Meeting Title: Client Hub and Stand-up Improvements Date: 2025-11-19 Meeting participants: Samuel Roberts, Mustafa Raja, Casie Aviles, Gabriel Lam


WEBVTT

1 00:00:30.930 00:00:32.579 Samuel Roberts: I’m muted. Okay, there we go.

2 00:00:36.310 00:00:43.170 Samuel Roberts: I mean… Now that I’m thinking about this, Stand-up thing.

3 00:00:43.540 00:00:45.449 Samuel Roberts: Or the summary thing.

4 00:00:49.090 00:00:50.090 Samuel Roberts: Yeah.

5 00:00:53.610 00:00:56.930 Gabriel Lam: I think the flow for now seems to be, like.

6 00:00:57.270 00:01:00.020 Gabriel Lam: The… from the meeting, we have the summary.

7 00:01:00.750 00:01:06.109 Gabriel Lam: Ideally splits into work streams, so there’s, like, You know, it depends.

8 00:01:06.580 00:01:07.980 Samuel Roberts: But I don’t think right now it does that right.

9 00:01:07.980 00:01:15.280 Gabriel Lam: It doesn’t do that. Yeah, and then, ideally, from the work streams, and it has… if it’s a client meeting, it’s like, okay, this is what the client needs.

10 00:01:16.080 00:01:19.599 Gabriel Lam: To do on that side, this is what we need to do on that side, and then…

11 00:01:19.770 00:01:29.980 Gabriel Lam: you know, it might be, like, data collection, this is what we need from the client, this is what we can do, and then it might be, like, the warehousing, and this is what the client can do, this is what we can do. And then ac…

12 00:01:30.350 00:01:37.630 Gabriel Lam: that… I think the summary would in some way include an action item. I don’t know if a summary really is helpful, just to be like, this is what the meeting is about, without…

13 00:01:37.630 00:01:38.340 Samuel Roberts: Right.

14 00:01:38.340 00:01:39.440 Gabriel Lam: That sort of thing.

15 00:01:39.690 00:01:43.240 Gabriel Lam: I think linear tickets, as a result of that.

16 00:01:44.150 00:01:46.850 Gabriel Lam: I think linear tickets can either come out

17 00:01:47.190 00:01:50.320 Gabriel Lam: Separately, or as a result of these action items.

18 00:01:50.840 00:02:00.300 Gabriel Lam: and I think that can live within each meeting like, object.

19 00:02:01.420 00:02:13.790 Samuel Roberts: Right, so when a meeting is ingested into the platform, like, summary action items in some form, whether that’s together or separate, get generated, and a bunch of potential linear tickets get generated?

20 00:02:15.250 00:02:16.850 Samuel Roberts: And then…

21 00:02:17.670 00:02:23.440 Samuel Roberts: Okay, I can see that. Like, based on the actions that we discussed in the meeting, here’s…

22 00:02:23.920 00:02:25.760 Samuel Roberts: Learning tickets to create.

23 00:02:26.200 00:02:32.800 Samuel Roberts: Then at some point, those could get created automatically, I suppose, or there could be a human in the loop to be like, yes, yes, yes, no.

24 00:02:32.800 00:02:40.049 Gabriel Lam: Yeah, I think… I think right now it is a human in the loop, and I still believe that a human should be in the loop. I don’t really trust the AI to.

25 00:02:40.050 00:02:40.500 Samuel Roberts: No, I agree.

26 00:02:40.500 00:02:49.520 Gabriel Lam: adequate tickets. I think the follow-up to that is, like, okay, if we have a ton of meetings and a ton of summaries and a ton of action items, like, what’s the most up-to-date?

27 00:02:49.670 00:02:56.559 Gabriel Lam: And then… and where does that live? Does it live in the client hub? Which maybe it will eventually, so you don’t have to…

28 00:02:57.100 00:03:02.980 Gabriel Lam: Or does it live in the stand-up position? Like, what is… It seems like… Most people…

29 00:03:03.430 00:03:07.929 Gabriel Lam: go through linear as, like, okay, that’s my source of truth, or, like, Slack is my source of truth.

