Meeting Title: Eden - Brainforge: AI Command Center Weekly Check-in Date: 2026-04-13 Meeting participants: Pranav Narahari, Daniel, Adam P, Robert Tseng
WEBVTT
1 00:07:34.470 ⇒ 00:07:36.710 Pranav Narahari: Hello, hello, welcome back!
2 00:07:37.000 ⇒ 00:07:38.970 Daniel: What’s going on, man? How you doing?
3 00:07:38.970 ⇒ 00:07:41.150 Pranav Narahari: Nothing much, nothing much. How you been?
4 00:07:41.410 ⇒ 00:07:44.890 Daniel: Oh, good. Just hacking through all sorts of stuff here.
5 00:07:45.150 ⇒ 00:07:51.709 Pranav Narahari: Yeah, back to… back to work, right? I think you were out last… since… You came back last Tuesday?
6 00:07:52.210 ⇒ 00:07:56.809 Daniel: Well, yeah, and then it was straight into Health OS stuff, so we’ve been grinding on that.
7 00:07:57.110 ⇒ 00:07:59.520 Pranav Narahari: Yeah, yeah, that’s how it usually goes.
8 00:07:59.520 ⇒ 00:08:02.440 Daniel: Seem in a good spot, so we’ll take that and run with it.
9 00:08:03.180 ⇒ 00:08:03.840 Pranav Narahari: Yeah.
10 00:08:03.840 ⇒ 00:08:07.120 Daniel: able to get some of this set up, I think.
11 00:08:07.120 ⇒ 00:08:09.999 Pranav Narahari: Oh, cool. I was gonna ask you about that. Yeah.
12 00:08:10.000 ⇒ 00:08:14.510 Daniel: from what I can tell… What is this master, Astra thing?
13 00:08:14.980 ⇒ 00:08:22.910 Pranav Narahari: Yeah, so Master Studio is kind of the user interface that we’re just using for right now. And so…
14 00:08:23.100 ⇒ 00:08:29.570 Pranav Narahari: we’re gonna… I’ll talk to you a little bit about that, how we can, like, make it look a little prettier. There’s a lot happening, super busy, right?
15 00:08:29.600 ⇒ 00:08:48.770 Pranav Narahari: But Master is just, like, a really good scaffolding for backend AI agents. And so, we’ve used it on, like, a few other clients. It makes it really simple for just building out, like, these chat interfaces, and then also building, like, integrations with, like, different data platforms. So for us, like, we’re using it for…
16 00:08:48.810 ⇒ 00:08:53.210 Pranav Narahari: We’re using it for Slack, we’re using it for Google,
17 00:08:53.350 ⇒ 00:09:00.790 Pranav Narahari: And then built in, out of the box, they have this thing called Master Studio. So we just kind of deployed that so you guys could use it as well, just for, like, this POC.
18 00:09:01.240 ⇒ 00:09:02.550 Daniel: Brad, I like it.
19 00:09:03.140 ⇒ 00:09:03.990 Daniel: Yeah.
20 00:09:04.500 ⇒ 00:09:09.759 Daniel: So, I’m just diving right in. What’s up, Adam? What’s up, Robert?
21 00:09:09.760 ⇒ 00:09:10.869 Pranav Narahari: Yeah, how’s it going, guys?
22 00:09:10.870 ⇒ 00:09:16.490 Daniel: But I’ve, I know it’s been… Some backend setup here, and…
23 00:09:17.140 ⇒ 00:09:26.439 Daniel: I’m gonna take a step back for this call, but I also want to make sure I dive in on this, which is, I think I’m all connected, I’m,
24 00:09:27.490 ⇒ 00:09:35.039 Daniel: Working through some of the use cases, how do you prompt this thing, what’s it looking for, right? It doesn’t inherently appear to have
25 00:09:36.930 ⇒ 00:09:45.140 Daniel: a great ability to go, like, look at my doc, see who I am, isolate a strategy, it’s very prompt-driven, right?
26 00:09:45.610 ⇒ 00:09:49.970 Daniel: I.e, the instructions and use case are, this is for me.
27 00:09:50.260 ⇒ 00:09:54.440 Daniel: You know, as COO, like, dive through, read this stuff, and things like that.
28 00:09:54.950 ⇒ 00:10:13.229 Daniel: take a look at… we have this layer of all of this public and a little bit private, which is, you know, we need to work through information from the company to drive some results. So, in the first instance, I want to play around with this, but I want to reserve some of this because
29 00:10:13.340 ⇒ 00:10:20.760 Daniel: I’m gonna be honest, there have been some really big launches in AI over the past 2 weeks. I think this is kind of gonna be, like.
30 00:10:20.860 ⇒ 00:10:27.519 Daniel: par for the course. And so I’ve started some discussions with Adam McBride and some of the things he’s working on, because
31 00:10:28.230 ⇒ 00:10:40.190 Daniel: what I was worried about to some extent is already happening, which is we’re gonna go through these efforts, build this thing, and then there’s gonna be some tool that’s gonna come up with a new feature.
32 00:10:40.720 ⇒ 00:10:48.460 Daniel: And I’m gonna struggle with adoption on this, because I’ll give you an example. Right now, Claude Co-work is live. Okay. Adam and Josh are basically living in Claude Co-work. Like…
33 00:10:48.680 ⇒ 00:10:58.720 Daniel: I don’t even know what the chances are that I could pull them out of Claude Cowork to go ask this thing some questions and integrate… like, I know the next question is gonna be, why can’t my Claude just integrate with this thing, right?
34 00:10:59.170 ⇒ 00:11:09.980 Daniel: And so I want to make sure that what we build isn’t in a box, that we can actually get adoption, where we… I see what’s happening right now, which is everybody was using ChatGPT for everything.
35 00:11:10.090 ⇒ 00:11:28.930 Daniel: Claude launches, you know, then Gemini came up with, whatever it is, 3.1, or whatever their last solution was, and they’re like, oh, it’s the best reasoning in the market, and so, yeah, everybody jumps on and starts using Gemini. Now Claude comes up with Co-work, okay, everything’s an anthropic switch, we gotta go get a BAA so we can implement this on my, you know, Mac Mini that I’m running on this side.
36 00:11:29.860 ⇒ 00:11:54.699 Daniel: exactly what I thought was gonna happen is gonna happen, which is we’re gonna go through these systems, we’re gonna build out this thing. I know for a fact Adam is just going to use Claude until he decides to use something else based on some new feature release or whatever. So I want to take a moment… so two things, I’m diving in on this, but I also want to take a step back, and this is why I wanted this structure as, like, buckets of hours, because if we actually… we should actually have a candid conversation about
37 00:11:54.700 ⇒ 00:11:56.720 Daniel: okay, what is Claude Co-work?
38 00:11:56.720 ⇒ 00:11:59.440 Daniel: Are we missing the mark on this thing? Do we need to dive…
39 00:11:59.440 ⇒ 00:12:07.130 Daniel: you know, deeper into maybe this is a different setup process, and I know it just happened since we started this project, but it’s, like, worth talking about.
40 00:12:07.860 ⇒ 00:12:15.469 Daniel: So, where do we want to start? Do we want to start at that higher level, or do you want… we want to dive in on some of the app and sort of where we’re at in that project?
41 00:12:16.280 ⇒ 00:12:21.840 Pranav Narahari: Yeah, I have a few… so a few different thoughts come up as you were… as you were talking.
42 00:12:21.870 ⇒ 00:12:33.720 Pranav Narahari: So, Adam’s current workflow for Claude Co-work, I want to kind of understand that to see, is what we’re building with Mastra, and just what we’re deploying into GCP,
43 00:12:33.720 ⇒ 00:12:51.140 Pranav Narahari: Is it a step behind? Is it actually building something completely different, where there’s still that value that’s gonna last for weeks and months into the future? So that’s kind of what I want to understand a little bit further, like, where is Claude Co-Work fitting into the current workflow?
44 00:12:51.140 ⇒ 00:12:55.000 Pranav Narahari: So, maybe we can start there.
45 00:12:55.030 ⇒ 00:12:58.490 Pranav Narahari: I think, and then that will kind of give me more insight to, like, okay.
46 00:12:58.580 ⇒ 00:13:00.549 Pranav Narahari: Where do we… where do we go from here?
47 00:13:00.870 ⇒ 00:13:08.630 Daniel: So, to start it all off, this was not something we missed. Like, literally, you know, last week, or the week before.
48 00:13:08.770 ⇒ 00:13:14.620 Daniel: Anthropic rolls out its co-work platform, which is, quite frankly, amazing, right? Yes.
49 00:13:15.030 ⇒ 00:13:20.950 Daniel: So, it’s worth recognizing that that happened. They did not have that, and they did.
50 00:13:20.950 ⇒ 00:13:21.350 Pranav Narahari: Yes.
51 00:13:21.350 ⇒ 00:13:34.510 Daniel: Now that they have it, I’ve started diving into some tools and systems and recognizing some things that are really interesting. Number one, there’s already an existing full Slack-connected, you know, setup within this Anthropic co-work solution.
52 00:13:34.960 ⇒ 00:13:54.769 Daniel: That, ironically, was actually one of the biggest pieces I was worried about, because the Gemini tool and this foundation didn’t have a really simple way to access that layer. Well now, all of a sudden, we’ve got Anthropic that has native… or built-out integrations and connect… they call them connectors, but connectors with Slack.
53 00:13:55.080 ⇒ 00:14:01.859 Daniel: and with the workspace, thus solving my biggest issue prior, which is I had nothing that could see everything.
54 00:14:02.020 ⇒ 00:14:03.139 Daniel: Now, ironically.
55 00:14:03.140 ⇒ 00:14:03.660 Pranav Narahari: Yep.
