Meeting Title: Brainforge x Mixpanel MCP Feedback Sync Date: 2026-02-25 Meeting participants: Greg Stoutenburg, Michael Armstrong
WEBVTT
1 00:01:07.850 ⇒ 00:01:09.070 Greg Stoutenburg: Hey, Michael!
2 00:01:09.240 ⇒ 00:01:11.060 Michael Armstrong: Hi, how are you doing?
3 00:01:11.060 ⇒ 00:01:12.760 Greg Stoutenburg: Hey, I’m doing alright, how are you?
4 00:01:12.760 ⇒ 00:01:18.150 Michael Armstrong: I’m wonderful, really excited to be talking with you. Thanks for making the time, I know you’re busy.
5 00:01:18.150 ⇒ 00:01:19.239 Greg Stoutenburg: Yeah, happy to connect!
6 00:01:19.370 ⇒ 00:01:21.220 Michael Armstrong: Yeah, where are you based out of?
7 00:01:21.410 ⇒ 00:01:25.520 Greg Stoutenburg: I’m in York, Pennsylvania, which is a little under an hour from Baltimore.
8 00:01:25.520 ⇒ 00:01:30.370 Michael Armstrong: Okay, very cool. I’m based in San Francisco.
9 00:01:30.370 ⇒ 00:01:30.930 Greg Stoutenburg: Okay, cool.
10 00:01:30.930 ⇒ 00:01:35.479 Michael Armstrong: And, I’ve been with Mixpanel for, I’d say, about 5 months now.
11 00:01:35.480 ⇒ 00:01:35.840 Greg Stoutenburg: Okay.
12 00:01:35.840 ⇒ 00:01:58.890 Michael Armstrong: manager, so a lot of my job is getting to talk to our customers, figuring out their aha moments in their zone of genius, and learning how they use the product to make sense of their data and make smart, insightful decisions. And so, I’m really excited for us to be able to have a short chat around
13 00:01:59.240 ⇒ 00:02:10.530 Michael Armstrong: your use of the MCP server, and also collect some feedback, if you have any, to pass on to the product team, because we’re always looking to improve. Yeah.
14 00:02:10.539 ⇒ 00:02:11.149 Greg Stoutenburg: Yeah, yeah.
15 00:02:11.150 ⇒ 00:02:13.170 Michael Armstrong: How does that sound in terms of agenda?
16 00:02:13.620 ⇒ 00:02:20.280 Greg Stoutenburg: Sure, yeah, sounds good. So, yeah, so we’ve got, we’ve got a few clients using Mixpanel.
17 00:02:20.480 ⇒ 00:02:26.690 Greg Stoutenburg: I’ve been doing work for one of them, so, you know, I know what their instance looks like, and I’ve spent some time in it.
18 00:02:26.690 ⇒ 00:02:36.809 Michael Armstrong: Before you dig into that, could you just, like, zoom out a little bit and tell me a little bit about the work of Brainforge and your role in particular, so that we’ve got all of that content with.
19 00:02:36.810 ⇒ 00:03:00.359 Greg Stoutenburg: Yeah, yeah, sure, yeah. So, yeah, BrainForge is a data and analytics consultancy. We help customers improve workflows, especially using AI. We help with things like data platform migrations. We help with things like improving user activation engagement by helping, to set up product analytics tools and making sure that events are instrumented correctly.
20 00:03:00.360 ⇒ 00:03:01.339 Greg Stoutenburg: Things like that.
21 00:03:01.340 ⇒ 00:03:06.880 Greg Stoutenburg: So, that’s… that’s sort of, like, the mission, sort of, you know, data help. Perfect.
22 00:03:06.880 ⇒ 00:03:15.249 Michael Armstrong: You’re the perfect person for us to be talking to, then. Alright, so tell me a little bit about your role, and then the organizations that you’ve been working with.
23 00:03:15.250 ⇒ 00:03:19.710 Greg Stoutenburg: Yeah, yeah, so I’ve been with Brainforge for a month, and…
24 00:03:19.710 ⇒ 00:03:20.490 Michael Armstrong: Boom!
25 00:03:20.490 ⇒ 00:03:27.840 Greg Stoutenburg: Yeah, yeah, thanks, everything… I know, just yesterday I was talking to CEOs, and I was like, I feel like I’ve been here for a year, and it’s been 6 weeks.
