that kind of covers me real quick.
25 00:02:05.950 ⇒ 00:02:06.760 Samuel Roberts: Go ahead.
26 00:02:06.980 ⇒ 00:02:10.579 Cole Carter: Cool. Yeah, I’m Cole, last name Carter.
27 00:02:10.690 ⇒ 00:02:16.989 Cole Carter: I went to UT Austin, I studied economics, philosophy, and data science.
28 00:02:17.260 ⇒ 00:02:17.790 Samuel Roberts: Oh, very cool.
29 00:02:18.670 ⇒ 00:02:29.019 Cole Carter: I was… yeah, I came out of school May 2025, so it’s been about a year. Okay. I was fortunate to get a job right out of school, which some of my friends haven’t been able to, which…
30 00:02:29.020 ⇒ 00:02:29.600 Samuel Roberts: I bet.
31 00:02:29.600 ⇒ 00:02:31.010 Cole Carter: From a crappy market out there.
32 00:02:31.140 ⇒ 00:02:32.380 Cole Carter: But,
33 00:02:32.980 ⇒ 00:02:44.590 Cole Carter: they hired me on as a data engineer, and my eventual kind of idea was to pivot into ML or AI at some point. That’s sort of what I studied and what I’ve always been pretty passionate about.
34 00:02:44.880 ⇒ 00:02:51.640 Cole Carter: Learning about… And then, just a couple months ago, I was promoted to AI engineer.
35 00:02:51.830 ⇒ 00:02:56.119 Cole Carter: And senior AI engineer, but it’s, like, startup senior, you know?
36 00:02:56.120 ⇒ 00:02:57.949 Samuel Roberts: I understand that completely.
37 00:02:58.560 ⇒ 00:03:06.109 Samuel Roberts: I’ve been CTO of many things even when I was right out of school, so I, yeah, I get that. Titles don’t mean much when it’s startup stuff in my mind, so I understand.
38 00:03:06.560 ⇒ 00:03:07.800 Cole Carter: But,
39 00:03:08.380 ⇒ 00:03:24.549 Cole Carter: I’ve sat on the product team the whole time, so at my company right now, it’s a newswire company, kind of like AP or Reuters, but smaller, and they serve brand content for free to publishers. So the idea is, like, they…
40 00:03:25.000 ⇒ 00:03:32.859 Cole Carter: have these articles that they source from brands, and then they put them in front of publishers like Houston Chronicle or whoever, who picked them up for free.
41 00:03:33.100 ⇒ 00:03:39.290 Cole Carter: But… For the brands, they needed somebody who… had…
42 00:03:39.860 ⇒ 00:03:43.689 Cole Carter: Data science experience, because they wanted someone who could…
43 00:03:44.250 ⇒ 00:03:47.739 Cole Carter: basically own perspectives on metrics, so that’s what I’ve been.
44 00:03:47.740 ⇒ 00:03:48.120 Samuel Roberts: Okay.
45 00:03:48.120 ⇒ 00:03:54.220 Cole Carter: The last year, coming up with different metrics to show value, basically, and implementing those into the portal.
46 00:03:54.420 ⇒ 00:03:54.970 Cole Carter: Because…
47 00:03:54.970 ⇒ 00:03:55.330 Samuel Roberts: Okay.
48 00:03:55.330 ⇒ 00:04:03.580 Cole Carter: I sit on the product team, so anything that has to do with… client-facing… Build-outs, platform, web stuff.
49 00:04:03.870 ⇒ 00:04:05.130 Cole Carter: I’m in there.
50 00:04:05.310 ⇒ 00:04:09.440 Cole Carter: But I mostly do, like, Python kind of backend stuff.
51 00:04:09.760 ⇒ 00:04:10.150 Samuel Roberts: Okay.
52 00:04:10.150 ⇒ 00:04:13.580 Cole Carter: The database, data management,
53 00:04:13.790 ⇒ 00:04:22.390 Cole Carter: Creating views for the metrics, coming up with the metrics, and then anything AI product is me designing and building it.
54 00:04:23.070 ⇒ 00:04:30.179 Cole Carter: One example of that, probably the biggest example that I have is, we did a recommendation system.
55 00:04:30.420 ⇒ 00:04:31.960 Cole Carter: Where we look at…
56 00:04:32.360 ⇒ 00:04:45.099 Cole Carter: How the articles are doing across the newswire, and surface different metrics, convert into advanced stats, feed that into a different… a couple different models that spit out
57 00:04:45.240 ⇒ 00:05:02.669 Cole Carter: different tables, and then I take those tables, format them into Markdown, and then give it to an AI to sort of, like, parse through, be like, hey, you guys could be hitting this money story angle, or this transportation story angle, based on, you know, recent trends, or even what day of the week, or what events are coming up.
58 00:05:02.940 ⇒ 00:05:06.910 Cole Carter: And then, right now, we’re working on…
59 00:05:07.850 ⇒ 00:05:10.409 Cole Carter: A project where we’re trying to figure out
60 00:05:10.980 ⇒ 00:05:13.250 Cole Carter: how well clients show up in LLMs.
61 00:05:13.740 ⇒ 00:05:14.650 Samuel Roberts: Okay.
62 00:05:14.650 ⇒ 00:05:22.260 Cole Carter: pivot in the, in sort of the news and journalism industry and SEO industry is, are people showing up in LLMs? You know, if somebody searches…
63 00:05:22.380 ⇒ 00:05:24.810 Cole Carter: what’s the best water bottle? Does Yeti pop up, you know?
64 00:05:24.810 ⇒ 00:05:25.710 Samuel Roberts: Right, right.
65 00:05:25.710 ⇒ 00:05:28.389 Cole Carter: So, working on building out the platform to…
66 00:05:28.540 ⇒ 00:05:33.420 Cole Carter: First of all, get a bunch of prompts for all of our clients, send them to all these different…
67 00:05:33.590 ⇒ 00:05:40.109 Cole Carter: LLMs, Claw, Gemini, whoever, get back the responses, do sentiment analysis on the responses.
68 00:05:40.110 ⇒ 00:05:40.640 Samuel Roberts: sequels.
69 00:05:40.640 ⇒ 00:05:50.039 Cole Carter: pull out brand mentions, you know, if Yeti was mentioned in the response at all, and then citations, which is our business’s biggest prop, which is, like.
70 00:05:50.040 ⇒ 00:05:50.580 Samuel Roberts: Right?
71 00:05:50.580 ⇒ 00:06:02.690 Cole Carter: Is there a HoustonChronicle.com slash your brand’s article, citation sourced for the… for the thing, and then coming up with all the summary stats that build all that out, aggregate all that, and then serve that to clients in the portal.
72 00:06:03.680 ⇒ 00:06:11.909 Samuel Roberts: Cool, cool. That’s… yeah, that sounds really neat. Yeah, I hadn’t thought so much about how SEO is changing. Like, I’ve seen articles, and I hadn’t really thought about…
73 00:06:12.380 ⇒ 00:06:17.149 Samuel Roberts: the… testing all that against the models and the training and stuff, that’s really interesting. Cool, cool.
74 00:06:18.170 ⇒ 00:06:28.439 Samuel Roberts: Great. Okay, yeah, so you actually covered a bunch of stuff I already had questions about, a lot of projects and things that way, so, maybe we’ll dig in a little bit more to certain… a few things, but let me just adjust this a little bit.
75 00:06:28.620 ⇒ 00:06:32.740 Samuel Roberts: Okay, so we kind of covered, you know, LM-based feature, you shipped…
76 00:06:33.190 ⇒ 00:06:45.660 Samuel Roberts: So, okay, so let’s dig in a little bit more then. Like, where in the AI stack have you spent the most time, like, building versus, like, experimenting? You know, so… and then, you know, whatever specific examples help with that is, you know, good, too.
77 00:06:46.530 ⇒ 00:06:52.389 Cole Carter: Where in the AI stack do you mean, as far as, like, foundation models, or do you mean, like, machine learning type of stuff?
78 00:06:52.390 ⇒ 00:06:58.690 Samuel Roberts: No, I mean, I mean, so, like, LLM in general, like, you know, if you’re building an LLM-based tool, like, where in that stack do you feel most…
79 00:06:58.690 ⇒ 00:06:59.119 Cole Carter: You’ve got you.
80 00:06:59.120 ⇒ 00:07:02.669 Samuel Roberts: Yeah, yeah, I’m not… yeah, I mean, we don’t have to go way back, kind of thing, yeah.
81 00:07:02.800 ⇒ 00:07:06.380 Cole Carter: Well, I use a lot of Claude code, pretty much exclusively.
82 00:07:06.860 ⇒ 00:07:11.560 Cole Carter: It started out maybe a year ago, me, like, tinkering with it.
83 00:07:11.710 ⇒ 00:07:12.530 Cole Carter: like…
84 00:07:12.650 ⇒ 00:07:21.279 Cole Carter: I don’t know, 30 minutes per day, and, like, doing a lot of copy-pasting and revision, and over the past year, it’s gotten to the point where I’m probably using clawed code for about…
85 00:07:21.640 ⇒ 00:07:24.060 Cole Carter: Like, 4 hours a day, I would say.
86 00:07:24.060 ⇒ 00:07:26.119 Samuel Roberts: Right? No, I get it.
87 00:07:26.120 ⇒ 00:07:26.530 Cole Carter: Yeah.
88 00:07:26.530 ⇒ 00:07:27.270 Samuel Roberts: Yeah, yeah, totally.
89 00:07:28.520 ⇒ 00:07:29.250 Cole Carter: Yep.