56 00:14:04.120 ⇒ 00:14:05.650 Daniel: We can.
57 00:14:05.860 ⇒ 00:14:20.700 Pranav Narahari: Right. I remember being… that being, like, the first step that you’re like, a huge value add would just be having my Google Workspace integration and just adding Slack to that. Now, I totally agree with you on that point. If that’s, like, the end, then that’s perfect.
58 00:14:20.890 ⇒ 00:14:33.959 Pranav Narahari: But what we were kind of building here was, like, the idea, and we can… we can pivot, too. I think the idea was, with how we kind of scoped this, is that we want to move with how AI moves, which is, week to week, it changes.
59 00:14:34.510 ⇒ 00:14:48.320 Pranav Narahari: with, what we initially thought of was that we want to be able to alias across all the different users within Eden, so then we can have domain-wide understanding of, like, what is going on.
60 00:14:48.320 ⇒ 00:14:57.760 Pranav Narahari: Currently with that Slack integration, you’re still probably just using your own user ID to authenticate, right? You’re not authenticating on a…
61 00:14:58.210 ⇒ 00:15:14.559 Daniel: The biggest single difference I see is that the cowork setup, its integrations, give Adam McBride the access Adam McBride has in the messages and the threads that Adam McBride is on, thus making it very useful for him in function.
62 00:15:15.330 ⇒ 00:15:15.840 Pranav Narahari: Exactly, yeah.
63 00:15:15.840 ⇒ 00:15:18.580 Daniel: I have been convinced to the point where I actually think
64 00:15:18.910 ⇒ 00:15:27.390 Daniel: the strategy of, you know, I’ve been calling it augmented employees, it’s not really catching on, but the strategy of being able to integrate some of this stuff
65 00:15:27.520 ⇒ 00:15:44.899 Daniel: The tool that at least we’re seeing and using from Anthropic is probably the best that we’re gonna get to giving all of our employees access to something that has all of their stuff in it, can give them responses on what they’re doing, find their files, etc.
66 00:15:45.050 ⇒ 00:16:02.500 Daniel: Where I’m seeing the white space here is, okay, but what does that mean for cross-collaborative work function, these types of things? That’s where the layer doesn’t exist. And so my thought in my head was, we need to build out that layer that shows everybody’s use space and access and data. Whatever tool crawls it.
67 00:16:02.910 ⇒ 00:16:07.630 Daniel: That’s neither here nor there, it just needs to be somewhere and be accessible.
68 00:16:07.760 ⇒ 00:16:09.569 Daniel: And then number 2 is.
69 00:16:09.590 ⇒ 00:16:29.529 Daniel: I thought, well, should we integrate these just into Claude Co-Works so we see it? I actually think we shouldn’t. I think this should be a very particular tool used for very particular purposes. Again, not to dive… I don’t want to… I don’t want to dive into people’s emails. I don’t want that. I just want to see general overall directionality of the business and have some feedback.
70 00:16:29.530 ⇒ 00:16:34.099 Daniel: I think there’s a heavy amount of value there, but we’re in alignment, those two things are separate, right?
71 00:16:35.240 ⇒ 00:16:37.010 Pranav Narahari: Yeah,
72 00:16:37.450 ⇒ 00:16:49.930 Pranav Narahari: And I think here, like you said, the simplicity of it is kind of important, right? We don’t want to just copy what Claude Kowick is doing, because like you said, they spent a ton of time doing that for this, like, super broad purpose thing.
73 00:16:50.130 ⇒ 00:17:06.810 Pranav Narahari: I think what is super interesting for… for you specifically, that is still kind of a gap with the co-work setup, is that domain-wide data. For the entirety of Eden, like, yeah, we don’t need to dive into specific emails, and we can talk about what that,
74 00:17:06.970 ⇒ 00:17:23.920 Pranav Narahari: what that, you know, filtering looks like, and at what level do we want to detect themes? But a huge gap that I still see is, okay, even if Adam McBride is able to authenticate into his own Slack messages, what is he missing in conversations that he is not a part of?
75 00:17:23.960 ⇒ 00:17:34.359 Pranav Narahari: I was talking to Adam a little bit last week, and he was saying how, like, a lot of, like, work gets done in just, like, private conversations. Maybe people create DMs.
76 00:17:35.070 ⇒ 00:17:43.499 Pranav Narahari: And so… Slack has ways of… if you guys have, like, the Business Plus and Enterprise,
77 00:17:43.980 ⇒ 00:18:01.150 Pranav Narahari: the subscriptions, we can then pull in that data in… in a roundabout way, so we can take… we can take advantage of this specific use case, which is, you know, themes aren’t going to get detected just based on one user’s perspective of all the images… all the messages coming in.
78 00:18:01.610 ⇒ 00:18:20.669 Pranav Narahari: I think it is important to, if we’re going to talk about themes going on within Eden, it’s to get the context of all the different conversations that are happening across the organization. So, to me, that still seems like a very unique use case for this, command center. One other thing I also want to mention, too, is that
79 00:18:20.810 ⇒ 00:18:28.030 Pranav Narahari: it’s really important that we restrict access as well for this command center, right? We don’t want, you know, having all this information.
80 00:18:28.040 ⇒ 00:18:36.699 Pranav Narahari: We don’t want, then, people to be able to update certain things, or delete certain things, right? Even by accident.
81 00:18:36.700 ⇒ 00:18:50.830 Pranav Narahari: And so, a big part of, like, what me and Ad have been working on the last couple weeks is, like, I’ve been pushing back on, like, hey, let’s not have write access, let’s not have delete access. Let’s make sure everything is read, just so…
82 00:18:51.010 ⇒ 00:19:04.399 Pranav Narahari: for this tool, we can feel very safe about what we’re doing, which is, you know, we are pulling in data from very… the idea is to pull in data from various people’s accounts, right? To detect the themes. We had to be really safe about that.
83 00:19:05.550 ⇒ 00:19:10.500 Daniel: Yeah, and I’ve also had lots of conversations with Adam McBride about that, because…
84 00:19:13.560 ⇒ 00:19:17.800 Daniel: No matter how you build this and how you trust, like…
85 00:19:17.940 ⇒ 00:19:23.950 Daniel: There’s a chance to use this in ways that can breach trust between the company and its employees, too.
86 00:19:24.380 ⇒ 00:19:25.100 Pranav Narahari: I can see that, yeah.
87 00:19:25.100 ⇒ 00:19:29.779 Daniel: So… so I do want to be very careful about understanding that.
88 00:19:30.680 ⇒ 00:19:31.310 Pranav Narahari: Yeah.
89 00:19:31.490 ⇒ 00:19:37.019 Daniel: here’s where I’m at. I want to back up, and that’s why I started at the top of this, just discussing goals, so…
90 00:19:37.140 ⇒ 00:19:53.829 Daniel: I had two main goals. My two main goals were get the company up to level in terms of having access to AI tools that we should, given the technology and the level of investment we put into this, and the level of importance we place on automating and optimizing these systems.
91 00:19:53.830 ⇒ 00:20:04.100 Daniel: The second was building out some level… high-level command pulse through this thing, where we can get bird’s-eye views of projects moving through the company, etc. Those have always been my two clear goals.
92 00:20:07.150 ⇒ 00:20:20.960 Daniel: I thought this would be accomplished in one, i.e, if we have this system and solution that would also allow us to integrate Slack into Gemini and give everybody, basically, a chatbot that runs on their system with their calendar and all this kind of stuff.
93 00:20:21.050 ⇒ 00:20:26.350 Daniel: I have kind of changed my mind after seeing Claude Co-work. I honestly think it is so good, and…
94 00:20:26.390 ⇒ 00:20:43.140 Daniel: I thought there would be some feature launches, I thought… I think maybe if we waited 6 months, we’d be in a different situation, but right now, the tool is so functional and so good that the second piece of that is now becoming a separate and distinct project in my mind, wherein I really want to
95 00:20:43.270 ⇒ 00:20:49.740 Daniel: at least spend some time thinking about how can we safely give our employees access to clawed co-work.
96 00:20:50.090 ⇒ 00:20:55.529 Daniel: with safe levels of information and guardrails, given what’s out of the box already with Cloud Cowork.
97 00:20:56.250 ⇒ 00:20:58.389 Daniel: It’s a big enough deal that…
98 00:20:58.900 ⇒ 00:21:14.240 Daniel: I want everybody to be able to unlock those same tools, functionality, and the chats and integrations. It is remarkably powerful, and I see very little reason to wait on Google to launch a competitive solution here. What are your guys’ general thoughts on that?
99 00:21:16.140 ⇒ 00:21:17.709 Pranav Narahari: Yeah, I mean,
100 00:21:19.170 ⇒ 00:21:26.709 Pranav Narahari: And so, let me… let me just kind of reiterate and, like, my ideas on how we would kind of move going forward.
101 00:21:27.060 ⇒ 00:21:37.759 Pranav Narahari: building this Claude Co-work setup, as you’ve seen, like, Adam McBride is kind of… and you are pretty convinced, and I understand why you’re convinced, too, of this just being…
102 00:21:37.910 ⇒ 00:21:54.929 Pranav Narahari: a huge leap in productivity and workflows that people can be having at Eden, we should now share that with everybody at Eden, right? So, now what our focus should be on is, okay, how do we have that, like, last mile tailored experience for each one of the
103 00:21:55.180 ⇒ 00:22:00.290 Pranav Narahari: employees at Eden. So, now that’s something that we can look into.
104 00:22:00.290 ⇒ 00:22:10.380 Daniel: And I expected this would be, again, from a similar repository with similar connected, because there were areas that were not accomplished when we talked about this a month ago, that now are, right?