26 00:03:27.840 ⇒ 00:03:30.879 Michael Armstrong: I know the feeling.
27 00:03:31.050 ⇒ 00:03:34.880 Greg Stoutenburg: Yeah, the ramp-up period is, you know, fly now.
28 00:03:34.880 ⇒ 00:03:57.649 Greg Stoutenburg: Yeah, and so, I work in data strategy, data and strategy. I am the person who would work with a client, like, okay, they know they need product analytics, but what do we do? And we go from those conversations around, what does your product do, and let’s sort of map out workflows and, you know, the conceptual part, and then work with our engineers to actually instrument events.
29 00:03:57.650 ⇒ 00:04:03.309 Greg Stoutenburg: And then help the team build funnels, understand what they’re looking at, things like that.
30 00:04:03.580 ⇒ 00:04:21.100 Greg Stoutenburg: I’m also for a client, doing a data platform migration. They’re on Tableau, and they’re moving to Omni, and so I’ve led the project to hook up the data sources that were going into Tableau, recreate those dashboards one-to-one, things like that. So, my background is in product-led growth.
31 00:04:21.100 ⇒ 00:04:28.720 Greg Stoutenburg: I was a growth PM for a few years, before ending up here at Brainforge, and, yeah.
32 00:04:29.190 ⇒ 00:04:51.449 Michael Armstrong: Fantastic. All right, great. Now that we’ve, set the lay of the land, tell me a little bit about some of the organizations that you’ve been working with, specifically using the MCP server, to want to be able to highlight a couple of examples of, you know, magic aha moments or efficiency gains specifically for these orgs.
33 00:04:51.950 ⇒ 00:04:54.100 Greg Stoutenburg: Yeah, I mean, I can give you a pretty good one, so…
34 00:04:54.100 ⇒ 00:04:54.770 Michael Armstrong: Great!
35 00:04:54.770 ⇒ 00:05:08.949 Greg Stoutenburg: Yeah, I mean, I don’t know if they want to be named, so I’ll leave their name out, but I’ll just say, an online, medical e-commerce company that’s, that’s relatively big, they use Vixpanel, and,
36 00:05:08.950 ⇒ 00:05:20.310 Greg Stoutenburg: So, I, you know, because we have the partnership with Mixpanel, I was asked to record demo videos a couple weeks back, where I set up the MCP, and, you know, I could have just asked it to
37 00:05:20.310 ⇒ 00:05:25.899 Greg Stoutenburg: build a chart. You know, I was using Claude on my desktop. I couldn’t ask to build a chart, and I thought, like, you know.
38 00:05:26.480 ⇒ 00:05:41.510 Greg Stoutenburg: my opinion is that, for the most part, if all you’re trying to do is just, like, build a chart, it’s faster to do it than it is to even bother with the MCP connection. Like, just log in and do it. So I was like, I want to give it something that’s, like, bigger, that I can sort of set and forget and come back to it. So…
39 00:05:42.010 ⇒ 00:05:56.130 Greg Stoutenburg: what I told, what I told Claude to do using MixedPanel MCP is look at a certain… look at a certain intake. I mean, like, basically, it’s like a, you know, like an activation funnel, to look at it, and then…
40 00:05:56.130 ⇒ 00:06:08.890 Greg Stoutenburg: using some additional sources that I gave upon… about experimentation and experiment prioritization, give me a principled roadmap for the next quarter to improve this funnel.
41 00:06:09.000 ⇒ 00:06:14.619 Greg Stoutenburg: And, and it did. And it gave this, like… you know, it took some… took some wrestling.
42 00:06:14.960 ⇒ 00:06:27.270 Greg Stoutenburg: as it does with AI tools, but, it gave me, like, this really detailed 14-page report that was, honestly pretty spot-on. It was pretty good. So, I was impressed by that.
43 00:06:27.900 ⇒ 00:06:39.230 Michael Armstrong: That’s… fantastic. I may, we’re actually in the process of launching an e-commerce campaign, so, maybe there’s a possibility for you to connect with them and see if they’d be.
44 00:06:39.230 ⇒ 00:06:39.570 Greg Stoutenburg: Yeah.