30 00:03:08.090 00:03:18.620 Gabriel Lam: To be like, if I want to know what’s the most up-to-date thing, I go to Slack, I go to Linear, and I know what to do. Do we play along with that and say, like, okay, at the end of every meeting.

31 00:03:18.870 00:03:22.569 Gabriel Lam: like, you get a Slack notification for the summary, so you don’t have to…

32 00:03:23.070 00:03:29.260 Gabriel Lam: you don’t have to, like, look through anything, or is it we have to get people to start using the Client Hub, or start using…

33 00:03:29.620 00:03:31.100 Gabriel Lam: The case study assistant.

34 00:03:31.650 00:03:32.140 Gabriel Lam: And.

35 00:03:32.140 00:03:32.540 Samuel Roberts: Yeah.

36 00:03:32.540 00:03:35.580 Gabriel Lam: Okay, change your behavior there.

37 00:03:36.660 00:03:38.379 Samuel Roberts: Two thoughts on that. I think…

38 00:03:40.130 00:03:46.600 Samuel Roberts: My initial gut is to go where people are, because… Changing behavior is difficult, and…

39 00:03:46.600 00:03:47.140 Gabriel Lam: Right.

40 00:03:48.010 00:03:56.060 Samuel Roberts: But, I will say, I think part of the reason people might not be using the client hubs is the… maybe we need to rest…

41 00:03:56.160 00:04:08.469 Samuel Roberts: I don’t know how to say restructure, but, like, the client hubs, and guys, correct me if I’m wrong about, like, the historical way this was set up, was, like, ingesting the meetings, organizing the clients, you know.

42 00:04:08.940 00:04:18.699 Samuel Roberts: hitting all those, transcripts and stuff when questions are asked, so we have, like, a way to access that information, but I don’t know if it’s ever really synthesized anywhere.

43 00:04:19.970 00:04:25.679 Samuel Roberts: In that… there’s no, like, I click on this client on the…

44 00:04:25.970 00:04:30.159 Samuel Roberts: platform, and I can see what the current state of that client is.

45 00:04:32.040 00:04:39.609 Gabriel Lam: There’s an assistant that I see that I can ask and be like, and that’s gonna synthesize every time, it’s gonna hit that…

46 00:04:39.610 00:04:45.079 Samuel Roberts: right now, N8N, agent, and go through and do that, so I’m now wondering…

47 00:04:45.620 00:04:52.860 Samuel Roberts: And maybe I’m… maybe I’m overstretching here a little bit, but if we’re gonna start, especially now that we have, like, the stand-up assistant.

48 00:04:53.170 00:05:00.010 Samuel Roberts: And we’re kind of… We want more… Overtime tracking of stuff.

49 00:05:01.830 00:05:06.380 Samuel Roberts: Excuse me. Might we need to rethink, like, what gets stored on a client at any given time?

50 00:05:11.260 00:05:20.489 Samuel Roberts: And then, again, I’m kind of still spitballing a little bit, where, like, what does the ideal client hub look like, and what is it now, is kind of my current thinking.

51 00:05:21.270 00:05:23.819 Samuel Roberts: Because… if we could have…

52 00:05:26.060 00:05:42.370 Samuel Roberts: action… well, I don’t know, maybe I’m going too deep into this, but… If we were to build this from scratch, what would it look like? It’s kind of my thought. Like, the things have been kind of… we have Zoom meetings brought in, we have Slack messages brought in, that all… the data’s there, but I don’t think there’s ever really, like, a…

53 00:05:43.130 00:05:49.169 Samuel Roberts: You know, now with the stand-up view, or the stand-up assistant, we have a little bit of that, but maybe we need to lean into that more.

54 00:05:51.230 00:05:57.950 Samuel Roberts: Such as action items after the meeting that feeds into something that is, like, a current status of the client.

55 00:06:01.640 00:06:02.270 Casie Aviles: Good luck.

56 00:06:02.390 00:06:05.690 Casie Aviles: Are you creating client objects, something like that.

57 00:06:05.690 00:06:12.369 Samuel Roberts: Or, no. Potentially, yeah, but I don’t even want to… I’m just thinking, like, when we think of, like, right now, the client hub.