105 00:22:10.520 ⇒ 00:22:13.329 Daniel: And I’m realizing, shit, I’m really…
106 00:22:13.690 ⇒ 00:22:32.539 Daniel: thought it was gonna be the same project, now it’s really two projects… projects. It’s adoption of the, you know, Claude system safely across the company, and helping Adam and I, Adam P. and I, understand, like, how can we make sure people aren’t gonna just open us up to a huge amount of data exposure challenges, right? That’s a concern I have.
107 00:22:32.550 ⇒ 00:22:47.999 Daniel: We need to have a safe way to do that. Not everybody’s going to be running a Mac Mini in a separate environment on this side, so a strategy of, okay, that was built while we’ve been talking, so I want to get that live across the company. You guys understand our data architecture better than anybody.
108 00:22:48.160 ⇒ 00:23:04.469 Daniel: How can we do this safely? And I’m now seeing that that is becoming a separate project of high-level importance, not superseding this work on this side, but, like, that is half my goal, and it’s already built out of the box if we can safely give people access to it.
109 00:23:05.850 ⇒ 00:23:06.480 Pranav Narahari: Right, yeah.
110 00:23:06.480 ⇒ 00:23:07.609 Robert Tseng: Yeah, I’ll jump in here.
111 00:23:07.610 ⇒ 00:23:08.930 Pranav Narahari: Pretty straightforward. Yeah.
112 00:23:08.930 ⇒ 00:23:25.380 Robert Tseng: Yeah, so I agree, I agree with it’s two projects. So one is, like, kind of, it’s just… it’s provisioning and access. The surface, we’re no longer married to, like, Gemini. I think we were… even with the beginning of the project, we were kind of like, okay, whatever we end up standing up is going to be agnostic to whatever, like, tool you end up plugging in.
113 00:23:25.380 ⇒ 00:23:31.660 Robert Tseng: we, you know, Gemini was your preference to start with, and that’s what we kind of, like, were moving towards, but you’re basically, like…
114 00:23:31.660 ⇒ 00:23:40.940 Daniel: we had a BAA with Google. Now, Adam and I are… we are entering in conversations. Anthropic is ready to give us a BAA.
115 00:23:41.160 ⇒ 00:23:41.710 Robert Tseng: Yeah.
116 00:23:42.050 ⇒ 00:23:44.710 Daniel: So that’s a huge unlock that wasn’t available either.
117 00:23:45.180 ⇒ 00:24:02.970 Robert Tseng: Okay, yeah, so that’s… that’s fine. I think I… I think we’re… we can definitely just… we’re switching… switching the environment, Gemini to… to Cloud. But everything else under the hood, the work that Vernav and Adam have already been kind of, like, going… going through, that’s all kind of part of the rules and provisioning that we need to have, and so when we switch environments.
118 00:24:02.970 ⇒ 00:24:09.300 Robert Tseng: I mean, we’re gonna… there’s gonna be some discrepancy about what’s… what was achievable in… and, you know, Google…
119 00:24:09.670 ⇒ 00:24:22.419 Robert Tseng: in the Google workspace versus what’s in Anthropic, and we’ll tease that out shortly. But yeah, I think that makes sense to us. Like, I… I don’t think this decision comes as a surprise, we’re not caught off guard or anything, like.
120 00:24:22.420 ⇒ 00:24:35.770 Robert Tseng: you know, at Brainforge, we have already been building for, like… I mean, even before Cloud Code Work came out, we’ve been building on OpenWork, which is basically an open source version of Cloud Code. So we’ve been doing this for clients for 6 months already, so…
121 00:24:35.860 ⇒ 00:24:55.420 Robert Tseng: Yeah, I think, like, the Cloud Cowork kind of release, I mean, it’s great. It looks good, it looks better than the open source version. But yeah, I think… I think we… we understand how to, like, rotate these different platforms onto the same data foundations. So, I think, yeah, totally hear you on that.
122 00:24:55.530 ⇒ 00:25:12.430 Robert Tseng: Ben, I think what Pranav is saying around last mile, in terms of what needs to make this particular sprint, like, very functional for you, is that, like, you know, we do have a… we are a few steps ahead in terms of, like, doing the fine-tuning to get what the command center output should be.
123 00:25:12.790 ⇒ 00:25:19.270 Robert Tseng: I think, like, obviously, like, I mean, I’ve seen some of the outputs that Adam McBride has put out.
124 00:25:19.380 ⇒ 00:25:37.100 Robert Tseng: I mean, frankly, it’s caused more… more work than… than not on the data team when he’s pushed us to kind of go in certain directions, but it’s… it’s clear. It’s just, like, it’s a general intelligence platform. It’s… the product analytics roadmap tracking plan that he gave us was pretty much garbage. Like, I had to rip it up and, like.
125 00:25:37.100 ⇒ 00:26:00.890 Robert Tseng: redo, like, 70% of it. So, like, I don’t think it’s, like, going to replace, like, the work that we… that we do yet. Like, we still need the… the people that… here, that… that actually, like, know how to architect the thing. So, I mean, I think… but there will be plenty of other workflows that are totally replaceable… replaceable. So, I think, like, you know, there are going to be
126 00:26:00.890 ⇒ 00:26:03.390 Robert Tseng: like, certain… I mean, there are already, like.
127 00:26:03.390 ⇒ 00:26:14.540 Robert Tseng: There are limits out of the box for that environment, but as a general intelligence tool given to all people to raise, like, the baseline, totally agree that that is probably what
128 00:26:14.570 ⇒ 00:26:22.140 Robert Tseng: you know, that’s probably the best… best solution for your team. So if you can get that BA in place, we are… we are… we can totally switch the surface.
129 00:26:22.510 ⇒ 00:26:29.330 Daniel: Yeah, and I don’t want to commingle too many efforts. I am not expecting that…
130 00:26:29.600 ⇒ 00:26:33.849 Daniel: You know, Katie Kramer on our farm and ops team is gonna start
131 00:26:33.850 ⇒ 00:26:49.790 Daniel: pulling UTM data and trying to determine if there’s a cat… that’s not the point. The point is, she has a meeting coming up with, you know, Beluga, our external doctors network, wants to pull the last meeting notes and put together a high-level summary to discuss on her agenda. This is very basic stuff.
132 00:26:49.800 ⇒ 00:27:04.839 Daniel: Yeah. People should be doing in Gemini, but Claude unlocks a couple of things. Number one is it brings Slack into the fold through the connector. And number three, or sorry, number two here is that it has a much better ability to read CSV data and output visualizations in PowerPoint.
133 00:27:04.840 ⇒ 00:27:05.200 Robert Tseng: Yeah.
134 00:27:05.200 ⇒ 00:27:20.269 Daniel: Something that Gemini has always struggled with. And those two boxes are really simple. I’m just looking for employees to have a general intelligence platform to say, hey, I have a meeting coming up, can you pull up the meeting notes from the last stand-up, help put together my agenda, and then put that in a PowerPoint presentation? I can do that.
135 00:27:20.380 ⇒ 00:27:39.749 Daniel: if I have Slack in there, which now Claude, you know, does, and it’s chatbots, I could give all of my employees access to a general intelligence platform to architect their meetings, help them with some basic scheduling, note-taking, yeah, all this stuff now exists out of the box, which was half of my goal in going through these efforts.
136 00:27:39.750 ⇒ 00:27:42.709 Daniel: And so I’m thinking, instead of us…
137 00:27:42.710 ⇒ 00:27:54.989 Daniel: rebuilding everything from, you know, a different platform, waiting for Google Workspace to launch the next thing. Why don’t we just help my team have a best practices for connecting their Claude instance.
138 00:27:55.180 ⇒ 00:27:55.700 Robert Tseng: Yeah.
139 00:27:55.700 ⇒ 00:28:10.960 Daniel: do a little roadshow, make sure it’s adopted to the right folks, and they know how to, you know, type in their Slack credentials. We’re talking very basic shit. I’m not going up to my level and Adam DeBride’s level on playing with this stuff, or your guys, or even some of the deep MarTech details. I’m talking just for intercompany operations.
140 00:28:11.790 ⇒ 00:28:29.429 Daniel: Yeah. So, my thought is, let’s accomplish that right out of the gate, because we can accomplish it today. We don’t need to build anything. All we need is a PowerPoint deck and a little presentation to walk some of my employees through how to connect their Claude setup and get onboarded. You know, Adam, P, and I, and you guys can run that, but basically, it should be one
141 00:28:29.470 ⇒ 00:28:39.950 Daniel: half-hour meeting on connecting Clod, now play with it, it has access to your stuff. Like, all of a sudden, I’ve got my employees cranking out 15% higher productivity right out of the gate, just by getting set up on the tool.
142 00:28:41.690 ⇒ 00:28:55.350 Daniel: And the prioritization is not because I think that is inherently a more important effort, it’s an effort we don’t have to do any coding on, we just need to make sure we apply safe guardrails and tell people how to safely use and install this stuff.
143 00:28:55.840 ⇒ 00:28:56.880 Robert Tseng: Yeah. Yep.
144 00:28:58.420 ⇒ 00:28:59.690 Robert Tseng: Yeah, understood.
145 00:28:59.960 ⇒ 00:29:13.929 Robert Tseng: Yeah, and I think I should take that on. I mean, obviously, Bernav will support, but, like, you know, I would want Bernard’s time kind of focusing on, like, the more specific use case. So, like, I think that’s kind of how we can… I mean, it’s…
146 00:29:14.050 ⇒ 00:29:18.979 Robert Tseng: I wouldn’t even really call it a pivot, but, like, just more clarity on kind of how we move forward from here.