90 00:07:29.550 ⇒ 00:07:31.230 Samuel Roberts: Alright, cool, cool.
91 00:07:31.590 ⇒ 00:07:40.950 Samuel Roberts: And so then, so you mentioned Python more back-end focus, but, are you doing front-end, like, are you doing JavaScript, TypeScript stuff at all, or where does that fall?
92 00:07:40.950 ⇒ 00:07:42.939 Cole Carter: No, I don’t do any front-end.
93 00:07:42.940 ⇒ 00:07:43.350 Samuel Roberts: Okay.
94 00:07:43.350 ⇒ 00:07:46.130 Cole Carter: Do you have a, like, a specific web dev team.
95 00:07:46.130 ⇒ 00:07:46.770 Samuel Roberts: Sure, okay.
96 00:07:46.930 ⇒ 00:07:51.780 Cole Carter: Python stuff, and then I manage all of our Postgres databases.
97 00:07:51.780 ⇒ 00:07:53.230 Samuel Roberts: Okay, okay, cool, cool.
98 00:07:53.230 ⇒ 00:07:55.380 Cole Carter: SQL or Python stuff.
99 00:07:55.380 ⇒ 00:07:55.700 Samuel Roberts: Yeah.
100 00:07:55.700 ⇒ 00:07:58.390 Cole Carter: Get the job done with that, then it’s in my wheelhouse.
101 00:07:58.590 ⇒ 00:08:00.839 Samuel Roberts: Cool, alright, cool, cool, that’s… that’s good.
102 00:08:00.990 ⇒ 00:08:07.070 Samuel Roberts: Alright, let’s talk a little bit about, non-technical stakeholders.
103 00:08:07.170 ⇒ 00:08:14.670 Samuel Roberts: So, you know, lots of news out there, AI, LLMs, I mean, like, people are hearing all kinds of things and may not be…
104 00:08:14.980 ⇒ 00:08:18.869 Samuel Roberts: fully, digesting what it all means.
105 00:08:19.110 ⇒ 00:08:27.400 Samuel Roberts: Is… how do you think about communicating the limitations of some of this technology to someone who’s a non-technical stakeholder, for example?
106 00:08:28.900 ⇒ 00:08:29.999 Cole Carter: That’s a good question.
107 00:08:30.860 ⇒ 00:08:36.900 Cole Carter: I think… can probably frame that in two different ways, both externally and internally.
108 00:08:36.900 ⇒ 00:08:37.770 Samuel Roberts: Yeah, definitely.
109 00:08:37.770 ⇒ 00:08:45.049 Cole Carter: Starting with internally, I think it’s a little bit easier, because you can be a little more direct, and just, you know, explaining things, and…
110 00:08:45.210 ⇒ 00:08:54.640 Cole Carter: Not necessarily calling your coworker a dumbass or anything like that, but, like, you know, just putting it down, like, hey, to get this, you need to understand that
111 00:08:55.140 ⇒ 00:08:57.420 Cole Carter: An LLM is essentially an autocomplete.
112 00:08:57.580 ⇒ 00:09:06.429 Cole Carter: algorithm. It takes, you know, the capital of France is, and then fills in Paris, and does that a bunch of times, so… understanding
113 00:09:07.280 ⇒ 00:09:14.360 Cole Carter: I guess how to explain to them those concepts, I think, is a little bit easier than external. For external.
114 00:09:14.990 ⇒ 00:09:18.050 Cole Carter: I would say the hardest part is to convey
115 00:09:19.780 ⇒ 00:09:27.490 Cole Carter: Information security, I would say, is a common thing that I see a lot. We have a lot of clients, especially in, like, the financial sector, who…
116 00:09:28.090 ⇒ 00:09:31.749 Cole Carter: Are very cagey about any data related to them being uploaded.
117 00:09:32.600 ⇒ 00:09:38.899 Cole Carter: So, explaining, I guess… how the whole LLM cycle works, and explaining our products.
118 00:09:39.090 ⇒ 00:09:45.920 Cole Carter: the entire LLM section of those products is… is pretty key. So what we’ve done in the past is…
119 00:09:46.270 ⇒ 00:09:48.619 Cole Carter: Basically, break it down.
120 00:09:48.960 ⇒ 00:09:52.800 Cole Carter: just the product. We do a lot of marketing blogs, so I think that helps, so we can just point.
121 00:09:52.800 ⇒ 00:09:53.130 Samuel Roberts: Sure.
122 00:09:53.130 ⇒ 00:09:54.270 Cole Carter: But…
123 00:09:54.270 ⇒ 00:09:54.970 Samuel Roberts: Great.
124 00:09:55.130 ⇒ 00:10:07.729 Cole Carter: Showing them exactly, okay, we have this data that comes from publishers, Wall Street Journal, or MSN.com, or whatever. It’s not used specifically, it’s just it’s metadata. It goes into LLMs.
125 00:10:08.030 ⇒ 00:10:08.940 Cole Carter: It…
126 00:10:09.270 ⇒ 00:10:17.739 Cole Carter: becomes autocomplete, it lives on ChatGPT servers, it lives on cloud servers or whatever, and then we get the responses back, and we use those
127 00:10:18.060 ⇒ 00:10:23.530 Cole Carter: on our end for Portal products, and also for…
128 00:10:23.780 ⇒ 00:10:32.530 Cole Carter: some internal tooling. We do use it for some internal tooling, and I feel like that usually takes the edge off of those clients, because they’re kind of concerned that we’re, like.
129 00:10:33.210 ⇒ 00:10:36.699 Cole Carter: Somehow… uploading their own data that they have.
130 00:10:36.700 ⇒ 00:10:37.589 Samuel Roberts: Right, right.
131 00:10:37.590 ⇒ 00:10:39.380 Cole Carter: And a lot of times, it’s just us
132 00:10:39.480 ⇒ 00:10:46.220 Cole Carter: transforming their data, or their metadata, or what we have on them, and using that for LLM, so that usually clears it up.
133 00:10:47.350 ⇒ 00:10:50.950 Cole Carter: Other common questions from external stakeholders include, like.
134 00:10:51.980 ⇒ 00:10:55.670 Cole Carter: which LLM providers we use. Some of them want us to be using.
135 00:10:55.810 ⇒ 00:10:57.469 Cole Carter: Cloud and ChatGPT, because they have…
136 00:10:57.980 ⇒ 00:11:01.389 Cole Carter: since Stacker’s kind of prop is a bit of a marketing kind of prop.
137 00:11:01.390 ⇒ 00:11:01.710 Samuel Roberts: Yeah.
138 00:11:01.710 ⇒ 00:11:09.130 Cole Carter: So, we do have some clients who are like, we want you guys to be using Claude and Gemini, because we want to be in the training data for those.
139 00:11:09.130 ⇒ 00:11:09.600 Samuel Roberts: Right.
140 00:11:09.600 ⇒ 00:11:10.560 Cole Carter: LLMs.
141 00:11:10.780 ⇒ 00:11:11.919 Samuel Roberts: Right, right, that makes sense.
142 00:11:11.920 ⇒ 00:11:13.730 Cole Carter: We’ve had questions where it’s like.
143 00:11:13.960 ⇒ 00:11:18.730 Cole Carter: can we be in ChatGPT? And we’re like, well, our workflow is built on Claude, so I’m sorry, and also.
144 00:11:18.730 ⇒ 00:11:19.229 Samuel Roberts: Yeah, I don’.
145 00:11:19.230 ⇒ 00:11:25.339 Cole Carter: I don’t think that… this drop in the bucket, in terms of the ChatGBT servers, is really gonna…
146 00:11:25.460 ⇒ 00:11:30.450 Cole Carter: move the needle for Yeti showing up in… Responses.
147 00:11:30.450 ⇒ 00:11:32.809 Samuel Roberts: Right, right. That makes sense, that makes sense.
148 00:11:33.140 ⇒ 00:11:41.859 Samuel Roberts: Cool, cool, yeah. Interesting. Yeah, I like… I like this angle. I hadn’t thought about the marketing side as much as I have, you know, other… other elements, so that’s… that’s really neat.
149 00:11:41.860 ⇒ 00:11:50.409 Cole Carter: Yeah, it’s really exploding, I guess, in the marketing industry, and yeah. Stacker’s kind of positioned well for it, but
150 00:11:51.350 ⇒ 00:11:55.270 Cole Carter: The biggest thing is just… Coming up with all the metrics for this stuff.
151 00:11:55.270 ⇒ 00:11:55.850 Samuel Roberts: Right.
152 00:11:55.850 ⇒ 00:11:59.019 Cole Carter: Kind of, like, mind-boggling, because it’s all new territory.
153 00:12:00.450 ⇒ 00:12:11.969 Samuel Roberts: Yeah, everything’s changing so much, like, just because of these LLMs and the tooling and everything now. That’s… yeah. Oh, I like this. Okay, cool, cool. Let’s see what else I can, kind of segue to there.
154 00:12:14.320 ⇒ 00:12:27.309 Samuel Roberts: So, okay, I mentioned things changing quickly, you know, we talked about how that… so is there… have there been any, like, trends that you were maybe initially excited about in AIL and stuff that eventually you decided not to adopt after
155 00:12:27.740 ⇒ 00:12:28.859 Samuel Roberts: learning more.
156 00:12:30.590 ⇒ 00:12:31.240 Cole Carter: Yeah.
157 00:12:33.350 ⇒ 00:12:38.439 Cole Carter: at least in my experience, I can talk about work, and I can talk about my personal experience in my personal.
158 00:12:38.440 ⇒ 00:12:38.800 Samuel Roberts: Yeah.