45 00:06:39.570 ⇒ 00:06:45.760 Michael Armstrong: in being included in maybe a broader customer story that connects all of us together, that would be fantastic.
46 00:06:45.760 ⇒ 00:06:46.320 Greg Stoutenburg: Yup.
47 00:06:46.320 ⇒ 00:06:52.340 Michael Armstrong: So, how long would you say that that normally would have taken without, using MCP Server?
48 00:06:52.800 ⇒ 00:06:56.050 Greg Stoutenburg: That would have taken… I mean…
49 00:06:59.090 ⇒ 00:07:10.930 Greg Stoutenburg: I mean, the hard part was really… was definitely the write-up, but the MCP set me up… set me up by… Sorry, sped me up by… oh, geez, I don’t know.
50 00:07:11.740 ⇒ 00:07:29.030 Greg Stoutenburg: at… I think at least an hour? Because one of the things that was cool about it is that it sounded like it just looked at the one chart, the same one that I was looking at, right? Like, it was using at the events used to build the chart, and then looking at that more comprehensively in the context of everything that’s available.
51 00:07:29.110 ⇒ 00:07:38.389 Greg Stoutenburg: In, in Mixpanel, in this Mixpanel instance. So, yeah, I mean, I guess I’ll say an hour for that, and then as far as, like…
52 00:07:38.660 ⇒ 00:07:42.900 Greg Stoutenburg: So if the story… here’s what I’m thinking as an editor, right?
53 00:07:42.900 ⇒ 00:07:43.370 Michael Armstrong: Yeah.
54 00:07:43.370 ⇒ 00:07:45.489 Greg Stoutenburg: is just MixPanel MCP.
55 00:07:46.190 ⇒ 00:07:47.540 Greg Stoutenburg: good story.
56 00:07:48.390 ⇒ 00:08:03.390 Greg Stoutenburg: not complete. If it’s like, how much was I able to speed up by using Mixpanel MCP and resources that were going to help me get the client a deliverable that they could act on and, you know, do something,
57 00:08:04.120 ⇒ 00:08:14.159 Greg Stoutenburg: extraordinarily faster. Like, probably, probably, you know, maybe then it’s, like, what would have taken a half a day, or three quarters of a day, took.
58 00:08:14.330 ⇒ 00:08:15.600 Greg Stoutenburg: Minutes.
59 00:08:15.730 ⇒ 00:08:21.319 Michael Armstrong: Yeah, that’s incredible. I also would love to know,
60 00:08:22.460 ⇒ 00:08:30.600 Michael Armstrong: Because I think, like, a really… I think what is most exciting about this is, you know, I think nowadays folks are trying to put
61 00:08:30.790 ⇒ 00:08:47.500 Michael Armstrong: AI into everything, like refrigerators, but, like, this is really purpose-built for making sense of mountains of data to make intelligent choices that really drive impact, bottom line, revenue. So I’m curious,
62 00:08:47.790 ⇒ 00:08:59.369 Michael Armstrong: How do you see this potentially changing the way that, maybe non-technical teams interact with data, for leadership to be able to.
63 00:08:59.370 ⇒ 00:09:04.639 Greg Stoutenburg: Get a direct through-line and make sense of this data so that…
64 00:09:04.640 ⇒ 00:09:15.230 Michael Armstrong: Everyone is really aligned and speaking the same data-driven conversations and building organizational alignment across teams, whether they’re technical or not.
65 00:09:15.230 ⇒ 00:09:31.220 Greg Stoutenburg: Yeah, yeah, I mean, I think what’s great about it, like, for the non-technical user, is that, if they’re… if they’re not so accustomed to logging in and looking at an analytics interface and, messing with those filters and things like that to… to adjust a chart.
66 00:09:31.290 ⇒ 00:09:36.039 Greg Stoutenburg: they can just ask a question directly using natural language, and I think that…
67 00:09:36.040 ⇒ 00:09:56.579 Greg Stoutenburg: I think being able to just rely on the way that you’re thinking about it naturally, rather than having to adjust the way you think to the way a tool is presented it, which is, you know, just not natural, you know, it’s built out of engineering code and formal logic and things like that, is just extraordinarily powerful, because it’s gonna… the way that I like to think of this is, like, data democratization.