147 00:29:19.340 ⇒ 00:29:43.770 Daniel: Well, it really accomplishes the second part of what I want to do, which is get our employees to access their own tools and data. Eventually, I want maybe some other stuff to live in there. I don’t know how we could connect up something like a Paycom for our employees or whatever, but, like, we can get there, but that’s a big accomplishment in terms of getting my employees access to General Intelligence Tool, because I know right now we only have 16 Claude users, that’s not everybody, so… and the way they’re using it is generally almost
148 00:29:43.770 ⇒ 00:29:48.470 Daniel: almost none of the… like, there’s, like, 5 people who have any of their workspace connected into it.
149 00:29:48.690 ⇒ 00:29:57.869 Daniel: So, I know we’ve got a huge lift on productivity I can accomplish just by getting a general rollout of anthropic tools to my people.
150 00:29:58.490 ⇒ 00:29:59.120 Robert Tseng: Okay.
151 00:29:59.520 ⇒ 00:30:11.879 Daniel: So, I think there’s a little bit of protection we need to do for the BAA stuff, Adam, and then in addition to that, I think we just need to make sure, like, do a little bit of thought on what could they possibly break doing this, and make sure they don’t.
152 00:30:12.130 ⇒ 00:30:19.869 Daniel: So I don’t know if there’s, like, a corporate admin function in Claude to be able to make sure people don’t give right access and things like that to our internal.
153 00:30:19.870 ⇒ 00:30:20.340 Robert Tseng: Yeah.
154 00:30:20.340 ⇒ 00:30:21.910 Daniel: sure how that all works.
155 00:30:21.910 ⇒ 00:30:40.960 Adam P: Yeah, and I’m gonna… I haven’t played with Cowork at all yet, so I’m gonna jump in and play with it, too. I think it’s gotta be limited to the user’s permissions at the end of the day, right? So, as long as the users themselves aren’t elevated in the tools they’re signing into, then they can only cause as much damage as they could have caused anyway, but I’m just gonna…
156 00:30:40.960 ⇒ 00:30:42.689 Robert Tseng: Just going to all those tools themselves, yeah.
157 00:30:42.690 ⇒ 00:30:48.199 Adam P: Right. Yeah. But yeah, I’ll poke around and make sure that that statement’s true, and go from there.
158 00:30:48.370 ⇒ 00:31:05.830 Daniel: Yeah, and again, my background reasoning on this isn’t that it’s better than us building, or anything to do with that. It’s just, like, available today, and people are starting to use it, and it’s functioning well, I’d love to put some gasoline and just immediately get folks sort of trained up on the right way to use it, so they don’t end up doing something really crazy. I’m watching Brad build up, like.
159 00:31:06.090 ⇒ 00:31:15.939 Daniel: different engines on top of his access layers, all the way down to our, database on orders, and using, like, a Google tool called,
160 00:31:16.120 ⇒ 00:31:17.840 Daniel: Air… something?
161 00:31:18.080 ⇒ 00:31:19.330 Adam P: Oh, yeah.
162 00:31:19.330 ⇒ 00:31:19.760 Pranav Narahari: But…
163 00:31:20.140 ⇒ 00:31:25.889 Adam P: Yeah, no, crap. It’s sort of like… it’s sort of like AIIDE, what’s it called?
164 00:31:26.380 ⇒ 00:31:30.099 Adam P: Man, I can’t remember now. But yeah, he loves that.
165 00:31:30.100 ⇒ 00:31:44.429 Daniel: super robust system that’s linked directly to our database, and I’m, like, not exactly sure how cool I am with that. I mean, maybe it’s non-issue, maybe it’s just read-only, but, like, I’m realizing if we don’t give people a playbook on how to do this, they’re gonna go do it themselves.
166 00:31:44.430 ⇒ 00:31:51.850 Daniel: And they are not experts in data architecture, so I have no idea what they’re gonna do in this process, and right now, it’s a free-for-all across the board.
167 00:31:52.290 ⇒ 00:31:53.050 Robert Tseng: Yeah.
168 00:31:54.310 ⇒ 00:31:54.700 Daniel: So…
169 00:31:54.700 ⇒ 00:32:01.270 Adam P: Yeah, and yeah, Dan, to the same point, like, other things I’ve seen, it is getting to a free-for-all fashion of, like.
170 00:32:01.400 ⇒ 00:32:04.900 Adam P: People are not coming to me asking for, you know, a specific
171 00:32:05.010 ⇒ 00:32:12.860 Adam P: access to whatever, because the tool they’re using told them they need this right access, when I’m… and… but if you look into it, it’s like, no, you don’t.
172 00:32:12.860 ⇒ 00:32:15.629 Robert Tseng: It’s, it’s… I was just telling them to ask for too much, yeah.
173 00:32:15.630 ⇒ 00:32:16.330 Adam P: Yeah.
174 00:32:16.740 ⇒ 00:32:35.630 Daniel: Now I’m like, okay, in the absence of a playbook on how to do this stuff, they’re gonna use their own tool, their own setup, they’re gonna Google it all, they’re gonna ask these tools what they need to get connected to, and build out this monster of a system on who knows what data drawing what business outcome from an unclear, you know, system setup. I’d rather.
175 00:32:35.630 ⇒ 00:32:36.120 Robert Tseng: Yeah.
176 00:32:36.120 ⇒ 00:32:43.069 Daniel: playbook into Claude, because I think everybody will be generally satisfied with the functionality of that general intelligence tool.
177 00:32:43.430 ⇒ 00:32:44.080 Robert Tseng: Yeah.
178 00:32:44.320 ⇒ 00:33:02.670 Robert Tseng: So… Well, so I’m not gonna talk you out of the Cloud thing, but I think this is why we’ve been offering the open source, like, custom build for enterprises, like, bigger than Eden, because in the… in that environment, you have the controls tighter, you can see every query that’s being run, like, whereas Claude, it’s like, you know, you don’t really get to control
179 00:33:02.670 ⇒ 00:33:07.090 Robert Tseng: That much, and so when we’re working with other clients that, like, want to have
180 00:33:07.140 ⇒ 00:33:30.530 Robert Tseng: don’t want people stepping out of their lane. They prefer to, like, build their own version of Cloud Cowork instead of actually using Cloud Co-Work. Because, yeah, every user’s, like, search history, or, like, how they use the tool is all being logged, and you can see all of that, and you can use it to retrain your systems, send out alerts that, like, hey, you’re, like, crossing the line here and there, whatever.
181 00:33:30.530 ⇒ 00:33:48.729 Robert Tseng: And so, like, that’s… at an enterprise… as an enterprise offering, we’re seeing that that’s what the market wants, and, like, yeah, they are freaking out about, like, oh, no, I don’t want my team… half my team going to Cloud Cowork, the other one going to Google, or whatever, like, going all these other tools, where all these third-party platforms
182 00:33:48.730 ⇒ 00:33:59.759 Robert Tseng: are all… they don’t really share back, like, how users are using, using, the… the… like, the… the queries, or using the… using the reasoning model. So, I think, like.
183 00:33:59.760 ⇒ 00:34:04.400 Robert Tseng: this, this, like, kind of free-for-all situation, like, I… we… I told…
184 00:34:04.410 ⇒ 00:34:08.589 Robert Tseng: Yeah, I think that’s what we’re seeing. So, you know, I would just, you know.
185 00:34:08.610 ⇒ 00:34:17.539 Robert Tseng: keep that in the back of your mind, that, like, I think that’s why the build-your-own version of Cloud Co-work, is, like, an approach that a lot of companies are taking these days.
186 00:34:17.860 ⇒ 00:34:36.920 Daniel: And I fully respect that. I also have to worry about adoption, though, right? Because I am not gonna be able to tell everybody at this company, you can’t use an anthropic tool. I’ve already tried, and not one, right? That’s not gonna work. So, what I need to do is try and at least instill basic guardrails towards, okay, nobody’s begging me for chat GPT access, just for the record.
187 00:34:37.110 ⇒ 00:34:43.080 Daniel: But right now, it’s Anthropic tools that everybody is going to install, installing anyway, even.
188 00:34:43.080 ⇒ 00:34:43.550 Robert Tseng: Yeah.
189 00:34:43.550 ⇒ 00:34:48.249 Daniel: poking through this process, and they’re using the Claude interface. They’re not…
190 00:34:48.449 ⇒ 00:35:00.320 Daniel: looking for an alternate solution, so I’m like, okay, at the very least, what I can do is have a best practices for your Claude co-work setup at Eden, and run with that. That is better than a free-for-all. It is not perfect.
191 00:35:01.130 ⇒ 00:35:01.480 Robert Tseng: Yeah.
192 00:35:01.480 ⇒ 00:35:17.589 Daniel: But if they’re at least using an Enterprise Claude instance, and they set it up using best practices and commit to doing so, it’s part of our employee handbook and rules, then at that point, it falls to the same level of delegation of my employee opening their computer at a coffee shop for access, you know what I mean? Like…
193 00:35:17.590 ⇒ 00:35:18.160 Robert Tseng: Yeah.
194 00:35:19.250 ⇒ 00:35:35.140 Daniel: That’s the least we can do. So that’s one deliverable that I think accomplishes a lot of that. That was a long soliloquy that’s taken away from this, but the reason is because I’m looking at the business objective, which is enabling general intelligence tools for my employees. That one is something we can actually solve in a short-term.
195 00:35:35.140 ⇒ 00:35:35.540 Robert Tseng: Yeah.
196 00:35:35.540 ⇒ 00:35:48.010 Daniel: It’s here if we just get a BAA with Anthropic, get a separately, Adam, whatever we need to do for those security requirements, everybody’s using in the Enterprise version. If we get that done, we can tell everybody, have at it. Here’s Claude.
197 00:35:48.130 ⇒ 00:35:58.749 Daniel: we’re gonna give you access to what you’re gonna get access to, here’s how you need to use it, don’t do these things. Then, at the very least, we have a controlled environment where that half of my business objective is accomplished.