159 00:12:38.800 ⇒ 00:12:42.069 Cole Carter: open claw is something that I am really all about.
160 00:12:42.660 ⇒ 00:12:44.489 Cole Carter: Hot on, and I was really hot on.
161 00:12:44.650 ⇒ 00:12:50.490 Cole Carter: I don’t really have an infinite wallet to really just, like, keep it on all the time, or… Yeah. I’m not willing to…
162 00:12:50.740 ⇒ 00:12:54.870 Cole Carter: Violate terms of service for my subscription plans.
163 00:12:54.870 ⇒ 00:12:55.320 Samuel Roberts: Totally.
164 00:12:55.320 ⇒ 00:13:02.240 Cole Carter: Is it… But it is something that I think about a lot, and I’m pretty bullish on long-term.
165 00:13:02.620 ⇒ 00:13:06.379 Cole Carter: Kind of just the idea of… Autonomous agents interact.
166 00:13:08.250 ⇒ 00:13:11.799 Cole Carter: Having workflows attached to them, even inventing workflows.
167 00:13:11.800 ⇒ 00:13:12.320 Samuel Roberts: Yeah.
168 00:13:12.320 ⇒ 00:13:17.090 Cole Carter: there’s, there’s, like, a scale of AI intelligence that I think…
169 00:13:17.310 ⇒ 00:13:20.139 Cole Carter: OpenAI maybe put out a few years ago, where they took.
170 00:13:20.140 ⇒ 00:13:20.770 Samuel Roberts: promote.
171 00:13:20.770 ⇒ 00:13:27.680 Cole Carter: Chatbot level, rolling up to the reasoning level, which now we see with thinking tokens, rolling up to the agent level.
172 00:13:28.300 ⇒ 00:13:30.400 Cole Carter: Which is kind of like OpenClaw or Cloud.
173 00:13:30.400 ⇒ 00:13:31.340 Samuel Roberts: You’re right.
174 00:13:31.340 ⇒ 00:13:36.800 Cole Carter: And then, past that, the organization level, where you have, like, hundreds of autonomous agents.
175 00:13:36.800 ⇒ 00:13:37.649 Samuel Roberts: Right, right.
176 00:13:37.650 ⇒ 00:13:40.919 Cole Carter: So, I’m… I’m really curious, at each… at each of those…
177 00:13:41.090 ⇒ 00:13:44.160 Cole Carter: kind of precipices, I’ve always been curious, like, what is…
178 00:13:44.440 ⇒ 00:13:45.990 Cole Carter: Going on today, that leaves us.
179 00:13:45.990 ⇒ 00:13:46.310 Samuel Roberts: Yeah.
180 00:13:46.310 ⇒ 00:13:48.599 Cole Carter: Precipice, and open claw, to me, is like…
181 00:13:48.860 ⇒ 00:13:50.980 Cole Carter: Close to getting to that next.
182 00:13:51.120 ⇒ 00:13:54.470 Cole Carter: precipice. Definitely, yeah. But it’s not something that…
183 00:13:55.250 ⇒ 00:13:59.440 Cole Carter: I really have the money to invest in, totally, but I like to think about it, and I like to.
184 00:13:59.440 ⇒ 00:14:15.890 Samuel Roberts: Yeah, yeah. Yeah, I played around a little bit, especially when it was first blowing up, because I needed to know, because I saw so much about it, but I was a little nervous because of all the security issues and everything, and I just put it on an isolated VM and barely gave it access to anything, and saw, like, it can do a lot, but I was scared to…
185 00:14:15.950 ⇒ 00:14:32.899 Samuel Roberts: let it at that point. And then I read about all the… how many tokens people were burning through, and I was like, oh, maybe I don’t need that right now. Yeah, I feel that, I feel that. Yeah. Cool, cool. Let’s see, okay, we’re almost halfway. I got one… one more thing I want to at least jump into, but…
186 00:14:33.010 ⇒ 00:14:40.820 Samuel Roberts: What is, okay, this is the last one we’ll do. If you had 6 months, no obligations.
187 00:14:41.210 ⇒ 00:14:43.089 Samuel Roberts: What would you work on?
188 00:14:45.090 ⇒ 00:14:46.749 Cole Carter: Definitely artificial intelligence.
189 00:14:46.750 ⇒ 00:14:51.850 Samuel Roberts: Yeah, okay, yeah, I mean, we can dig into that a little bit, maybe, but yeah.
190 00:14:51.850 ⇒ 00:14:52.240 Cole Carter: For sure.
191 00:14:52.240 ⇒ 00:14:53.160 Samuel Roberts: Yeah, totally.
192 00:14:53.310 ⇒ 00:14:56.030 Cole Carter: I really like…
193 00:14:56.830 ⇒ 00:15:02.310 Cole Carter: Well, I studied economics in college, and I really like economics. I also like philosophy and ethics, so I think.
194 00:15:02.310 ⇒ 00:15:05.250 Samuel Roberts: I’m… yeah, I wanna hear more about that too, that’s cool, yeah.
195 00:15:05.250 ⇒ 00:15:09.979 Cole Carter: I think both of those two angles kind of inform
196 00:15:10.430 ⇒ 00:15:16.780 Cole Carter: like, what I’m most deeply interested in. Like, I really love AI, and I love… philosophically AI.
197 00:15:17.150 ⇒ 00:15:22.469 Cole Carter: and also using AI as a tool to solve questions in both those domains, I think, is really interesting.
198 00:15:22.700 ⇒ 00:15:23.510 Samuel Roberts: So, from, like, the.
199 00:15:23.510 ⇒ 00:15:29.210 Cole Carter: economics angle, The rate at which…
200 00:15:29.350 ⇒ 00:15:39.590 Cole Carter: information can be processed with AI is just vastly higher, and the way that that information processing power is distributed is much more democratic than it ever has been.
201 00:15:39.730 ⇒ 00:15:49.760 Cole Carter: Anybody can, you know, download booklets and booklets of FRED data, and come up with models, and things like that, and, you know, audit different
202 00:15:50.070 ⇒ 00:15:54.799 Cole Carter: Like, are there really 150,000 more jobs this month or not? You know, people have a lot more.
203 00:15:54.800 ⇒ 00:15:55.460 Samuel Roberts: Right.
204 00:15:55.920 ⇒ 00:16:05.810 Cole Carter: themselves. So, something that I’ve been thinking a lot about is… Having, kind of, independent Consumer-driven audits of…
205 00:16:06.340 ⇒ 00:16:08.080 Cole Carter: economic information.
206 00:16:08.880 ⇒ 00:16:17.650 Cole Carter: And, like, jobs reports and stuff like that. I think now there’s a lot more, especially now that so many of the, like, BLS has been gutted and stuff like that.
207 00:16:17.880 ⇒ 00:16:19.119 Cole Carter: Now there’s a lot.
208 00:16:19.220 ⇒ 00:16:24.809 Cole Carter: more leeway for… I guess, private actors to take on those roles.
209 00:16:24.990 ⇒ 00:16:25.390 Samuel Roberts: And…
210 00:16:25.390 ⇒ 00:16:30.990 Cole Carter: especially with AI, I think… You know, pretty much anybody can do it, so building models to…
211 00:16:31.340 ⇒ 00:16:39.600 Cole Carter: to be able to assess stuff like jobs reports is something that I’ve been interested in. Other interests include, like, prediction markets.
212 00:16:39.940 ⇒ 00:16:43.949 Cole Carter: I’m not really, like, a gambler person, but…
213 00:16:44.250 ⇒ 00:16:54.449 Cole Carter: the rate… the way that all of them can be interacted with via API, and the fact that, like, 70% of traffic is bots, just makes me feel like…
214 00:16:56.230 ⇒ 00:16:59.599 Cole Carter: it can’t be that hard to make another bot, you know?
215 00:16:59.600 ⇒ 00:17:00.400 Samuel Roberts: Right, right, yeah.
216 00:17:00.400 ⇒ 00:17:05.479 Cole Carter: And, you know, come up with theories that way, and analyze, like, mass book data, stuff like that.
217 00:17:06.470 ⇒ 00:17:10.100 Cole Carter: And then on the philosophy side, definitely ethics. I’ve been…
218 00:17:10.800 ⇒ 00:17:13.360 Cole Carter: That was my main focus.
219 00:17:13.550 ⇒ 00:17:13.859 Samuel Roberts: Okay.
220 00:17:13.869 ⇒ 00:17:20.759 Cole Carter: philosophy, studying philosophy, was both consciousness and ethics, and both of those kind of line up with artificial intelligence in a lot of ways.
221 00:17:21.109 ⇒ 00:17:27.369 Cole Carter: So… Something… something that… where that actually relates on… on product work, I guess?
222 00:17:28.200 ⇒ 00:17:35.459 Cole Carter: And my company is sort of the philosophy of how do we improve an AI product, and how do we evaluate it?
223 00:17:35.690 ⇒ 00:17:43.889 Cole Carter: And we’ve taken different approaches, but something that I like to do is keep the human in the loop as much as possible, until you can finally
224 00:17:44.080 ⇒ 00:17:50.479 Cole Carter: create, or… or, automate that step and create an LLM as a judge kind of phase.
225 00:17:52.210 ⇒ 00:18:02.690 Cole Carter: that’s not… that’s not super, like, ethics questions, but it’s just one way that philosophy intersects with product work. But as far as ethics questions, I definitely think about it a lot.
226 00:18:02.860 ⇒ 00:18:12.019 Cole Carter: And… because I have enough Like, study of neural networks and transformer architectures and stuff like that.