58 00:06:12.740 00:06:16.640 Samuel Roberts: Is just a list of meetings with a chat next to it, right?

59 00:06:17.260 00:06:17.600 Casie Aviles: Yeah.

60 00:06:17.600 00:06:18.100 Mustafa Raja: Yeah.

61 00:06:18.100 00:06:21.189 Casie Aviles: and also some SLAP data, so it’s really just…

62 00:06:21.190 00:06:25.280 Samuel Roberts: But the chat goes to the Slack data, it goes to the transcripts, it pulls all that, right?

63 00:06:25.280 00:06:25.600 Casie Aviles: Yeah.

64 00:06:25.600 00:06:31.729 Samuel Roberts: I’m now thinking, like, what is… what else would I want to see if I click on a client… the client…

65 00:06:32.670 00:06:33.610 Samuel Roberts: link.

66 00:06:33.780 00:06:35.930 Samuel Roberts: on the forge, and what do I see?

67 00:06:38.430 00:06:43.299 Gabriel Lam: Which I think is what, sort of, Utam is saying, which is, like, if I click on the client, then…

68 00:06:44.090 00:06:46.300 Gabriel Lam: I would want to know what…

69 00:06:46.910 00:06:49.710 Gabriel Lam: The client’s asking for at the moment, or what’s most up-to-date.

70 00:06:49.710 00:07:05.679 Samuel Roberts: Right, like, what’s currently in progress, what is… what are the current outstanding at… like, the stuff that is kind of brought into the stand-up assistant in some way is probably better to, like, live on a client object of some sort, or, like, current status, or something like that.

71 00:07:06.270 00:07:11.920 Samuel Roberts: That is not necessarily something we can just jump to, but the current…

72 00:07:12.120 00:07:24.240 Samuel Roberts: setup, but maybe the first steps to get to something like this, if that’s, like, the end goal, is, like, you click on the client page, you see the meetings, you see the chat, but you also have, like, a summary of the current state of the client at the top.

73 00:07:24.940 00:07:42.249 Samuel Roberts: whether that’s outstanding action items, or, you know, what’s blocked from the client side, or whatever it is. Like, we can think about that a little bit more. But maybe the first few steps to that is every meeting gets action items and blockers and things kind of pulled out of it.

74 00:07:44.870 00:07:49.229 Samuel Roberts: That can get synthesized into something that’s a current state.

75 00:07:50.200 00:07:51.690 Samuel Roberts: At any given time.

76 00:07:53.930 00:07:55.889 Gabriel Lam: Right, so it’s like, it’s almost like…

77 00:07:55.960 00:07:57.820 Samuel Roberts: At the end of every meeting, we have.

78 00:07:58.140 00:08:03.040 Gabriel Lam: In each meeting object, we have a summary, we have action items, we have tickets as a result.

79 00:08:03.280 00:08:07.990 Gabriel Lam: And we’ll just let that sort of be the first, like, the zero.

80 00:08:08.950 00:08:13.579 Samuel Roberts: Yeah, I think that’s, like, what was the last thing discussed? Yeah, potentially. I mean, there’s definitely, like…

81 00:08:14.970 00:08:20.409 Samuel Roberts: Edge cases there, where, like, there might be multiple meetings that are applied to the client that, like.

82 00:08:20.550 00:08:23.050 Samuel Roberts: Don’t necessarily supersede each other.

83 00:08:24.190 00:08:31.629 Gabriel Lam: And I think that’s where the human can sort of step in and be like, okay, this is… this is an irrelevant ticket to add, like, we can just cross it off, which is sort of what happens now.

84 00:08:31.630 00:08:44.479 Samuel Roberts: Yeah, I think that’s true, I think that’s true. I’m thinking even just, like, if there’s a… you know, the last meeting was the stand-up, and that’s got a bunch of client stuff, and then there was another meeting where two people got on to discuss some detail about that.

85 00:08:45.060 00:08:50.900 Samuel Roberts: you know, that might move certain things along, but not everything. But then we can… we can sort that out. As long as it’s all there.