198 00:35:59.460 ⇒ 00:36:00.040 Robert Tseng: Yeah.
199 00:36:00.820 ⇒ 00:36:01.480 Daniel: So…
200 00:36:01.700 ⇒ 00:36:07.529 Daniel: Taking a step back, saying, that’s that second half. I thought it was gonna come after we got all the connected tools.
201 00:36:07.530 ⇒ 00:36:08.370 Robert Tseng: Yeah, but it’s happening.
202 00:36:08.370 ⇒ 00:36:27.640 Daniel: ironically come before, so I just want to make that clear. Now we can jump back to the first piece, which is how can we have general intelligence over the workspace as a whole to see ongoing improvements, and what does this MCP layer data setup that we have, what is that going to enable us to do as soon as we get some of these other functions and connectivity there?
203 00:36:30.420 ⇒ 00:36:34.490 Pranav Narahari: Yeah, and so for that part of things, what we’re building is…
204 00:36:34.720 ⇒ 00:36:52.040 Pranav Narahari: is still, I think, in the right direction. Like Robert said, like, these are kind of two things that we can kind of do in parallel. Robert, yeah, I mean, pull me into, like, what you need from there. I think I can probably help you with, like, the research in terms of best practices, how we can put in the right guardrails. But yeah, then hopping back into this,
205 00:36:52.070 ⇒ 00:36:55.429 Pranav Narahari: we are still, yeah, still in a good position. What…
206 00:36:56.140 ⇒ 00:37:02.230 Pranav Narahari: So what I kind of wanted to do here is… and Adam, it sounds like you’re already kind of set up with your instance. You were able to, like.
207 00:37:02.350 ⇒ 00:37:09.590 Pranav Narahari: Do certain testing, with, like, the Slack integration and Google, or…
208 00:37:09.810 ⇒ 00:37:13.470 Adam P: I was in… like I said earlier, like, it’s kind of…
209 00:37:13.910 ⇒ 00:37:21.910 Adam P: I didn’t really have any good use cases in mind of how to really tax it or whatever, so I was generally querying or asking it, you know, how to
210 00:37:22.010 ⇒ 00:37:28.669 Adam P: a certain info, like that one example I gave you, it was, you know, who do I talk to the most on Slack? Sort of those, like.
211 00:37:28.830 ⇒ 00:37:40.339 Adam P: not exactly a very one-and-done query, but more of, like, it would have had to have holistically looked at everything to then respond to me. But yeah, those are kind of things, the test I was running through it.
212 00:37:40.900 ⇒ 00:37:53.359 Pranav Narahari: Right. So, yeah, Danny, I remember, like, in the… when we were first talking about this project, like, how are we gonna assess if Slack is actually bringing in the context in the right way? You know, not just arbitrarily pulling in certain messages.
213 00:37:53.360 ⇒ 00:38:04.469 Pranav Narahari: this is, I think, the point where it would be really helpful to, like, understand, and maybe you give me, like, a bank of questions that you would like to be asking this command center, and getting responses that are valuable.
214 00:38:04.470 ⇒ 00:38:10.160 Pranav Narahari: And then, now we can test against that. Because, yeah, to Adam’s point, it’s like, yeah, we can…
215 00:38:10.160 ⇒ 00:38:28.389 Pranav Narahari: right now, where we’re at is, like, I just wanted to test that integrations are being made. You’re able to get the data on all these edge cases, like, okay, what was my last message? What’s… what topics are being talked about in certain channels on, maybe, like, for engineering, marketing, etc.
216 00:38:28.390 ⇒ 00:38:33.490 Pranav Narahari: But now, okay, what are actually gonna be the valuable insights that we get? So, Danny, like, if you can…
217 00:38:33.490 ⇒ 00:38:39.060 Pranav Narahari: Give us a bank of those questions, then we can refine this, like, command center.
218 00:38:39.060 ⇒ 00:38:47.880 Daniel: And this is what’s super helpful in understanding that second piece, which is why I want to do it, because at this point, I have access to my Slack. I can help me.
219 00:38:47.920 ⇒ 00:39:01.810 Daniel: So now, it removes that requirement from this project entirely, because now we can step back and say, what are workspace-wide initiatives that we need to have some access and some understanding of? Perfect.
220 00:39:02.040 ⇒ 00:39:08.540 Daniel: So, I can make it really clear, right? Eventually, I’d want this to help me identify bottlenecks.
221 00:39:08.710 ⇒ 00:39:25.910 Daniel: And so, in essence, I would want to say something to the effect of, what is the blocker on our lead-up project for Eden Pharmacy, right? And I want enough context that this can understand a message for needing a call with such and such has been brought up 4 times.
222 00:39:25.950 ⇒ 00:39:34.040 Daniel: Right? They’re waiting on this. Like, what is the context layers we would need for me to ask it, what is the bottleneck with X project? Right?
223 00:39:34.040 ⇒ 00:39:35.070 Pranav Narahari: Yes, exactly.
224 00:39:35.070 ⇒ 00:39:38.549 Daniel: what is the context I would need to ask it, you know.
225 00:39:38.640 ⇒ 00:39:50.729 Daniel: hey, what are the latest updates on the 2026 growth plan for telehealth? And it can recognize and identify what these things mean.
226 00:39:50.730 ⇒ 00:40:01.490 Daniel: And so, when I’ve been searching it, and, you know, I’ve tried a little bit, it’s really specific when I know the exact project I’m asking for, and it’s… it doesn’t even… actually, it doesn’t even have to be overly specific.
227 00:40:01.490 ⇒ 00:40:02.160 Daniel: So I’m asking
228 00:40:02.750 ⇒ 00:40:12.909 Daniel: Where’s the Health OS project in terms of status? And it will pull for me all of the last 9 context messages to set up status of that project.
229 00:40:13.530 ⇒ 00:40:14.100 Daniel: It’s actually.
230 00:40:14.100 ⇒ 00:40:14.460 Pranav Narahari: Absolutely.
231 00:40:14.460 ⇒ 00:40:17.559 Daniel: Cool. It’s almost 2 in the weeds.
232 00:40:18.260 ⇒ 00:40:18.610 Pranav Narahari: Yes.
233 00:40:18.610 ⇒ 00:40:25.069 Daniel: I’m thinking of that, right? I don’t really care about the message KDK sent about… I mean, I’ll see what it shows.
234 00:40:25.440 ⇒ 00:40:36.110 Daniel: You know, it’s pulling this and saying, Katie Kramer noted the sync visit feature was launched, clearing all sync errors. Actually, it’s honestly pretty good when I know the project I’m asking.
235 00:40:36.230 ⇒ 00:40:37.110 Daniel: But I’ve tried.
236 00:40:37.110 ⇒ 00:40:38.120 Pranav Narahari: Yes, but that’s…
237 00:40:38.120 ⇒ 00:40:49.099 Daniel: that are a little less, like, a little less context? What are the three largest projects the company is working on right now in terms of message blocks and team members involved? And I don’t get a response.
238 00:40:49.280 ⇒ 00:40:52.580 Daniel: The bot thinks, thinks, thinks, thinks, thinks, and then has no output.
239 00:40:53.730 ⇒ 00:41:03.629 Pranav Narahari: Okay, yeah, so… couple things. Adam also mentioned a specific situation, which I think, Adam, you asked, like, who am I talking to the most, right?
240 00:41:04.010 ⇒ 00:41:04.440 Adam P: Yep.
241 00:41:04.440 ⇒ 00:41:19.689 Pranav Narahari: These are going to be questions that are going to use up a lot of the context window, and this is a very raw prototype that we kind of built, right? Just so we can make sure that you guys have the integrations. And so this is good to know, because,
242 00:41:19.860 ⇒ 00:41:23.579 Pranav Narahari: We need to make sure that the context window is at a certain point, where, like, it could…
243 00:41:23.740 ⇒ 00:41:30.909 Pranav Narahari: answer these types of questions that, you know, Danny… Danny just mentioned. Another thing, too, is, yeah, we don’t…
244 00:41:30.910 ⇒ 00:41:38.219 Daniel: I wouldn’t… so what’s different about this and another bot, which basically is gonna force itself to answer the question, right?
245 00:41:39.350 ⇒ 00:41:50.470 Pranav Narahari: Yeah, so it has, like, a hard cap. It’ll only go as deep as it has context window for. What I did here was just, like, I want you to make sure, like, you can just get all the integrations in, but…
246 00:41:50.610 ⇒ 00:41:59.809 Pranav Narahari: the benefit of what we’re building here is that we define all these parameters. It’s not just some arbitrary platform that defines it for you.
247 00:41:59.860 ⇒ 00:42:17.479 Pranav Narahari: And so that’s good to hear, because I didn’t want to cap us, right? Because we… there’s a lot of models now that have crazy context windows, that we can take advantage of, specifically because we’re talking about themes. We’re not just talking about, like, hey, what was my last message? We’re talking about, like, hey, there’s maybe…
248 00:42:17.510 ⇒ 00:42:23.299 Pranav Narahari: 20 or dozens of people working on a single project. Get me all of that context, and
249 00:42:23.370 ⇒ 00:42:26.469 Pranav Narahari: create themes based off of that.
250 00:42:26.810 ⇒ 00:42:34.090 Pranav Narahari: Okay, so this is actually really good insight. I’m realizing a big use case for you, Danny, is, like, just understanding, okay, what
251 00:42:34.220 ⇒ 00:42:39.650 Pranav Narahari: If you ask about the project, you’re able to get that insight. However, if…
252 00:42:39.810 ⇒ 00:42:47.180 Pranav Narahari: you just want to know about the projects, or know about the bottlenecks, it is doing a bad job at that. Is that kind of your experience?