227 00:18:12.210 ⇒ 00:18:21.619 Cole Carter: it’s very plain to me that there’s no consciousness in LLMs. Yeah. So, I have been thinking about, like, creating a blog just to debunk that, because I know there’s so many people who are getting
228 00:18:21.980 ⇒ 00:18:25.779 Cole Carter: I guess… Psychosis, or whatever, into believing.
229 00:18:25.780 ⇒ 00:18:26.630 Samuel Roberts: Yeah, some of those articles.
230 00:18:26.630 ⇒ 00:18:34.179 Cole Carter: LLMs are, you know, they’re conscious, and they’re trying to come alive, and they’re asking their person to help them prove to the world.
231 00:18:34.460 ⇒ 00:18:37.380 Cole Carter: So I think it’s really important that…
232 00:18:37.900 ⇒ 00:18:41.809 Cole Carter: First of all, those people don’t… Get psychosed, or psychoso.
233 00:18:41.810 ⇒ 00:18:42.470 Samuel Roberts: Yeah.
234 00:18:43.010 ⇒ 00:18:45.550 Cole Carter: So yeah, having some kind of…
235 00:18:45.810 ⇒ 00:18:48.809 Cole Carter: blog about that. It was something I’ve been thinking about.
236 00:18:49.750 ⇒ 00:18:50.659 Samuel Roberts: Oh, yeah.
237 00:18:50.770 ⇒ 00:19:05.690 Samuel Roberts: Yeah, I’m with you. I studied mechanical engineering, physics, and history and philosophy of science, and then got into startups and web tech, but this is also, like, 10 years ago, so yeah, I imagine it’s a different world studying philosophy of consciousness and stuff now.
238 00:19:06.090 ⇒ 00:19:08.930 Cole Carter: There was a lot of, like, new papers to read.
239 00:19:08.930 ⇒ 00:19:09.800 Samuel Roberts: Yeah.
240 00:19:09.800 ⇒ 00:19:17.169 Cole Carter: really cool. Even, something that I really liked studying was mechanistic interpretability of LLM.
241 00:19:17.170 ⇒ 00:19:17.830 Samuel Roberts: Hmm.
242 00:19:18.030 ⇒ 00:19:21.840 Cole Carter: I don’t know if you’re familiar with that field, but it’s… Basically, where…
243 00:19:22.740 ⇒ 00:19:26.700 Cole Carter: they’ll cut an LLM in half, essentially, and look at the weights in the middle of the.
244 00:19:27.630 ⇒ 00:19:30.209 Cole Carter: Model, and then try to figure out
245 00:19:30.370 ⇒ 00:19:36.029 Cole Carter: Which weights are attributed to which, like, real-world objects or columns?
246 00:19:36.030 ⇒ 00:19:36.410 Samuel Roberts: Yeah.
247 00:19:36.980 ⇒ 00:19:40.070 Cole Carter: And… they can get an idea of…
248 00:19:40.690 ⇒ 00:19:45.869 Cole Carter: Where there’s vulnerabilities, and where there’s, like, overtuning and stuff like that.
249 00:19:45.870 ⇒ 00:19:46.330 Samuel Roberts: Yeah.
250 00:19:46.330 ⇒ 00:19:53.269 Cole Carter: There was a… there was a Claude paper where they… they did that with one of the older Claude models, and they… they found the Golden Gate Bridge.
251 00:19:53.450 ⇒ 00:19:55.480 Cole Carter: Sort of theme in the ways.
252 00:19:55.920 ⇒ 00:20:04.539 Cole Carter: just jacked it up to, like, a thousand, and then they… they tested the model some more, and pretty much every response involved something to do with San Francisco, or the Golden Gate Bridge, or Alcatraz.
253 00:20:04.540 ⇒ 00:20:13.979 Samuel Roberts: That’s real interesting. Yeah, I’ve seen a little bit about people looking at, like, internal weights, but not… I hadn’t seen that. That’s really neat. That’s very cool. All right.
254 00:20:13.980 ⇒ 00:20:17.650 Cole Carter: But I think there’s a lot of parallels between that and then philosophy of mind.
255 00:20:17.650 ⇒ 00:20:25.149 Samuel Roberts: Totally, totally. Yeah, I can see why people that may not fully understand how the tech works, or the fact that, you know, these are
256 00:20:25.250 ⇒ 00:20:35.809 Samuel Roberts: stateless and the chatbots are all kind of built around them, and, you know, I think there’s a lot there that people might not fully get, but there’s still interesting things to discuss about it, and, you know, parallels, and, you know…
257 00:20:35.870 ⇒ 00:20:47.500 Samuel Roberts: I like it. I think it’s really cool. It’s one of those things I always wish I had more time to dig into, because it was, like, my third major, and so it was always, like, one class a semester, and I haven’t really done much with it since, unfortunately, but I…
258 00:20:47.500 ⇒ 00:20:57.930 Samuel Roberts: read as much as I can, when I get time, so… That’s cool. All right, yeah. Okay, so we’re a little over halfway now, I want to make sure we have time for any questions you’ve got, so,
259 00:20:58.360 ⇒ 00:21:02.159 Samuel Roberts: Yeah, anything, role, company, whatever I can answer for you, yeah, totally.
260 00:21:02.160 ⇒ 00:21:04.539 Cole Carter: Yeah, I wrote down some, if you don’t mind me pulling up.
261 00:21:04.540 ⇒ 00:21:05.220 Samuel Roberts: Of course, yeah.
262 00:21:12.010 ⇒ 00:21:12.890 Cole Carter: Hang on…
263 00:21:14.620 ⇒ 00:21:15.439 Samuel Roberts: Yeah, no worries.
264 00:21:16.960 ⇒ 00:21:17.620 Cole Carter: Okay.
265 00:21:17.950 ⇒ 00:21:26.400 Cole Carter: Yeah, I was curious, so, I haven’t done any specifically consulting work. I’ve done… Product-facing, or sorry, client-facing.
266 00:21:26.760 ⇒ 00:21:28.080 Cole Carter: stuff, but…
267 00:21:28.190 ⇒ 00:21:34.189 Cole Carter: I was kind of curious what the day-to-day looks like in terms of interacting with clients, and then…
268 00:21:34.430 ⇒ 00:21:42.909 Cole Carter: How you guys, based on problems and, you know, triaging, like, what tech stack might be needed for this, or that client, how you guys…
269 00:21:43.390 ⇒ 00:21:47.669 Cole Carter: Would potentially give me a, like, a task to work on, or a.
270 00:21:48.060 ⇒ 00:21:48.870 Cole Carter: work on.
271 00:21:50.440 ⇒ 00:22:09.509 Samuel Roberts: Sure, sure. Yeah, so… excuse me. So yeah, we, it’s an interesting question, because I came from a much more product background, and so, you know, I’d freelanced a bit, so I had, you know, customers, I guess, clients in that sense. But yeah, consulting, you know, is another beast a little bit. We…
272 00:22:09.660 ⇒ 00:22:18.550 Samuel Roberts: try to be very open with… with, our clients. So we have people that, you know, manage the relationship and are kind of the main,
273 00:22:20.390 ⇒ 00:22:36.410 Samuel Roberts: input point, I guess? You know, so, like, they’re the people that, like, get called when something, you know, needs to happen. But we don’t necessarily restrict, you know, engineers especially, from communicating on Slack on joint channels and things like that. So, you know, we try to keep them in the loop as we’re doing things. You know, we don’t want to be that…
274 00:22:36.410 ⇒ 00:22:41.859 Samuel Roberts: Like, black box that, you know, you hire a consultant, ask them to do something, they come back.
275 00:22:41.890 ⇒ 00:22:53.749 Samuel Roberts: two weeks, three weeks, a month later, and it’s not the right direction, all these things, so we try to kind of keep that, like, almost iteration cycle that, you know, you would have for a product company.
276 00:22:53.830 ⇒ 00:23:07.980 Samuel Roberts: how we think about, you know, triaging projects and stuff, it’s been changing a little bit. You know, as we’ve been growing, you can imagine going from smaller teams to bigger teams, or more clients, bigger projects. We’ve been refining that a bit.
277 00:23:08.110 ⇒ 00:23:23.539 Samuel Roberts: You know, there’s a process all the way from, like, sales and, like, prospecting, all the way through the statement of work, through, you know, building the project and the milestones and stuff, and so, what we’ve started to do is kind of pull
278 00:23:23.640 ⇒ 00:23:33.449 Samuel Roberts: technical people a little further up into that process, so that by the time the project gets to a point when we’re planning out literal tickets and, you know, specific work.
279 00:23:33.450 ⇒ 00:23:54.110 Samuel Roberts: there’s been thought, and it’s not just, here’s a statement of work, figure out how to do it, kind of thing. And so, with that, you know, we have some tools that we lean on. I come from a more TypeScript background, so I’ve, you know, been using Mastra, I don’t know if you’ve seen any frameworks like that, TypeScript-wise, but we used some Langchain in the past, we were using N8N a bit for prototyping,
280 00:23:54.110 ⇒ 00:23:58.979 Samuel Roberts: probably a little beyond prototyping some points, and hit some real bottlenecks with that.
281 00:23:59.000 ⇒ 00:24:01.909 Samuel Roberts: And so then, the way it tends to work is…
282 00:24:02.390 ⇒ 00:24:11.239 Samuel Roberts: You know, depending on how many clients we have, how many projects are going, you may work on a few different ones, so there’s some context switching there, that is…
283 00:24:11.520 ⇒ 00:24:30.189 Samuel Roberts: manageable, I would say, you know, you’re not ping-ponging constantly. But, you know, we get tickets on the board, we kind of, as a team, figure out, you know, who’s best positioned to do this, who has the time, who’s allocated for the right number of hours, who has the context already, how do they sync with the next person who does.