68 00:09:57.130 ⇒ 00:10:03.449 Greg Stoutenburg: you know, you can just… you can just get answers the way you would normally ask them, and then get on with what you really want to do, right? Like…
69 00:10:03.450 ⇒ 00:10:04.050 Michael Armstrong: Hmm.
70 00:10:04.400 ⇒ 00:10:10.779 Greg Stoutenburg: the way that I… the way that I put it to, cheer on, you know, your… your AI PM is, like.
71 00:10:10.810 ⇒ 00:10:30.050 Greg Stoutenburg: It’s like, you know, the more I think about it, I don’t think anyone has ever actually wanted to look at a dashboard. Like, that’s never been the goal, you know, for anything. You know, what they really want to do is, like, improve revenue, you know? And that requires some measurements, then you have to, like, look at these measurements, but, like, no one’s goal is ever to look at a dashboard, so, you know,
72 00:10:30.570 ⇒ 00:10:42.740 Greg Stoutenburg: the lesson then being for those non-technical users, that MCP, natural language, enables you to get the answers that you really need to get, so you can, you know, get on with whatever it is you’re actually trying to accomplish.
73 00:10:42.920 ⇒ 00:10:51.519 Michael Armstrong: I love that. So I’m actually in the process of, building out a webinar with, McGaw.
74 00:10:51.550 ⇒ 00:11:08.200 Michael Armstrong: And the content is a walkthrough of an e-commerce workflow example, because it really, like, integrates all the elements of tech, from product development, QA, experimentation.
75 00:11:08.200 ⇒ 00:11:24.630 Michael Armstrong: Our metric trees feature so that, you know, you really go from high-level executive, line of sight, into impact of work, down through the dashboards that are brought to life for day-to-day practitioners of the product.
76 00:11:24.630 ⇒ 00:11:38.830 Michael Armstrong: I’m curious your, your take on the benefits that some of these executive C-suite-level folks who really need to get their finger on the pulse of
77 00:11:38.830 ⇒ 00:11:46.359 Michael Armstrong: the, metrics that matter that are going to drive successful KPIs, if not to be able to pivot and make
78 00:11:46.360 ⇒ 00:11:59.000 Michael Armstrong: data-informed decisions, when it matters, and not in a, you know, post-mortem, after the KPI has been missed. Because in, you know, I think the beauty of this e-commerce use case is
79 00:11:59.020 ⇒ 00:12:04.420 Michael Armstrong: It integrates all the aspects of, you know, traditional tech, but…
80 00:12:04.550 ⇒ 00:12:18.480 Michael Armstrong: it really demonstrates high-stakes value, because 1% can make or break a quarter. And so, taking that use case as an example, how do you see the benefit,
81 00:12:18.710 ⇒ 00:12:27.759 Michael Armstrong: for some of those, executive sponsors of the product, and how they would use MCP to achieve those results.
82 00:12:28.030 ⇒ 00:12:47.599 Greg Stoutenburg: Yeah, that’s… I feel like that’s a tougher question. So, like, I haven’t talked to executives about metric trees specifically. I know, I mean, you know, I knew the concept already and worked with the concept, in other contexts. I would think that what… what that’s… one of the things that that’s definitely useful for is for managers trying to understand how to get all their teams to align.
83 00:12:47.600 ⇒ 00:12:58.949 Greg Stoutenburg: Because it’s just… it’s just not obvious in any, you know, at a glance, why, you know, some, sad… some advertising work
84 00:12:59.420 ⇒ 00:13:07.149 Greg Stoutenburg: leads to revenue, ultimately, right? Like, that is a lot of work. So to be able to visualize that, and to be able to formalize and standardize
85 00:13:07.290 ⇒ 00:13:17.909 Greg Stoutenburg: those connections is, I think, I think awesome. Yeah, and then, you know, MCV enables you to look at that as well. You know, so an executive could do something like
86 00:13:19.550 ⇒ 00:13:33.299 Greg Stoutenburg: have, create a workflow that uses Mixpanel MCP, uses what they’ve built for metric trees and for the events that they’ve instrumented, and get, like, a, you know, get a weekly report.