86 00:08:51.980 00:08:56.880 Samuel Roberts: Like, you could even see, like, okay, the last meeting discussed this, there’s action items, this got brought forward.

87 00:08:58.960 00:09:03.569 Samuel Roberts: That might be, yeah, the V0 is just, like, this was the last meeting, this was the last updates.

88 00:09:05.460 00:09:09.299 Gabriel Lam: Yeah, and I wonder if that same… I think my hope is that

89 00:09:10.830 00:09:17.319 Gabriel Lam: the synthesized summary gets put somewhere. And maybe it’s a stand-up assistant for now, where it’s like…

90 00:09:18.280 00:09:21.020 Gabriel Lam: You know, tomorrow morning, when we have a call, we’re like, okay.

91 00:09:21.460 00:09:25.090 Gabriel Lam: This is a summary of all the calls, and… Yeah. Action.

92 00:09:25.090 00:09:26.150 Samuel Roberts: 100%. I think.

93 00:09:26.360 00:09:27.120 Gabriel Lam: And I, yeah.

94 00:09:27.120 00:09:32.959 Samuel Roberts: One thing that made the stand-up assistant harder to do was the fact that we didn’t have all this kind of, like… we had to resynthesize a bunch of stuff.

95 00:09:33.090 00:09:40.419 Gabriel Lam: Yeah, and I think we’ll see the ticket moves, and I think it’s fine to not have tickets be done in the stand-up assistant, because…

96 00:09:41.950 00:09:44.659 Gabriel Lam: that… Like, once we have them eating.

97 00:09:45.130 00:09:49.710 Gabriel Lam: We can go into the client hub, or we can go into the dashboard and…

98 00:09:50.360 00:09:52.230 Gabriel Lam: approve. And maybe it’s, like.

99 00:09:52.350 00:09:57.820 Gabriel Lam: you know, we get a Slack notification at the end, and it’s like, hey, the meeting’s up, can someone approve it?

100 00:09:58.590 00:09:59.760 Samuel Roberts: Right.

101 00:10:02.600 00:10:06.089 Gabriel Lam: just some thoughts, but I think for now… Yeah.

102 00:10:06.090 00:10:12.660 Samuel Roberts: Yeah, I’m definitely thinking, like, long… yeah, but I think, yeah, the first thing is probably… update that summary.

103 00:10:12.660 00:10:16.790 Mustafa Raja: Yeah… Yeah, I’m just, wondering one thing.

104 00:10:17.000 00:10:17.550 Samuel Roberts: Yeah.

105 00:10:17.720 00:10:29.050 Mustafa Raja: So right now, the way this captures the meetings is by the title, and we haven’t been… I believe we haven’t been titling the stand-ups pretty good.

106 00:10:29.330 00:10:31.789 Mustafa Raja: Based on, so, so…

107 00:10:32.020 00:10:46.379 Mustafa Raja: we talk about Insomnia every day, but none of the stand-ups would be titled as Insomnia stand-up or something like that, right? So, for the current implementation, we can say that Insomnia,

108 00:10:46.950 00:10:50.349 Mustafa Raja: Client hasn’t been receiving any meetings at all.

109 00:10:51.750 00:10:55.410 Samuel Roberts: Yeah, that’s… that’s a… an issue.

110 00:10:55.760 00:10:59.660 Samuel Roberts: in a few different ways. I think the title is definitely…

111 00:10:59.850 00:11:02.369 Samuel Roberts: Or the basing it on the title is one thing.

112 00:11:03.510 00:11:12.290 Samuel Roberts: I think the fact that multiple clients are discussed in one big meeting is another thing we have to work out better than we have been in general.

113 00:11:17.420 00:11:20.229 Samuel Roberts: I don’t have a good solution for that yet, besides, like.

114 00:11:21.550 00:11:24.890 Samuel Roberts: Maybe we need to change how we’re classifying

115 00:11:28.170 00:11:29.220 Samuel Roberts: Well, yeah.

116 00:11:29.220 00:11:37.460 Mustafa Raja: I mean, after the meeting, I believe we can, we can sort of, in the database, have a field

117 00:11:37.580 00:11:56.639 Mustafa Raja: The clients talked about in this meeting, and then, have an AI agent go through the meeting and populate from the client’s table the names of… proper names or proper IDs of the clients, and then we’ll have to also change the way we are ingesting all of the data in, in vector stores.