284 00:24:30.300 ⇒ 00:24:32.650 Samuel Roberts: Sorry, my throat’s getting a little dry.
285 00:24:32.670 ⇒ 00:24:33.819 Cole Carter: Okay, I’m sick.
286 00:24:33.820 ⇒ 00:24:36.439 Samuel Roberts: And so, yeah,
287 00:24:37.050 ⇒ 00:24:44.750 Samuel Roberts: So yeah, I mean, it depends, you know, I think it kind of comes in, like, you know, especially if you’re coming in with a certain skill set, and it’s like, you know, Python backend database stuff is…
288 00:24:44.750 ⇒ 00:24:59.330 Samuel Roberts: you know, things we do, we kind of run the full stack when we’re developing these things, so, sometimes there’s a UI that we’re plugging into, sometimes we’re, you know, working with the data team to bring data in, because there’s a whole data side and kind of AI automation side.
289 00:24:59.330 ⇒ 00:25:14.100 Samuel Roberts: So, Brainforge was kind of started as the data consultancy, and then the AI kind of spun out of that. So, you know, we have resources, so it’s not like I’m necessarily putting together pipelines if I don’t, you know, need to be. I can be leaning on some of the data folks to help with that, and the expertise there that they have.
290 00:25:14.240 ⇒ 00:25:19.770 Samuel Roberts: And so then, from our side, you know, we’ve put together the right pipelines and stuff, we’ve done chatbots, we’ve done,
291 00:25:20.640 ⇒ 00:25:38.279 Samuel Roberts: simpler automations, I’d say, you know, people copying and pasting into Claude all the time, and like, okay, we can automate this into a, you know, multi-agent loop and things like that. And so, like, where, you know, an AI engineer gets hired would jump in would, you know, probably start on a client, and then as other things pop up, we try to…
292 00:25:38.620 ⇒ 00:25:39.390 Samuel Roberts: you know.
293 00:25:39.900 ⇒ 00:25:59.480 Samuel Roberts: be across different clients, so there’s not, you know, if someone’s out, it’s not, oh, so-and-so has all the context, we try to make sure that there’s enough, you know, in the repo, in Slack. We have a whole internal platform, that’s the other side of it, is we’ve done a lot of internal, you know, tooling, build for ingesting meetings and transcripts and case studies and things like that, that…
294 00:25:59.650 ⇒ 00:26:08.770 Samuel Roberts: you know, the context is there, and it’s not… you’re not just, oh, I have nothing, no idea what to do with this ticket, you know, the tickets get flushed out, you know, pretty well. And so.
295 00:26:09.450 ⇒ 00:26:24.039 Samuel Roberts: I think, you know, we’re looking to hire and grow that way, so I think it will maybe change a little bit, but right now, you know, people have kind of their main clients, I guess, that they’re, you know, most familiar with, but we like to make sure that people are jumping around and pulling tickets at other clients if there’s…
296 00:26:24.550 ⇒ 00:26:28.459 Samuel Roberts: the right resources for that, if that makes sense. Yeah.
297 00:26:28.890 ⇒ 00:26:33.939 Samuel Roberts: But yeah, I mean, it’s, it’s nice, because it’s different projects, it’s not, you know.
298 00:26:34.000 ⇒ 00:26:46.519 Samuel Roberts: there’s pros and cons to, like, always working on one product versus working on multiple client projects. I find it interesting because I’m not just, you know, doing a RAG pipeline that I’m tuning and refining all the time and getting, you know, really, you know, in-depth with
299 00:26:46.520 ⇒ 00:26:57.090 Samuel Roberts: But, you know, someone is doing that when it needs to happen. But, you know, we get to jump around a little bit and try different things, different tools, especially on the internal stuff, when there’s, you know, we get to play around with it a little bit.
300 00:26:57.090 ⇒ 00:27:02.649 Samuel Roberts: Kind of more bleeding-edge thing that we might not necessarily put into production for clients yet.
301 00:27:03.020 ⇒ 00:27:07.439 Samuel Roberts: So there’s kind of a lot of it that way. I mean, if you’re, you know, interested in
302 00:27:07.440 ⇒ 00:27:24.859 Samuel Roberts: more than just, like, back-end, there’s definitely will be front-end stuff, and how those tie together, and, you know, full-stack kind of vertical slices is how we like to think of it, so, you know, you don’t go a week, and it’s like, alright, we got the whole backend ready, but we can’t show anything to the client yet, because nothing’s there. So, kind of, thinking of those things, you know, more vertically than horizontally is…
303 00:27:24.860 ⇒ 00:27:40.499 Samuel Roberts: is something we like to do so that we’re all a little more, like, well-rounded when something happens. It can be like, alright, you know, Cole’s got this, he knows who to talk to for the things he might not have, but using the coding agents and the context in the repo and stuff, you know, can kind of hit the ground running that way, so…
304 00:27:40.900 ⇒ 00:27:44.560 Cole Carter: I like that a lot. Something that, obviously being…
305 00:27:44.800 ⇒ 00:27:48.280 Cole Carter: You know, having only about 2 years’ experience.
306 00:27:48.280 ⇒ 00:27:49.100 Samuel Roberts: Yeah.
307 00:27:50.080 ⇒ 00:27:58.799 Cole Carter: growth is, like, something that’s at the forefront of my mind. Like, where can I learn and get the opportunity to, you know, do some of those, like, bleeding edge kind of things, either.
308 00:27:58.800 ⇒ 00:27:59.240 Samuel Roberts: Yeah.
309 00:27:59.240 ⇒ 00:28:02.039 Cole Carter: tools, which I think is really cool, and then also…
310 00:28:02.810 ⇒ 00:28:08.750 Cole Carter: The idea of being on multiple projects and having multiple clients with different kind of tech stacks and solutions and…
311 00:28:10.530 ⇒ 00:28:12.380 Cole Carter: That, to me, is really cool, because…
312 00:28:12.990 ⇒ 00:28:15.299 Cole Carter: I would learn so fast, and especially, like.
313 00:28:15.300 ⇒ 00:28:15.690 Samuel Roberts: It’s…
314 00:28:15.690 ⇒ 00:28:16.400 Cole Carter: environment.
315 00:28:16.450 ⇒ 00:28:16.960 Samuel Roberts: Yeah.
316 00:28:16.960 ⇒ 00:28:21.960 Cole Carter: At my current role, Something that is a little bit lacking is…
317 00:28:22.250 ⇒ 00:28:28.090 Cole Carter: having, I don’t know, that guy who’s right there, who’s like, yeah, looking over your shoulder, like, hey, maybe you should do this, or try.
318 00:28:28.090 ⇒ 00:28:28.640 Samuel Roberts: Yeah.
319 00:28:28.640 ⇒ 00:28:29.320 Cole Carter: sway.
320 00:28:29.600 ⇒ 00:28:32.289 Cole Carter: So, like, I have gotten…
321 00:28:32.930 ⇒ 00:28:37.690 Cole Carter: to dip my toes in basically whatever I want to, because it is a startup, and I have so much ownership.
322 00:28:37.810 ⇒ 00:28:45.950 Cole Carter: Over my… my area, but… still, I feel like having that external push factor of, like, hey, you guys.
323 00:28:46.780 ⇒ 00:28:54.160 Cole Carter: you need to go work on this TypeScript kind of front-end type of deal. I think getting me out of my box is something that I would love, actually.
324 00:28:54.160 ⇒ 00:29:06.240 Samuel Roberts: That’s… that’s… yeah, that’s definitely a big part of it. And remote and, you know, somewhat async sometimes is something that I’ve had to deal with in the past, because I’ve basically never worked in an office, like.
325 00:29:06.690 ⇒ 00:29:09.909 Samuel Roberts: It’s always been hybrid or remote for me in the last, you know.
326 00:29:10.430 ⇒ 00:29:14.250 Samuel Roberts: 10 plus years. And so it’s something that I’ve been really…
327 00:29:14.620 ⇒ 00:29:32.659 Samuel Roberts: focused on, like, making sure that there’s, like, office hours or things. You know, we don’t necessarily have a daily stand-up, because, you know, we can kind of manage things on linear, and it’s a little async, and we have Slack, but I still want to be able to have that synchronous time where if someone has a question they need answered, you know, they know that there’s a time to do that. But also, you know, people reach out, and we chat about things on Slack, and kind of…
328 00:29:32.910 ⇒ 00:29:49.270 Samuel Roberts: trying to keep that culture of, you know, asking the right questions and growing and learning is big, because I, you know, came from the startup world. I pretty much self-taught in terms of coding, because, like I said, I studied mechanical engineering mostly, so, you know, that’s how I figured things out and learned from people, so, yeah.
329 00:29:49.270 ⇒ 00:29:58.619 Samuel Roberts: Yeah, yeah. Yeah, so I hope that, you know, covers, any other, like, question. We’re kind of at times, but I don’t want to, you know, I want to respect your time, so I didn’t want to…
330 00:29:58.620 ⇒ 00:30:00.580 Cole Carter: I mean, would you mind if I threw in another one?
331 00:30:00.580 ⇒ 00:30:03.249 Samuel Roberts: Totally, that’s… yeah, I just want to make sure that if you had a stop, yeah, by all means.