253 00:42:47.380 ⇒ 00:42:56.650 Daniel: So my experience is, if I ask… so, for example, there’s not a ton of projects in the company that we have this, but HealthOS became…
254 00:42:57.800 ⇒ 00:43:01.849 Daniel: There’s a couple of reasons. It’s super accurately pulling stuff.
255 00:43:02.200 ⇒ 00:43:07.559 Daniel: On, on, on… oh my god, just give me one second.
256 00:43:13.830 ⇒ 00:43:16.570 Daniel: It’s super accurate on stuff that, A,
257 00:43:17.100 ⇒ 00:43:28.269 Daniel: how can I phrase this? It’s really accurate on HealthOS. So there’s something about that project that’s working really well with this setup. There’s a couple of factors, I think, that are involved. Number one.
258 00:43:28.500 ⇒ 00:43:31.939 Daniel: The data only extends to about 6 weeks back.
259 00:43:32.570 ⇒ 00:43:51.370 Daniel: That’s when that project was introduced, right? This isn’t something that’s been going on forever at the company, where there’s a ton of context needed. It’s only been introduced last 6 weeks, and it’s been introduced at a very high velocity. So there’s a ton of communications, and they’re in channels called, like, HealthOS Launch Squad, HealthOS, you know, Dev Team.
260 00:43:51.370 ⇒ 00:43:52.080 Pranav Narahari: Gotcha.
261 00:43:52.080 ⇒ 00:43:57.969 Daniel: So all of a sudden, it has this really tight window with very clear stacking of data.
262 00:43:58.570 ⇒ 00:43:58.920 Pranav Narahari: Right.
263 00:43:58.920 ⇒ 00:44:18.320 Daniel: That then is spitting back really accurate examples. Here’s a summary of what’s happening in that channel to date, here’s a summary of what’s happening in this channel to date, and it’s literally pulling them by channel by channel. Here’s the Health OS Launch Squad channel, here’s the Health Dev channel, right? It’s going into detail across these channels with this HealthOS MX concept, right?
264 00:44:18.320 ⇒ 00:44:22.640 Daniel: The second thing I think which is interesting about it is it’s…
265 00:44:22.640 ⇒ 00:44:29.889 Daniel: There is no real background context, so it’s not telling me anything about what HealthOS is, really.
266 00:44:30.060 ⇒ 00:44:35.430 Daniel: It’s summarizing the conversation topic, and then giving me the latest notes about it.
267 00:44:35.440 ⇒ 00:44:49.950 Daniel: When I ask it about the LIDA project, that’s a very interesting contrast. The LIDA project is also sometimes called the NM Project, or the Economic Development Bond Project, right? It’s got several title names.
268 00:44:49.950 ⇒ 00:44:59.359 Daniel: But it’s basically this project we’re working on with the state is giving us a bunch of money for our pharmacy down in New Mexico. It is over many channels, not just one channel.
269 00:44:59.710 ⇒ 00:45:02.580 Daniel: And it’s been going on for almost a year.
270 00:45:03.720 ⇒ 00:45:07.830 Daniel: That one, it’s just spinning out and not giving me an answer to where that’s at.
271 00:45:08.850 ⇒ 00:45:12.310 Pranav Narahari: so much context, right? It’s kind of that context window thing, and so…
272 00:45:12.310 ⇒ 00:45:19.590 Daniel: huge communicated project. There’s only a handful of emails and messages about it, but it still can’t feel that context well.
273 00:45:20.580 ⇒ 00:45:37.509 Pranav Narahari: So there’s a couple things here. What you mentioned about just, like, just summarizing conversations, it’s because it doesn’t have a backbone to understand what is the… what are these projects, what’s a high-level summary of what is… people are talking about, right? And so that’s something that.
274 00:45:37.870 ⇒ 00:45:41.949 Daniel: it knocks it out of the park on that. Can I… can I do the share screen on this?
275 00:45:41.950 ⇒ 00:45:42.560 Pranav Narahari: Yeah.
276 00:45:42.760 ⇒ 00:45:53.729 Daniel: it knocks it out of the park on that. I mean, when we’re sharing this, it’s like, okay, you had the OptiRx manual processing, those 20 transmitted orders are being clarified and processed, tracking is expected, like, this is super detailed.
277 00:45:54.060 ⇒ 00:46:08.179 Daniel: Like, this is exactly what I would hope in terms of, like, give me a project breakdown. Five orders required. Okay. I mean, this just summarized all of this and showed which JIRA tickets it’s linked to in terms of the scrum planning. Like, this is amazing.
278 00:46:09.870 ⇒ 00:46:15.460 Daniel: Super detailed analysis, high level, on where that project is, but this is the only one that got really right.
279 00:46:16.340 ⇒ 00:46:18.479 Pranav Narahari: Yeah. Yeah, I mean, this is…
280 00:46:18.830 ⇒ 00:46:23.579 Pranav Narahari: this is a good case study, then. And this is the one that, like.
281 00:46:24.510 ⇒ 00:46:30.559 Pranav Narahari: You know what, I’m interested to see, too, because how I built this out, too, wasn’t necessarily…
282 00:46:30.880 ⇒ 00:46:33.039 Pranav Narahari: It was a lot of…
283 00:46:33.450 ⇒ 00:46:45.400 Pranav Narahari: what I’m… what I’m realizing now is, like, a big bottleneck for this implementation is the amount of context and the memory that we haven’t fully implemented yet for this command center. And so.
284 00:46:45.440 ⇒ 00:46:49.509 Pranav Narahari: I’m going to do some testing with the… with the team on…
285 00:46:49.520 ⇒ 00:47:07.879 Pranav Narahari: this second prompt to see, is it because, you know, a lot of this chat thread context window was just absorbed by that first question, which could very much be the case. So like, you know, if you refreshed, ran it again, do you see the same issue? Because then, you know, the context window gets cleared?
286 00:47:08.470 ⇒ 00:47:11.629 Pranav Narahari: I’m gonna look into that.
287 00:47:12.150 ⇒ 00:47:18.129 Pranav Narahari: there’s a few different things here, and I think now that I have, like, this bank of questions, I can… I can dive deeper.
288 00:47:18.420 ⇒ 00:47:27.709 Daniel: And I can give you a clear bank of questions. Like, what I’ll go through is I’ll keep… I’ll keep pushing on this. This isn’t… I haven’t done weeks of testing. I set it up this morning, so, you know, I’m doing.
289 00:47:27.710 ⇒ 00:47:28.230 Pranav Narahari: Yeah.
290 00:47:28.230 ⇒ 00:47:31.050 Daniel: But what I’m also recognizing in what you’re saying.
291 00:47:31.290 ⇒ 00:47:45.269 Daniel: This tool is only valuable if we’re able to get a bird’s-eye perspective of that lens, right? So it’s kind of this odd thing where it’s like, okay, don’t worry, it’ll get better as it corrects memory and does this. There’s actually already a ton of memory and backup data
292 00:47:45.380 ⇒ 00:47:52.660 Daniel: We just need the ability for it to have enough tokens to synthesize that into… like, this summary is beautiful.
293 00:47:52.800 ⇒ 00:48:08.110 Daniel: And so what I’m thinking is it was just only had to go 6 weeks back, it was only 4 or 5 channels, this didn’t really crawl my email, it appeared, right? But maybe it did. But this context is… this is, like, an amazing expression. I would have given this expression in terms of that.
294 00:48:08.110 ⇒ 00:48:08.880 Pranav Narahari: And so this is…
295 00:48:08.880 ⇒ 00:48:09.560 Daniel: really involved in it.
296 00:48:09.560 ⇒ 00:48:24.280 Pranav Narahari: This is one thing that I was thinking about, too, is that there is going to be for… there’s going to be sprawling data from longer than just 6 weeks ago, right? Like, for that other project that you mentioned, over 1 year back, even though there’s maybe
297 00:48:24.280 ⇒ 00:48:31.730 Pranav Narahari: really few emails going on, or maybe there’s, like, a ton of different channels, but just a few messages per channel.
298 00:48:31.840 ⇒ 00:48:36.460 Pranav Narahari: I think we need to have a more tailored experience than just this master studio.
299 00:48:36.830 ⇒ 00:48:44.910 Pranav Narahari: We need to have a… maybe potentially a prompt library that has the correct… and, you know, let me also,
300 00:48:45.680 ⇒ 00:48:55.670 Pranav Narahari: I can share this… this other UI mockup that I had that I think maybe better fits the use case that you would have, Danny, for this type of,
301 00:48:55.850 ⇒ 00:49:01.569 Pranav Narahari: For… for the command center. But you just… you just ran a prompt. Do you want to kind of go over that, or…
302 00:49:02.460 ⇒ 00:49:19.450 Daniel: I’m just continuing to ask it, like, there’s this HR reset thing that Jared is working on, right? But he’s called it, like, HR Month of Shitty Stuff. He’s called it, like, he’s got all these names, and it doesn’t quite have enough context to recognize, because I know we had a really big meeting on this.
303 00:49:19.650 ⇒ 00:49:22.730 Daniel: Yesterday, and we had a ton of notes surface from it.
304 00:49:22.860 ⇒ 00:49:30.970 Daniel: Right, but it’s not finding that. So, I’m curious, what is the overall context window that we need? How do we prioritize things like
305 00:49:30.970 ⇒ 00:49:43.219 Daniel: like, meeting notes in a way that helps give it context to search the Slack channels and doc creation. Like, balancing some of where it searches first for context might be a really fruitful exercise.
306 00:49:43.220 ⇒ 00:49:45.990 Daniel: Because most of this comes in person, like.
307 00:49:45.990 ⇒ 00:50:05.220 Daniel: from our Google Meet meetings, we have objectives, like, okay, here’s the HR thing, and then with that context, maybe it’ll have an ability to go narrower and deeper on that context. Exactly. I’m just trying to think of how to architect which layer it searches first, as opposed to just telling it, go right to Slack, because that’s what it’s doing right now.