118 00:11:59.010 00:12:03.580 Samuel Roberts: Yeah, potentially, I think that’s… That’s definitely…

119 00:12:04.190 00:12:06.230 Samuel Roberts: That’s the other side of this, is what…

120 00:12:06.340 00:12:12.700 Samuel Roberts: ends up in the client hub? Is, like, is every client hub going to have the information from all these stand-ups?

121 00:12:17.540 00:12:18.060 Gabriel Lam: Hmm.

122 00:12:18.060 00:12:23.749 Samuel Roberts: Like, is that meeting, like, is that stand-up meeting in the morning gonna end up on every client hub page?

123 00:12:27.260 00:12:32.080 Gabriel Lam: Perhaps, I mean… It’s the internal, facing…

124 00:12:32.790 00:12:35.779 Gabriel Lam: like, interactions, I don’t think it really matters.

125 00:12:36.520 00:12:40.039 Samuel Roberts: I’m just thinking how that affects, like, what ends up in the summaries and stuff.

126 00:12:41.910 00:12:50.159 Samuel Roberts: I think we need… yeah, okay. We might need to rethink a little bit how we’re, yeah, fully ingesting these meetings completely, and classifying and organizing them.

127 00:12:52.270 00:12:56.599 Samuel Roberts: To be a little more robust to handle these, like, multi-client meetings.

128 00:12:58.620 00:12:59.520 Samuel Roberts: In a way that…

129 00:12:59.520 00:13:00.070 Gabriel Lam: Done.

130 00:13:01.190 00:13:05.300 Samuel Roberts: You know, because if we have… if we discuss, you know, however many clients on the stand-up.

131 00:13:05.560 00:13:08.610 Samuel Roberts: The action items need to be tied to those clients.

132 00:13:09.910 00:13:14.969 Gabriel Lam: which I think at least they are editable for now, so when… I think the review stage…

133 00:13:15.450 00:13:18.109 Gabriel Lam: Covers a good chunk of that, which is…

134 00:13:18.670 00:13:26.220 Samuel Roberts: Yes, yes. No, I see what you’re saying, yeah. I’m just thinking, like, if I… Yeah…

135 00:13:32.090 00:13:41.369 Samuel Roberts: Yeah, okay, you’re right. So if we have it pull out, and it thinks this client needs this action item, etc, etc, someone can go through and make sure that that’s right.

136 00:13:42.030 00:13:45.010 Gabriel Lam: Yes. As they would do to create linear tickets anyway.

137 00:13:47.550 00:13:48.050 Samuel Roberts: Okay.

138 00:13:48.050 00:13:54.080 Gabriel Lam: So… Maybe to go back to, like, what needs to be in the tickets?

139 00:13:54.250 00:14:00.550 Gabriel Lam: And, like, what we need to do, ideally, for the end of the week, we want to fix the streaming, or, like, add…

140 00:14:00.800 00:14:04.899 Gabriel Lam: A streaming layer to… Prevent the timeout?

141 00:14:06.890 00:14:11.359 Gabriel Lam: From the ticket generation, we need to refine the prompts, so the…

142 00:14:12.210 00:14:20.679 Gabriel Lam: So the tickets are actually helpful, and the descriptions… the titles and the descriptions of the prompt are more helpful.

143 00:14:21.500 00:14:26.780 Gabriel Lam: We need to do a summary.

144 00:14:27.340 00:14:28.900 Gabriel Lam: slash action item.

145 00:14:30.420 00:14:30.920 Samuel Roberts: Yeah.

146 00:14:30.920 00:14:35.490 Gabriel Lam: To match the exit, like, whatever standards we’re hoping to get with…

147 00:14:36.180 00:14:40.340 Gabriel Lam: For that to show up for each meeting. And then also…

148 00:14:44.340 00:14:50.109 Gabriel Lam: Is there anything else on top? Like, if we were to click… If I was to click

149 00:14:50.970 00:14:58.309 Gabriel Lam: create linear tickets, and I’m reading through the descriptions, and I’m like, would it be helpful to have a timestamp of where it’s from, or like.