332 00:30:04.070 ⇒ 00:30:14.220 Cole Carter: I was just curious, too, you were mentioning that you came from a product background, so I’m curious what brought you to consultancy, what you like about it, and why you haven’t
333 00:30:14.640 ⇒ 00:30:16.180 Cole Carter: Back to a product team.
334 00:30:16.390 ⇒ 00:30:17.980 Samuel Roberts: Yeah, so,
335 00:30:18.350 ⇒ 00:30:34.989 Samuel Roberts: I had a few different startups that I worked on over the years with more or less success, some less technical than others. So I had a number of different roles, you know, CTO of a two-person company kind of thing, where we’re just coding and, you know, my friends raising money and selling, and that’s…
336 00:30:35.270 ⇒ 00:30:49.400 Samuel Roberts: great sometimes. It’s also hell sometimes. I’m also, you know, 10 years past my degree, or more than 10 years out, so I, you know, coming out of college and, learning a lot that way is great, and the, you know,
337 00:30:49.470 ⇒ 00:30:59.139 Samuel Roberts: upside can be big, but the downside could also be, you know, you’re out of a job when the funding runs out, and if you haven’t made sales, kind of thing. So I’ve gone through a few different stages like that,
338 00:30:59.650 ⇒ 00:31:04.089 Samuel Roberts: It’s been… good learning experience, and I… I’m not necessarily…
339 00:31:04.350 ⇒ 00:31:08.960 Samuel Roberts: done with the startup game, I would say. But, as I’ve…
340 00:31:09.130 ⇒ 00:31:14.100 Samuel Roberts: you know, gotten older, and married, and house, and kids, it’s a little bit…
341 00:31:14.550 ⇒ 00:31:30.380 Samuel Roberts: less appealing on its face, maybe. I still like product… I mean, you know, I think the consulting thing kind of came from me doing more freelance work, for a bit, and so, you know, after my last company, I was in London for a couple years for my wife’s job, and so I worked with a guy there.
342 00:31:30.380 ⇒ 00:31:43.359 Samuel Roberts: I went to London saying, I’m just gonna get a coding job so we can travel, and I don’t have to start a company, and ended up getting sucked into another startup, because it’s kind of where I end up most of the time. And so, after we left London and kind of spun that down.
343 00:31:43.360 ⇒ 00:31:51.840 Samuel Roberts: freelancing a bit, playing with tools. I liked the idea of working on different things and playing with new technology and not
344 00:31:51.840 ⇒ 00:31:53.290 Samuel Roberts: Necessarily going really
345 00:31:53.580 ⇒ 00:32:02.499 Samuel Roberts: deep on one main thing, and worrying about all the other stuff that comes along with that, company-wise, startup-wise. I think…
346 00:32:03.420 ⇒ 00:32:20.710 Samuel Roberts: the thing about consulting that was new to me was a lot of dealing with the clients and the context switching and things like that that is manageable. You know, it’s not… it’s similar to freelancing, but I might have more projects that I’m, you know, working on or helping, you know, more junior engineers with in a way that I might not have had if I was just freelancing.
347 00:32:21.050 ⇒ 00:32:35.189 Samuel Roberts: But I think, you know, there’s definitely something in me that I also miss, you know, going real deep with, you know, my partner, who is more of a designer, and we could really just spend days thinking things through, testing things, iterating, you know. It’s… there’s something…
348 00:32:35.190 ⇒ 00:32:46.399 Samuel Roberts: it’s not like that’s not here, I would say, but it’s a much different cycle than, like, a big, like, okay, we gotta revamp this piece and see how that affects users and things. Like, we’re doing that on a smaller scale, and I think that’s also
349 00:32:46.440 ⇒ 00:33:01.380 Samuel Roberts: things are different now. When I was working on that, ChatGPT was just coming out, it wasn’t like we had all these agent decoding agents and stuff, so we can move at different speeds on multiple projects, which is different. So I don’t know how that would necessarily change if I were back in a product role, but I like,
350 00:33:01.380 ⇒ 00:33:08.039 Samuel Roberts: to some degree, I like the context switching, because it keeps me thinking about different things and growing and learning still. You know, testing…
351 00:33:08.040 ⇒ 00:33:22.969 Samuel Roberts: some model on our RAG pipeline, and then learning we have to switch to, you know, GCP, because that’s where the client has all their infrastructure and stuff, and so then we gotta use Gemini, and all of a sudden, like, we have to test all this new stuff and go deep on GCP and Gemini. It’s nice for that.
352 00:33:23.340 ⇒ 00:33:31.030 Samuel Roberts: But yeah, I do miss a little bit sometimes, the, like, okay, this is my… my baby, this is how we’re gonna figure it out,
353 00:33:31.080 ⇒ 00:33:46.210 Samuel Roberts: But like I said, there’s pros and cons to both of those, and right now, I think, at this point in my life, I like this situation where I can, not necessarily worry about company things, payroll things, you know, but can still help culturally with things, and tech, you know, and,
354 00:33:46.780 ⇒ 00:33:57.329 Samuel Roberts: technically, like, you know, grow myself while also helping others, you know. So, that’s sort of where I’m at mentally with it, if I’m being totally honest. But yeah, it’s different.
355 00:33:57.980 ⇒ 00:34:15.350 Samuel Roberts: you know, I don’t think it’s bad, though. You know, I think, like, sometimes consulting can be really bad, and when I’ve talked to other friends that have, you know, maybe not tech as much, but, you know, people that I know that went to big consulting places, small consulting places, they run a certain way, and I… I think there’s been a very intentional
356 00:34:16.469 ⇒ 00:34:21.139 Samuel Roberts: decision, or series of decisions to try to run Brainforge in a way that is not
357 00:34:22.110 ⇒ 00:34:38.270 Samuel Roberts: that, if that makes sense. You know, I’ve been talking to Utam way back, you know, it’s… we’re… we were working with another dev shop, or we were taking over a project from another dev shop, and these clients, like, weren’t even able to get some of the code, and it was hard to, like, extract it from the proprietary code that the…
358 00:34:38.380 ⇒ 00:34:46.980 Samuel Roberts: other dev shop had written to try to, like, keep them in their ecosystem and things like that, and we don’t like to operate that way, you know, I think,
359 00:34:47.210 ⇒ 00:34:59.809 Samuel Roberts: you know, for that client, we were like, hey, it’ll live on your GitHub, we will use your… you know, it’s… we’re very, like… we want to bring value rather than, you know, extract something and then try to pretend we’re bringing value, and, you know, I didn’t have a…
360 00:35:00.250 ⇒ 00:35:09.819 Samuel Roberts: real negative, necessarily… let me just rephrase that. I didn’t have an automatic negative consult, thought about consulting, but, there was a little bit of, like, a…
361 00:35:10.030 ⇒ 00:35:15.030 Samuel Roberts: Okay, well, we’ll see how this plays out initially, and I think this has been a really good place to…
362 00:35:15.390 ⇒ 00:35:20.750 Samuel Roberts: Do consulting, because it’s not quite that, you know… Throw together a, you know.
363 00:35:21.150 ⇒ 00:35:32.260 Samuel Roberts: slide deck and try to teach people something that they probably already know and get paid for, you know what I mean? Like, there’s a… there can be a vibe that is wrong a little bit sometimes with consultants when they’re… their motivations aren’t…
364 00:35:32.300 ⇒ 00:35:46.280 Samuel Roberts: their interests might not be quite aligned, and I think we’ve done a… we’re still doing a good job of trying to keep our interests aligned with the clients, and deliver value to them, and outcomes for them that are good for them, and that means it’s good for us, kind of thing, so…
365 00:35:46.800 ⇒ 00:35:52.899 Cole Carter: That’s really cool. I like that. I like that, you know, I feel like that’s a lot more gratifying than trying to…
366 00:35:53.260 ⇒ 00:35:58.940 Cole Carter: I don’t know, become the… The thing they can’t throw away. You know, that just seems, like, parasitic.
367 00:35:58.940 ⇒ 00:36:13.379 Samuel Roberts: Right, you want to be the thing they can’t throw away because it helps them, and it brings them value, and not that they’re, you know… I always have hated the idea of, like, vendor lock-in from a technical standpoint, from, like, a CTR as I’m building things, like, I don’t love…
368 00:36:14.500 ⇒ 00:36:19.930 Samuel Roberts: knowing I can’t switch out of something easily, or what the cost will be associated with that. And so I think…
369 00:36:19.930 ⇒ 00:36:38.849 Samuel Roberts: the idea that, you know, they’re gonna stick with us because we drive outcomes, we drive ROI, we bring value, they like working with us, you know, they appreciate that we have other services that they might want to pull on at some point. It’s not, oh, you’re stuck because we own the code, and you don’t have anyone that can read it, and that kind of stuff. You know, it’s a different relationship that way.
370 00:36:38.850 ⇒ 00:36:40.770 Cole Carter: Cool. That’s cool. I like that.
371 00:36:41.100 ⇒ 00:36:43.780 Cole Carter: I had one other question, if that’s okay?
372 00:36:43.780 ⇒ 00:36:55.239 Samuel Roberts: Yeah, let me just double check before I make sure, because there was something else on my calendar, and I don’t remember exactly what time that was, but I think we’re good. But yeah, by all means, just start asking, and I will confirm that I’m not about to jump to something else.
373 00:36:55.500 ⇒ 00:36:59.180 Cole Carter: So I’ve worked as a contractor before.
374 00:36:59.180 ⇒ 00:36:59.790 Samuel Roberts: Yeah, we’re good.
375 00:36:59.790 ⇒ 00:37:02.219 Cole Carter: Sort of in college, and or.