308 00:50:05.220 ⇒ 00:50:06.040 Daniel: I mean, you can literally see.
309 00:50:06.040 ⇒ 00:50:06.630 Pranav Narahari: Right.
310 00:50:06.630 ⇒ 00:50:10.080 Daniel: read Slack, search public and private, right?
311 00:50:10.080 ⇒ 00:50:10.480 Pranav Narahari: Yep.
312 00:50:10.480 ⇒ 00:50:16.479 Daniel: and run GWS, and I’m realizing, oh shit, if that was flipped, and it had our main notes first.
313 00:50:16.620 ⇒ 00:50:20.340 Daniel: Then it could go search Slack for the context related to those projects.
314 00:50:21.250 ⇒ 00:50:40.589 Pranav Narahari: Yeah, and so how a lot of, like, when we build, like, these chat interfaces for clients, like, prompt libraries… I don’t know if you’ve heard of, like, the concept of prompt libraries, it’s basically just, like, a one-click. It already has, like, the tools defined on what it would be using, so let’s say…
315 00:50:41.210 ⇒ 00:50:57.470 Pranav Narahari: Yeah, you don’t care for it to look through Slack first. Maybe that’s gonna be too much context to go through at first. The most relevant, biggest value-add context is gonna be coming from the meeting notes. Okay, then for that specific prompt, just look at the meeting notes first.
316 00:50:57.500 ⇒ 00:51:11.769 Pranav Narahari: And then what you can have is, like, additional toggles or additional, like, follow-up prompts that are… that are basically dive deeper, look into this area, look into that area. It’s basically giving additional context,
317 00:51:11.790 ⇒ 00:51:18.980 Pranav Narahari: But you can drive it a little bit further. Right now, it’s really driving a lot more than
318 00:51:23.350 ⇒ 00:51:24.030 Pranav Narahari: Interesting.
319 00:51:24.030 ⇒ 00:51:25.120 Daniel: Pretty good updates.
320 00:51:25.500 ⇒ 00:51:26.360 Pranav Narahari: Yeah.
321 00:51:27.050 ⇒ 00:51:28.579 Daniel: all from Slack, right?
322 00:51:29.410 ⇒ 00:51:33.689 Pranav Narahari: Yep. Yeah, and it’s… it is doing that preference, right? For whatever reason, it’s…
323 00:51:33.880 ⇒ 00:51:43.399 Pranav Narahari: calling Slack first. And so, we can update that, because it’s sounding like, you know, meeting notes are gonna be, like, the biggest value add to…
324 00:51:44.200 ⇒ 00:51:45.849 Pranav Narahari: To, like, the responses.
325 00:51:49.390 ⇒ 00:51:51.680 Adam P: Yeah, that response was, like…
326 00:51:51.900 ⇒ 00:51:56.260 Adam P: Not all… all three of those bullet points don’t apply to the command center project.
327 00:51:56.700 ⇒ 00:51:57.490 Daniel: Just one.
328 00:51:57.640 ⇒ 00:51:58.470 Adam P: Just one of them.
329 00:51:58.470 ⇒ 00:51:59.330 Pranav Narahari: That’s the top one, yeah.
330 00:51:59.330 ⇒ 00:52:01.969 Adam P: But the other ones are, like, AI included.
331 00:52:02.220 ⇒ 00:52:02.950 Daniel: But I’m realizing.
332 00:52:02.950 ⇒ 00:52:03.800 Pranav Narahari: Because AI…
333 00:52:03.800 ⇒ 00:52:05.329 Daniel: with the context.
334 00:52:05.720 ⇒ 00:52:11.630 Daniel: searching there is, like, predefining that, so I’m curious if… I wanted to try one and then flip it.
335 00:52:14.850 ⇒ 00:52:21.739 Pranav Narahari: Yeah, basically with, when I was talking about that backbone, it’s basically when there’s certain ambiguous
336 00:52:21.860 ⇒ 00:52:27.579 Pranav Narahari: terms, like AI Command Center, how is it supposed to know that AI Command Center isn’t just, like, a…
337 00:52:27.700 ⇒ 00:52:47.420 Pranav Narahari: an AI initiative, any type of project, right? Because you could think of Claude Cowork as an AI command center. It may not necessarily define what we’re working on here as the AI command center. And so with that backbone, it’s essentially a knowledge base that it can refer to as a source of truth of projects, initiatives.
338 00:52:47.420 ⇒ 00:52:53.669 Pranav Narahari: topics that are going on at Eden. And so, we just need to build that out to give this a little bit more direction.
339 00:52:53.670 ⇒ 00:53:10.410 Daniel: But that’s what I wanted to ensure as part of this, right? Because I… as soon as we unleash this, it’s really going to be, like, define the top 5 things my company is working on now, right? How well are my verticals working together, right? There’s a ton of context that I think is going to need to be understood
340 00:53:10.500 ⇒ 00:53:13.869 Daniel: In, like, some type of…
341 00:53:14.350 ⇒ 00:53:17.260 Daniel: Knowledge backbone here that tells us…
342 00:53:17.260 ⇒ 00:53:17.740 Pranav Narahari: Exactly.
343 00:53:17.740 ⇒ 00:53:31.619 Daniel: hey, here are the five verticals. Here’s who’s in, you know, who works in those verticals. Let’s see how much they’re working together. Let’s see what the cross-functional projects are working on. Like, that’s the type of context. So I was talking to that company, Worklytics.
344 00:53:32.340 ⇒ 00:53:51.519 Daniel: And they basically use HRIS data to build out your teams, and then tell you who’s working with who across time zones or geo groups, and then tell you, like, based on those notes and messages, how engaged are teams, how unengaged are, and it’s really to evaluate, like, management of teams. And they wanted to charge, like, $3,600 a month.
345 00:53:51.650 ⇒ 00:54:06.859 Daniel: to basically give us a tool that would say, these teams are working well together, by the way, these teams are the most engaged, unengaged. And that’s really when I kicked this off, because I’m like, I have all that context. But what I’m realizing is the value in a lot of what they’ve created is using that HRIS data to…
346 00:54:07.110 ⇒ 00:54:09.749 Daniel: Form what your teams are, who they are.
347 00:54:09.750 ⇒ 00:54:10.100 Pranav Narahari: Yes.
348 00:54:10.100 ⇒ 00:54:24.869 Daniel: working together, you know, and then I wanted to go one step deeper and say, okay, I don’t care about just how much they’re talking with each other, I want to know what they’re working on together, too, because that shows me a lot of context there. So I think maybe it would be fruitful for us to spend some time building out that backbone
349 00:54:25.230 ⇒ 00:54:31.350 Daniel: Maybe we could do something like an HRIS export, or… Like a…
350 00:54:31.620 ⇒ 00:54:38.109 Daniel: maybe a, like an org… like, a literal org chart with all of our people on it to help identify some of these teams.
351 00:54:38.620 ⇒ 00:54:43.989 Pranav Narahari: that was gonna be something I brought up today, too. Like, yeah, that would be super helpful.
352 00:54:44.220 ⇒ 00:55:01.230 Pranav Narahari: Because it is kind of like a source of truth for the prompt to go after. So, if it knows, like, you know, maybe even certain people are associated with certain projects, then it specifically goes into the channels in which they are the most active. Instead of right now, it definitely crawls more…
353 00:55:01.330 ⇒ 00:55:15.859 Pranav Narahari: more wide scopes, and it’s going to eat up a lot of tokens that way. These, I mean, these reasoning models are very, very smart. However, they also eat up a lot of tokens, and what that’s… the problem with that sometimes is that it’s gonna…
354 00:55:15.940 ⇒ 00:55:33.869 Pranav Narahari: it’s gonna eat up the entire context window, and you’re gonna have issues with, it just not giving you an answer for some… for some prompts. Because there are gonna be certain projects, I’m sure, at Eden, where there is a ton of messages being sent, there’s a ton of activity in the drive, there’s a ton of emails.
355 00:55:34.040 ⇒ 00:55:53.029 Pranav Narahari: that’s all useful, right? And we want to be using all of that information for the themes, and to get a summary of how the project is currently… currently going. And so for that, having that org chart that you just mentioned, having that knowledge store is going to be… is going to help direct the AI a little bit more.
356 00:55:53.920 ⇒ 00:55:54.370 Daniel: Okay.
357 00:55:54.370 ⇒ 00:55:55.490 Pranav Narahari: So, yeah.
358 00:55:55.490 ⇒ 00:56:02.409 Daniel: I can start here. What I can do is at least work to produce a revised organization chart.
359 00:56:03.940 ⇒ 00:56:05.619 Daniel: I think that would be helpful.
360 00:56:06.310 ⇒ 00:56:06.830 Pranav Narahari: Sure.
361 00:56:06.830 ⇒ 00:56:08.360 Daniel: So, we can do that.
362 00:56:08.620 ⇒ 00:56:14.300 Daniel: By the way, I just tried to see if it was already built out. No, unfortunately, it does need to be revised. Ugh.
363 00:56:14.300 ⇒ 00:56:31.700 Pranav Narahari: And what I can probably do first, too, is let me kind of first come up with that schema. Like, what would actually be useful information for us? Like, I don’t want you to have to, like, put in a ton of information in that org chart that we may not need. Yeah, so I will… I will get that to you first, and then you can let me know, like.
364 00:56:31.790 ⇒ 00:56:39.300 Pranav Narahari: you know, how we can… if maybe all this information we can’t bring in, I’ll let you know, like, what is the most important, what would be nice to have as well.