87 00:13:33.300 ⇒ 00:13:48.529 Greg Stoutenburg: that’s like, you know, here’s how… here’s how all of our departments are contributing to the bottom line, right? Like, they can look at something like that and do performance reporting. And I think something like that, like, the more automated that is, and then it uses some template that they’ve
88 00:13:48.610 ⇒ 00:14:03.250 Greg Stoutenburg: they’ve created, it’s just going to speed them up, right? Rather than something that it might have been, again, their couple of hours before the weekly meeting is something that, you know, they click a button, or maybe they don’t even click a button, it’s just scheduled, and it’s just, here it is, you know?
89 00:14:04.650 ⇒ 00:14:08.080 Michael Armstrong: Yeah, it’s remarkable. I love that.
90 00:14:09.900 ⇒ 00:14:21.980 Michael Armstrong: I am also interested, because I think, even just being at Mixpanel for the last 5 months, I’ve had a front-row seat to the launch of a whole range of…
91 00:14:22.080 ⇒ 00:14:27.069 Michael Armstrong: Brand new, exciting products that are,
92 00:14:27.380 ⇒ 00:14:40.459 Michael Armstrong: having the ability to make sense of all of your data in one platform, and so I’m really curious, especially with your experience in the broader industry.
93 00:14:41.080 ⇒ 00:14:43.139 Michael Armstrong: the impact of…
94 00:14:43.400 ⇒ 00:15:00.409 Michael Armstrong: the MCP server to be able to make sense of all of this data when we’re approaching things through a tool consolidation perspective. Because, I mean, I used to work at Asana, and so, like, a lot of that work was
95 00:15:00.790 ⇒ 00:15:03.400 Michael Armstrong: Viscerally feeling and understanding…
96 00:15:03.700 ⇒ 00:15:18.840 Michael Armstrong: the impact of data silos, and having to use multiple tools, there’s no central source of truth, and so I’m curious, your take, considering your experience, and, specifically about how MCP Server could help in that regard.
97 00:15:18.840 ⇒ 00:15:36.930 Greg Stoutenburg: Yeah. Yeah, I mean, I guess one of the things that’s cool about MCP Server is that, you know, you can… you can connect more than one MCP Server to your AI tool. So, you know, you could, in this way, do work just using a single interface and be pulling data from Mixpanel and, you know, whatever other sources.
98 00:15:37.010 ⇒ 00:15:39.980 Greg Stoutenburg: And turning that into something actionable.
99 00:15:40.200 ⇒ 00:15:45.960 Greg Stoutenburg: So, I think that… I think that the consolidation potential is definitely… definitely there.
100 00:15:46.200 ⇒ 00:15:58.639 Greg Stoutenburg: Yeah, and now here, and now here, this is a case where it’s not necessarily Mixpanel doing the consolidating, it’s Mixpanel partly enabling the consolidating by, you know, providing MCP server. But, you know, no less valuable.
101 00:15:59.890 ⇒ 00:16:09.890 Michael Armstrong: Wonderful. Well, this has been super helpful. Before I let you go and give some time back, I’d love to… if you have any kind of,
102 00:16:10.350 ⇒ 00:16:15.469 Michael Armstrong: Feedback, you know, like, well, I guess, you know, what do you love, and what would you love to see?
103 00:16:15.470 ⇒ 00:16:15.930 Greg Stoutenburg: Yup.
104 00:16:16.480 ⇒ 00:16:17.040 Greg Stoutenburg: Yeah.
105 00:16:17.470 ⇒ 00:16:30.040 Greg Stoutenburg: Yeah, I mean, so my experience using it was, like, it did what I wanted it to do, and it did it very fast, and even though, like I said, it took some wrestling, it still got there.
106 00:16:30.040 ⇒ 00:16:38.609 Greg Stoutenburg: Something that became sort of pretty quickly evident to me is that when I was typing in for natural language, like, event descriptions.
107 00:16:38.610 ⇒ 00:16:40.650 Michael Armstrong: Just was not getting picked up.
108 00:16:40.650 ⇒ 00:16:48.419 Greg Stoutenburg: accurately. Like, I… and I use the exact right event name for what we’ve instrumented, and it, like…
109 00:16:48.580 ⇒ 00:16:52.790 Greg Stoutenburg: it, like, couldn’t find it, or it would say, oh, I think you mean this other event, and it’s like.