376 00:37:02.220 ⇒ 00:37:02.550 Samuel Roberts: Okay.
377 00:37:02.550 ⇒ 00:37:04.050 Cole Carter: on top of a W-2.
378 00:37:04.350 ⇒ 00:37:08.640 Cole Carter: Sure. I’ve never just, like, straight up worked for just the 1099, so I’m curious.
379 00:37:08.640 ⇒ 00:37:09.040 Samuel Roberts: Yeah.
380 00:37:09.460 ⇒ 00:37:14.470 Cole Carter: like, what are the sort of ramifications of that, both in, I guess.
381 00:37:14.650 ⇒ 00:37:23.030 Cole Carter: You know, benefits to, like, how you manage that, and how you guys… I guess…
382 00:37:23.210 ⇒ 00:37:26.540 Cole Carter: manage employee expectations?
383 00:37:26.710 ⇒ 00:37:31.760 Cole Carter: you know, is this a locked-in deal, or is this gonna be around in a month from now, you know? How you guys.
384 00:37:31.760 ⇒ 00:37:41.240 Samuel Roberts: Right. Yeah, so, I mean, a lot of it comes down to, you know, allocating resources and, like, deal flow in general, and I think that’s been something that’s been…
385 00:37:41.580 ⇒ 00:37:52.570 Samuel Roberts: you know, growing in general, that’s why we’re hiring, you know, there’s… there’s deals in the pipeline, there’s contacts out there that are interested, and we may not have the resources to help them. So, like, there’s… I…
386 00:37:52.900 ⇒ 00:38:10.949 Samuel Roberts: when I joined, it was definitely like, okay, we’ll see how this goes, we’ll test it out. Back in, you know, July, kind of starting, ramping up a little bit. Since then, you know, there’s been plenty of work to do. And so, it’s not been, you know, as, you know, if you’re… I mean, there are some more nitty-gritty ramifications in terms of, like.
387 00:38:11.290 ⇒ 00:38:12.970 Samuel Roberts: Taxes and… and…
388 00:38:12.970 ⇒ 00:38:13.330 Cole Carter: Sure.
389 00:38:13.330 ⇒ 00:38:21.330 Samuel Roberts: how that all manages, and so there’s been some thought that I’ve had. But I, like I said, coming from a freelancing world, that wasn’t crazy to me. But if you’re coming from a, you know.
390 00:38:22.180 ⇒ 00:38:33.320 Samuel Roberts: a W-2 all the time, like, it might be a little weird to, like, get that initially… it might… you have to figure that out a little bit. Expectation-wise, I think it’s… it’s… you know, you’re not thought of as, like, in a…
391 00:38:33.490 ⇒ 00:38:44.400 Samuel Roberts: appendage that’s, like, also there. Like, there, you know, there’s some times where, like, you bring in a contractor to, like, do something very specific, and, you know, I think the path is definitely towards, you know.
392 00:38:44.520 ⇒ 00:38:48.900 Samuel Roberts: employee, and, you know, no one’s treated differently, necessarily.
393 00:38:49.900 ⇒ 00:38:57.200 Samuel Roberts: it’s… it’s more of a timing thing, I think, than anything. You know, like, it’s… instead of being, like, a probationary period, maybe, it’s… it’s like a…
394 00:38:57.300 ⇒ 00:39:00.940 Samuel Roberts: you know, ramp up into it, kind of thing.
395 00:39:01.610 ⇒ 00:39:04.960 Samuel Roberts: But I, I will also say, like, I’m not the, you know.
396 00:39:05.340 ⇒ 00:39:18.469 Samuel Roberts: people ops kind of side, I’m not necessarily the, like… So, like, there’s… there may be things that I’m not aware of or not answering quite correctly, so take everything with a grain of salt, but as someone who’s been on… on your side of what, you know, you’re talking about, it’s not been… not been crazy, so…
397 00:39:18.780 ⇒ 00:39:19.750 Cole Carter: Cool, cool.
398 00:39:20.330 ⇒ 00:39:21.950 Samuel Roberts: You know, people… Go ahead.
399 00:39:21.950 ⇒ 00:39:23.490 Cole Carter: That was… that was pretty much it, but I guess.
400 00:39:23.490 ⇒ 00:39:24.190 Samuel Roberts: Go ahead.
401 00:39:24.970 ⇒ 00:39:29.180 Cole Carter: As far as expectations, like, if I were to be hired or whatever.
402 00:39:29.180 ⇒ 00:39:29.630 Samuel Roberts: Yeah.
403 00:39:29.630 ⇒ 00:39:39.219 Cole Carter: would that kind of on-ramp look like? And then, as far as, like, delivering value slash being assessed, I’m curious, you know, what would it take for me to knock that out of the park?
404 00:39:39.670 ⇒ 00:39:41.960 Samuel Roberts: Yeah, yeah, so, okay, so,
405 00:39:42.490 ⇒ 00:40:01.839 Samuel Roberts: if everything goes well, I haven’t really talked about the full process of the hiring, but I can mention that at the very end. But yeah, so say you get an offer, you know, you join, you’d probably just be really jump right in. You know, we are adding more, as we’re growing now, structure around that onboarding. You know, when I was onboarded in July, it was a little bit more.
406 00:40:01.970 ⇒ 00:40:09.550 Samuel Roberts: Just in the deep end, and asking tons of questions, and that works when it’s a certain size, but as we’re growing, it’s… people are, you know.
407 00:40:09.550 ⇒ 00:40:22.499 Samuel Roberts: now that we have, kind of, people ops and, you know, recruitment now. Like, we didn’t really have these things before, so there’s a little bit more structure there. Bit of a, you know, buddy system for, like, checking in with folks and making sure things are good.
408 00:40:22.960 ⇒ 00:40:30.020 Samuel Roberts: making sure you’re finding things and all that sort of stuff, you know, from the operational side of things. From the work side of things, you know, it’s… it’s…
409 00:40:31.260 ⇒ 00:40:36.250 Samuel Roberts: it’s more about the outcomes, I guess, and the, like, you know,
410 00:40:37.600 ⇒ 00:40:53.360 Samuel Roberts: That sounds maybe not quite right, but I’m trying to think of the best way to phrase it, but, you know, it’s a different world than it was, you know, a few years ago, even, in terms of, like, what can be expected in a certain time period. That being said, like, we still want to produce quality code, which is something that is…
411 00:40:53.720 ⇒ 00:40:59.570 Samuel Roberts: you know, the industry itself is dealing with, and so I think, you know, there’s a lot of,
412 00:40:59.730 ⇒ 00:41:15.970 Samuel Roberts: chatter, I guess, out there in general about, you know, like, vibe coding versus, you know, like, engineering with the agent and things like that. And I think I’m sort of finding my way with that as a personal thing, but at the end of the day, like, you know, we’re trying to deliver results for clients, and so, you know, it’s…
413 00:41:16.830 ⇒ 00:41:19.469 Samuel Roberts: Finding that balance and knowing where the right
414 00:41:19.830 ⇒ 00:41:27.199 Samuel Roberts: corners to cut are and where to focus on things is always important, even, you know, any kind… especially at a startup, especially.
415 00:41:27.600 ⇒ 00:41:34.529 Samuel Roberts: And so I think, like, from a, like, how you’re doing, kind of, how you’re gonna knock it out of the park, it’s really, you know,
416 00:41:36.360 ⇒ 00:41:53.539 Samuel Roberts: we try to do some measurement of tickets and velocity and things like that, as you can imagine, but it’s also, like, how is it feeling? How’s it going? Like, is the work getting done right? Are they asking the right questions? Are they asking the wrong ques… are they not asking… you know, there’s a lot of that kind of stuff that’s, like, and…
417 00:41:53.840 ⇒ 00:42:12.390 Samuel Roberts: you can imagine, like, certain people, like, hit the ground running and do really well, and learn the right things the right way, because they’re asking the right questions, and certain people, don’t ask the right questions, and you don’t realize they’re going down the wrong rabbit hole until it’s a little too late. And you kind of get nudged certain ways and learn that. I think to, like, knock it out of the park, it really is, like.
418 00:42:12.570 ⇒ 00:42:14.590 Samuel Roberts: Be ready to learn as much as you can.
419 00:42:15.540 ⇒ 00:42:19.400 Samuel Roberts: I would say, like, you know, the work is the work, and you can…
420 00:42:19.590 ⇒ 00:42:23.990 Samuel Roberts: Measure some of those outcomes. But it is really, like, you know,
421 00:42:24.170 ⇒ 00:42:27.809 Samuel Roberts: be ready to jump into different things. You may not necessarily have the most,
422 00:42:27.960 ⇒ 00:42:35.930 Samuel Roberts: familiarity with initially, and figure out how to get that familiarity. You know, it’s… I’ve been thinking a lot of it with the agents, like, providing the right context to the agent.
423 00:42:36.340 ⇒ 00:42:48.360 Samuel Roberts: is important, because then you can maybe let the agent run a lot better than if you were just saying, like, build me this and nothing else. And, you know, people managing agents, the… thinking of that as, like, a manager managing engineers and things like… like, there’s a lot of parallels.
424 00:42:48.360 ⇒ 00:42:57.749 Samuel Roberts: And I think it really is, like, you know, it’s my job, it’s, you know, certain people’s job to provide the right context for executing on tasks and building the right things.
425 00:42:57.750 ⇒ 00:42:59.569 Cole Carter: But I think you’re hitting on something which is, like.
426 00:42:59.700 ⇒ 00:43:02.299 Cole Carter: It’s about the questions that are asked, too.