365 00:56:39.680 ⇒ 00:56:44.989 Daniel: Okay. I definitely think an org chart would be super helpful, so it understands the teams. I also think that…
366 00:56:45.130 ⇒ 00:56:49.100 Daniel: Some of, like, the mission statement, goal, priorities…
367 00:56:49.100 ⇒ 00:56:49.510 Pranav Narahari: Yes.
368 00:56:49.510 ⇒ 00:56:57.419 Daniel: to figure out a way to express that, and I can probably spend some time, working on sort of what we do and what we care about.
369 00:56:57.420 ⇒ 00:57:01.639 Pranav Narahari: probably certain documents, like Q1 planning, Q2 planning, like, all that stuff.
370 00:57:01.670 ⇒ 00:57:18.800 Daniel: quarterly planning. I was also thinking, like, mission statement, like, as silly as those things are, the context models, as soon as I put that in the cloud, it’s like, oh, so this is DCC, so you care about order velocity and retention rate. This is the pharmacy, you care about cost of goods sold, you know, like, it knows right away
371 00:57:18.800 ⇒ 00:57:22.020 Daniel: Once it gets that context, what those businesses are.
372 00:57:22.630 ⇒ 00:57:39.419 Pranav Narahari: Yeah, and there… I don’t know if this is just gonna be in one folder, I doubt it, just, like, within Google Drive, but there’s probably certain executive memos that are sent out on, like, a quarterly basis, or just ad hoc, that are gonna give a lot of direction to just, like, what everybody else works on.
373 00:57:39.420 ⇒ 00:57:43.219 Pranav Narahari: Right? And so just having the context of, like, hey, this is what…
374 00:57:43.300 ⇒ 00:57:55.699 Pranav Narahari: is supposed to be worked on in the company. This is what’s set… this is what is, sent over from the highest level. Then everybody starts working on the sub-projects, the mini-projects, the filling in the gaps in between.
375 00:57:55.930 ⇒ 00:58:09.410 Pranav Narahari: I think that information will also be, like, super… super… super useful. And yeah, that’s exactly what people would use for a knowledge base for, basically as the backbone, as the source of truth of, like, okay, this is what people are going to be…
376 00:58:09.440 ⇒ 00:58:18.430 Pranav Narahari: this is… this is, like, direction for you to start off with, and then from there, using the MCP servers, using the GW CLI to fill in the gaps.
377 00:58:19.040 ⇒ 00:58:29.340 Daniel: Okay, love that as a next step takeaway on this. I also, quite frankly, I know it was, like, a separate project. I can actually be much more focused on this initiative.
378 00:58:29.510 ⇒ 00:58:50.360 Daniel: by not having to worry about how does everybody get access to it, because if that’s taken care of, then this basically just becomes an executive command center to look at some of these, you know, overall velocity functions, which means I don’t have to worry about credentialing as much. I, like, all that stuff’s kind of taken care of in other projects. That’s why I wanted to start with that, because it helped my mindset really narrow in on this goal and initiative here.
379 00:58:50.960 ⇒ 00:58:53.150 Pranav Narahari: Right, it simplifies things, yeah.
380 00:58:53.430 ⇒ 00:59:01.180 Daniel: and then I don’t have to worry about how am I going to get people access to this, what, you know, then I can just use this towards update, you know, me on, here are our queue…
381 00:59:01.390 ⇒ 00:59:10.139 Daniel: update me on our Q2 OKR initiatives, and it should literally be able to pull that context straight out of these tools and deliver me an async result, which is cool.
382 00:59:10.550 ⇒ 00:59:11.360 Daniel: Yep.
383 00:59:11.360 ⇒ 00:59:11.990 Pranav Narahari: Yup.
384 00:59:11.990 ⇒ 00:59:19.809 Daniel: Let’s get it some context, tell me if you help. I’m gonna go… I need to go redo our org chart, obviously, in our HRIS, because it’s very wrong. So, I’ll just…
385 00:59:20.170 ⇒ 00:59:21.570 Daniel: Just out of that, which is probably good.
386 00:59:21.570 ⇒ 00:59:22.100 Pranav Narahari: Okay.
387 00:59:22.250 ⇒ 00:59:38.879 Daniel: food for thought anyway. In addition to that, I’ll continue to throw this some search prompts and see what’s picking up in terms of project velocity. I may start to prompt it with some cross-team initiatives and see how that changes anything, but it’s a pretty good setup so far. I like having this underpinning MCP backbone.
388 00:59:39.610 ⇒ 00:59:41.740 Pranav Narahari: Yes, yeah.
389 00:59:41.870 ⇒ 00:59:53.789 Pranav Narahari: Also, when you get a chance, take a look at, or if you have a couple more minutes, I can just show you what this, like, new UI looks like, too. Or if you don’t think we can just… yeah, let’s just do that real quick.
390 00:59:54.330 ⇒ 00:59:59.520 Adam P: I’m gonna jump for the guy who’ll get a kid from the bus, so, like, connect with me later, but I don’t need anything else from the day.
391 00:59:59.520 ⇒ 01:00:02.010 Pranav Narahari: Definitely. Definitely, definitely. Talk to you later.
392 01:00:02.010 ⇒ 01:00:02.790 Adam P: 100, guys.
393 01:00:05.400 ⇒ 01:00:09.339 Pranav Narahari: Yeah, so, much simpler, less busy interface here.
394 01:00:09.340 ⇒ 01:00:12.009 Daniel: This was the, it’s in Lebanon.
395 01:00:12.010 ⇒ 01:00:19.509 Pranav Narahari: It’s the lovable. Yes, it’s the lovable. It’s much simpler than… but I think it encompasses everything that we need here.
396 01:00:19.720 ⇒ 01:00:34.129 Pranav Narahari: there’s a few things from our conversation from just, like, ideas that I had before that is still not here, but that prompt library thing is gonna be here, too. I think what we can also have here is certain toggles, like.
397 01:00:34.270 ⇒ 01:00:46.940 Pranav Narahari: On… on the actual message text area, which is how to enable org chart, enable, like, maybe… maybe those… maybe those backbones can just be enabled
398 01:00:48.610 ⇒ 01:00:57.530 Pranav Narahari: We need to do some testing here. Based on our conversation, I think that should just always be on, right? The org chart. It’s always going to be the right context.
399 01:00:57.530 ⇒ 01:01:09.889 Daniel: we know it’s gonna be used just by me and a couple other people for super high level, like, you know, the weekly summary one makes sense, but, like, meeting prep? If I’ve got that accomplished in another tool that knows me and what I’m working on, that’s not a big deal.
400 01:01:10.180 ⇒ 01:01:10.740 Pranav Narahari: Yeah.
401 01:01:10.740 ⇒ 01:01:14.390 Daniel: Very much more, like, overall company objectives.
402 01:01:14.840 ⇒ 01:01:26.940 Pranav Narahari: Totally, yeah. These threads here don’t make as much sense anymore, because we don’t care anymore about the use… the individual user. We’re talking more about, like, the org-scoped, data. So…
403 01:01:27.380 ⇒ 01:01:35.600 Pranav Narahari: I mean, it narrows our focus here, which is great, right? We can really just, like, focus on and kill that part of, part of what we talked about, so…
404 01:01:35.750 ⇒ 01:01:51.059 Pranav Narahari: Okay, that’s good that you saw this. There’s some updates that I’ll make to this, and then I’ll send it again into the Slack. We definitely should meet again before Friday. I think maybe on Wednesday I can, like, set up a meeting, and then we can kind of…
405 01:01:51.170 ⇒ 01:01:56.720 Pranav Narahari: by that time, I will have a schema for what I think would be a good
406 01:01:57.530 ⇒ 01:02:06.770 Pranav Narahari: good knowledge, or, a good data format for our knowledge base. Org chart’s gonna be good. I’m gonna also come up with another one in terms of,
407 01:02:07.420 ⇒ 01:02:08.809 Pranav Narahari: maybe just…
408 01:02:09.250 ⇒ 01:02:19.870 Pranav Narahari: on a person-to-person basis, like, what metadata should we have per person? Could be part of the org chart, maybe we just… it’s a separate, document.
409 01:02:19.990 ⇒ 01:02:30.370 Pranav Narahari: I will look into that a little bit for you, and then maybe I’ll see if I can get some time on your schedule on Wednesday, or we just connect about it async, but I’ll put time in on Wednesday.
410 01:02:30.610 ⇒ 01:02:38.029 Daniel: Yeah, Wednesday I’m good after 11… give me till 11.30, Mountain Time.
411 01:02:38.650 ⇒ 01:02:40.399 Pranav Narahari: Perfect. Yeah, we can do that.
412 01:02:40.870 ⇒ 01:02:56.449 Daniel: Okay, cool. Yeah, this is exciting stuff. At first, I was trying to, like, wreck my brain around what does this all mean, until I realized, oh, they’re actually two separate things. And then when we do that, all of a sudden, it’s very clear for me on what we need to go. We need to enable the team with their tools for augmenting them.
413 01:02:56.700 ⇒ 01:03:01.410 Daniel: But on our side, this is still a holistic effort to understand the organization workflows as a whole.
414 01:03:02.410 ⇒ 01:03:09.550 Pranav Narahari: Yeah, yeah, it’s a lot more clear to me, too, now, and I think we’re not doing two different things in parallel. Yeah.
415 01:03:09.860 ⇒ 01:03:10.710 Pranav Narahari: Okay.
416 01:03:10.880 ⇒ 01:03:11.779 Pranav Narahari: Awesome. Sounds good to me.
417 01:03:11.780 ⇒ 01:03:12.579 Daniel: Great, thanks, Pranav.
418 01:03:13.110 ⇒ 01:03:18.480 Pranav Narahari: Cool, yeah, we need to extend these meetings. Yeah, yeah, cool. Appreciate you.
419 01:03:18.590 ⇒ 01:03:20.459 Pranav Narahari: Talk to you later. See ya.