110 00:16:52.920 ⇒ 00:17:11.830 Greg Stoutenburg: Nope. This is what I said. So, maybe, I don’t know, further work on, like, the schema that’s used to pick up, like, so the natural language input has to, you know, follow some kind of schema to identify, like, this is the right event, or this is the right event, or this is the right property. So that did make it, like.
111 00:17:12.329 ⇒ 00:17:14.050 Greg Stoutenburg: Yeah, that was a challenge.
112 00:17:14.510 ⇒ 00:17:32.249 Greg Stoutenburg: The second challenge was, it was just slow. Like, it was a lot slower than I thought it was going to be. Even before getting to, like, the experimentation reporting part, it just seemed like… and now, who knows, maybe it was… maybe Claude was having a bad day, you know, maybe it had nothing to do with the mix panel side, but, like, it was… it was really slow, and I was like.
113 00:17:32.250 ⇒ 00:17:40.899 Greg Stoutenburg: okay, you know, if I didn’t have to intervene in this process at all, and it was just gonna run in the background, maybe I wouldn’t even care. But, you know, for live demo purposes, I was sitting there.
114 00:17:41.120 ⇒ 00:17:48.480 Greg Stoutenburg: And just trying not to move, because I knew that the video editor wouldn’t have to see my head, you know, shaking all around, because they sped it up.
115 00:17:48.480 ⇒ 00:17:48.980 Michael Armstrong: Yeah.
116 00:17:48.980 ⇒ 00:18:00.810 Greg Stoutenburg: Yeah, so, yeah, I mean, so those are… those were the negative parts, but yeah, the positive was… it was clearly very powerful, and the potential for what it could do is, I mean, is, yeah, it’s enormous, just…
117 00:18:01.150 ⇒ 00:18:10.330 Greg Stoutenburg: Use natural language, take your data source, and, have it do things for you, is, is great. Yeah.
118 00:18:10.880 ⇒ 00:18:11.580 Michael Armstrong: Fantastic.
119 00:18:11.580 ⇒ 00:18:12.600 Greg Stoutenburg: Ben, yep.
120 00:18:13.300 ⇒ 00:18:14.580 Michael Armstrong: Alright, well,
121 00:18:14.630 ⇒ 00:18:33.300 Michael Armstrong: this has been super helpful, and I will… I will take a look at, our transcript and put together a couple of, examples for, you know, potential slide decks, to include in slide decks, as well as, to integrate within.
122 00:18:33.300 ⇒ 00:18:38.589 Michael Armstrong: upcoming blogs, so as that happens, I’d love to share with you, and…
123 00:18:38.590 ⇒ 00:18:52.479 Michael Armstrong: get your approval, before launching, and then also, I’d love to, like, do you have routine conversations or syncs with the, e-commerce organization that you’ve been working with?
124 00:18:52.480 ⇒ 00:18:53.950 Greg Stoutenburg: We talk all the time.
125 00:18:53.950 ⇒ 00:19:00.010 Michael Armstrong: Amazing. Well, I’d love it if you could surface, you know, our…
126 00:19:00.170 ⇒ 00:19:19.689 Michael Armstrong: our upcoming, like, our current e-commerce campaign, and see if they would be interested in having their story of success and how they’re using tech to, to increase efficiency and move from, you know, kind of insight, that window. So, yeah, I’d love it if this is
127 00:19:19.690 ⇒ 00:19:21.280 Michael Armstrong: beginning of a broader conversation.
128 00:19:21.280 ⇒ 00:19:21.710 Greg Stoutenburg: You’re out.
129 00:19:21.710 ⇒ 00:19:32.400 Michael Armstrong: And let me know how I can help in that regard. And if that’s… if they’re not interested, that’s okay, but I would love to at least see if their voice could be a part of this broader campaign.
130 00:19:32.400 ⇒ 00:19:35.230 Greg Stoutenburg: Yeah, sure, sounds great. Yeah, this is great, I appreciate it, Michael.
131 00:19:35.230 ⇒ 00:19:37.450 Michael Armstrong: It’s a wonderful… I appreciate it too. Take care of yourself.
132 00:19:37.450 ⇒ 00:19:39.110 Greg Stoutenburg: See ya. Bye.