427 00:43:02.300 ⇒ 00:43:09.999 Samuel Roberts: That’s what I was gonna say, and so, like, if I’m not giving the right piece of information, and someone doesn’t, you know, the agent makes an assumption, or the engineer makes the assumption, like.
428 00:43:10.510 ⇒ 00:43:21.910 Samuel Roberts: are they making good assumptions? And if they’re not, like, do they know they’re not making good assumptions? Do they know where those assumptions are? Like, there’s a lot there, and I think it really is about communication at that point. You know, you don’t want to…
429 00:43:21.910 ⇒ 00:43:33.449 Samuel Roberts: sit at your computer for a number of hours, just, like, grinding gears. You want to make sure that that’s being surfaced sooner, and that, things escalate when there’s problems, and figure out, like, how to solve them as a team.
430 00:43:33.620 ⇒ 00:43:35.710 Samuel Roberts: Rather than just, like.
431 00:43:36.050 ⇒ 00:43:42.480 Samuel Roberts: oh, I didn’t know what to do, so I tried this one thing that was bad, and spent hours on it, and it’s a waste of time now, you know?
432 00:43:42.790 ⇒ 00:44:01.330 Samuel Roberts: that, I think, is really how you hit the ground running and do a really good job. Like, you’ll… the rest of it, like, what the outcomes and all that kind of comes along with that, I would say. Assuming, you know, you’re capable of doing the work and stuff, like, you know, technically I, you know, we’re having a more of a kind of cultural fit interview now, but imagining that, you know, you can do that and can learn quick.
433 00:44:01.720 ⇒ 00:44:20.519 Samuel Roberts: and can, you know, communicate well internally, externally, there’s that side of it I haven’t really touched on, I guess, as much, but, you know, we don’t restrict people from communicating to clients, necessarily, on, like, external Slack channels, the way some places might say, like, oh, no, no, no, the engineering goes through the PM, and the PM, you know, whatever. But that being said, like, you know.
434 00:44:20.780 ⇒ 00:44:25.270 Samuel Roberts: you’re not putting together the presentations, you need to make sure that whoever is knows what’s been happening. There’s a lot of…
435 00:44:25.270 ⇒ 00:44:25.950 Cole Carter: Hmm.
436 00:44:26.200 ⇒ 00:44:30.139 Samuel Roberts: Stuff like that to make sure that, you know, flow is happening. And some of that is…
437 00:44:31.120 ⇒ 00:44:34.550 Samuel Roberts: Changing, and as we grow, you know, like any other startup, it’s…
438 00:44:35.090 ⇒ 00:44:40.500 Samuel Roberts: malleable, and we’re learning, you know, building the plane as you’re flying it kind of… kind of deal, so,
439 00:44:41.730 ⇒ 00:44:47.000 Samuel Roberts: Yeah, I don’t know what else I have to add about that, but I hope that answers enough of it, you know, a few questions, though.
440 00:44:47.280 ⇒ 00:44:53.359 Cole Carter: Because, yeah, I guess… Obviously, there’s, you know, Ascertaining value just from…
441 00:44:53.790 ⇒ 00:44:57.559 Cole Carter: delivery, but I do like that there is this more, like.
442 00:44:57.830 ⇒ 00:45:13.389 Cole Carter: general or strategic or holistic kind of assessment of, like, is this person even, you know, are they in… are they using their brain towards the problems in, like, an intelligent way? Totally. Yeah, I think if you, you know, if you can clear that bar, that’s…
443 00:45:13.830 ⇒ 00:45:15.000 Cole Carter: I’m fine by that.
444 00:45:15.140 ⇒ 00:45:19.720 Samuel Roberts: Right, you know, I mean, there’s one world, I guess, or one type of…
445 00:45:20.120 ⇒ 00:45:37.970 Samuel Roberts: engineer that, you know, can figure it all out and not ask the right… but, like, alignment on that is just as bad as an alignment with an AI, you know, you need to make sure that even if they’re going off on their own, they have the right information at the start. You know, if you’re doing a spike on something, and the whole point is to answer a bunch of questions, which questions are being answered, you know, that kind of stuff.
446 00:45:37.970 ⇒ 00:45:39.230 Cole Carter: Like, you know.
447 00:45:39.650 ⇒ 00:45:41.509 Cole Carter: Okay. Well, that’s cool. I like that.
448 00:45:41.510 ⇒ 00:45:42.530 Samuel Roberts: Yeah, yeah, totally.
449 00:45:42.910 ⇒ 00:45:48.230 Cole Carter: Thanks for answering that, and I know we went over, thank you for answering my extra questions.
450 00:45:48.230 ⇒ 00:45:55.060 Samuel Roberts: Yeah, yeah, no worries, no worries. Yeah, I’m glad there was nothing read after, so I was a little nervous for a second that I’d have to not answer, but I’m glad we had the time.
451 00:45:55.380 ⇒ 00:45:57.820 Samuel Roberts: Real quick, so this is kind of the first…
452 00:45:58.130 ⇒ 00:46:00.680 Samuel Roberts: Culture interview, then the next one is a little more…
453 00:46:01.010 ⇒ 00:46:08.139 Samuel Roberts: role-focused. It’s… it’s like a role-focused, more technical interview, but I don’t like saying technical interview, because you’re not…
454 00:46:08.320 ⇒ 00:46:27.409 Samuel Roberts: you’re not, like, doing leak code or coding in front of someone, necessarily. After that, if it’s, you know, if you pass that gate, it would be a, like, take-home kind of tech assessment. So we have a project that we’d share with you, and you would have kind of free reign to build the solution, and
455 00:46:27.530 ⇒ 00:46:38.829 Samuel Roberts: present that to us as, like, the third interview, would be basically, so you’d go off-build, come back, you know, present in some way to, I think it’s three of us on that call, usually. Okay.
456 00:46:39.130 ⇒ 00:46:58.710 Samuel Roberts: And then that’s, you know, that’s kind of the, like, the final, like, okay, yes, they’re… they’ve passed the few bars, and this one proves that they can do it, and then after that would be offer or not. So, yeah, kind of three interviews. And we like to move relatively quickly. I keep saying the main thing is just scheduling time, because, you know, that’s the kind of biggest thing, so, sure.
457 00:46:58.710 ⇒ 00:47:02.810 Samuel Roberts: Yeah, but that’s the… that’s the process, cool.
458 00:47:03.520 ⇒ 00:47:08.060 Cole Carter: Just any tips for the technical interview at all?
459 00:47:09.390 ⇒ 00:47:10.820 Cole Carter: Prep 4, specifically?
460 00:47:10.880 ⇒ 00:47:18.720 Samuel Roberts: No, I mean, like, I… that’s the thing, I don’t like calling it technical, because it’s, like, it’s more about, like, the role and how you think about certain things, it’s not necessarily, like, a…
461 00:47:18.880 ⇒ 00:47:29.279 Samuel Roberts: you know, reverse this, whatever, linked list kind of stuff. You know, it’s less of that, you know, that… and especially in the age of AI, you know, everyone’s figuring out how to assess
462 00:47:29.600 ⇒ 00:47:38.269 Samuel Roberts: what people can do when all, you know, the agents can do so many things, like, where’s the right piece? So, like, I wouldn’t say, you know, I’ve never really been a fan of those types of, like.
463 00:47:38.490 ⇒ 00:47:41.259 Samuel Roberts: Leap Code interviews in general.
464 00:47:41.690 ⇒ 00:47:55.579 Samuel Roberts: But, you know, there’s still a certain amount of, like, understanding the tech, and that’s sort of the third part. So I think it’s more of, like I said, role-focused, you know, we got into, like, a little bit of, like, LM stuff, but they’ll probably go a little bit deeper, and, like, how would you make certain decisions kind of thing, but… Okay, cool. Yeah.
465 00:47:55.580 ⇒ 00:47:57.110 Cole Carter: So, bring a strategy brain.
466 00:47:57.710 ⇒ 00:48:08.839 Samuel Roberts: Yeah, yeah, yeah, I wouldn’t necessarily worry about, like, clanking away at the keyboard kind of thing, but yeah, you want to be able to answer questions about if you were kind of given a task, how would you attack it sort of questions, that sort of thing.
467 00:48:09.610 ⇒ 00:48:10.380 Samuel Roberts: Right?
468 00:48:10.640 ⇒ 00:48:14.629 Cole Carter: I will… I guess you’ll send an email for scheduling that, and then…
469 00:48:14.630 ⇒ 00:48:26.350 Samuel Roberts: Yeah, yeah, so, I mean, like, it’s the way I… normally, if everything, I mean, this is, you know, if you pass this one, you’ll get an email from recruitment, one way or another in a day or two, ideally. Certainly by the end of the week, I’d say.
470 00:48:27.080 ⇒ 00:48:32.520 Samuel Roberts: And then, yeah, that’ll be a whole scheduling with the next person. So, the other side of it is meeting more of the team as well, so…
471 00:48:32.520 ⇒ 00:48:35.810 Cole Carter: Sure, sure. Cool. I’ll look forward to that.
472 00:48:35.810 ⇒ 00:48:37.809 Samuel Roberts: Alright, yeah, thank you so much, Cole.
473 00:48:37.810 ⇒ 00:48:42.490 Cole Carter: So, Sam, I appreciate you reaching out and talking to me and talking through all my questions.
474 00:48:42.490 ⇒ 00:48:45.600 Samuel Roberts: Yeah, I’m glad I could answer some of them, so yeah, great.
475 00:48:46.060 ⇒ 00:48:48.930 Samuel Roberts: Thank you. Alright, look forward to reading you again. You too. Bye-bye.