150 00:14:58.730 00:15:03.050 Gabriel Lam: For it to show the context, probably… .

151 00:15:03.050 00:15:06.920 Samuel Roberts: If there’s… if there’s questions and someone wasn’t in the meeting and doesn’t know, yeah, I think…

152 00:15:07.200 00:15:11.780 Samuel Roberts: Sourcing that information is… Definitely a benefit.

153 00:15:12.770 00:15:13.190 Gabriel Lam: Yeah.

154 00:15:13.190 00:15:13.659 Samuel Roberts: We’re showing.

155 00:15:13.660 00:15:18.650 Gabriel Lam: I mean, Utam just, tagged me in this…

156 00:15:19.790 00:15:26.569 Gabriel Lam: Where Henry is talking to the meeting assistant, and he’s like, hey, can you find a timestamp where people spoke about this?

157 00:15:27.030 00:15:32.610 Gabriel Lam: I don’t know how much of that we want to put in, I think it’s more just…

158 00:15:35.340 00:15:42.739 Gabriel Lam: Is it, like, part of the… Highlight part of the… transcript that says.

159 00:15:44.940 00:15:48.810 Samuel Roberts: Totally. Yeah, I mean, yeah, I would say it would be, like, pulling the…

160 00:15:50.020 00:15:53.310 Samuel Roberts: Either pulling the text out or linking to that in the transcript file.

161 00:15:59.140 00:16:08.509 Samuel Roberts: I don’t know exactly, I’m not… maybe I’m having a hard time picturing exactly, like, you’re thinking, like, on the ticket, in the description, having some kind of, like, quote or something?

162 00:16:14.990 00:16:17.360 Gabriel Lam: Perhaps. I’m not sure. Yeah.

163 00:16:17.360 00:16:19.190 Samuel Roberts: Yeah. We… I mean, we could…

164 00:16:19.330 00:16:22.790 Samuel Roberts: play with that. I think it’s either that, or it’s…

165 00:16:23.930 00:16:31.550 Mustafa Raja: linking to the timestamp, you can look into the transcript and see. Yeah, I was thinking the same. We have the ability to do that now, so…

166 00:16:31.550 00:16:32.020 Samuel Roberts: Yeah.

167 00:16:32.020 00:16:32.690 Gabriel Lam: Okay.

168 00:16:34.010 00:16:37.050 Gabriel Lam: Okay.

169 00:16:37.960 00:16:42.619 Samuel Roberts: Yeah, I think… I think just… yeah, I think what you have there is right, though, for the rest of the…

170 00:16:43.330 00:16:46.059 Samuel Roberts: Rest of the plan for the week.

171 00:16:48.830 00:16:49.790 Gabriel Lam: Yeah. I’m…

172 00:16:49.790 00:16:55.960 Samuel Roberts: I’m also thinking maybe we need to start really diagramming out a… Like, what…

173 00:16:56.280 00:17:01.100 Samuel Roberts: What would be an ideal flow for the new ingestion, if we’re gonna have to do that, to make this work better?

174 00:17:02.720 00:17:06.040 Samuel Roberts: But that’s not necessarily… This week.

175 00:17:06.440 00:17:09.310 Samuel Roberts: But… Thought might need to go into that.

176 00:17:09.490 00:17:10.180 Gabriel Lam: Yeah.

177 00:17:11.280 00:17:19.499 Gabriel Lam: Are there things that we can distribute? I’m happy to take up the… now that I’m in LangFuse, I can take up a lot more of the prompt refinement,

178 00:17:19.790 00:17:22.220 Gabriel Lam: And I can start on that flow. I think it’s more just…

179 00:17:22.530 00:17:24.630 Gabriel Lam: Maybe a question for you guys is, like, what…

180 00:17:27.319 00:17:36.800 Gabriel Lam: platform, or, like, is FigJam the best place for us to… Like, look over that as…

181 00:17:36.960 00:17:38.509 Gabriel Lam: I guess what works for you guys.

182 00:17:38.700 00:17:39.450 Gabriel Lam: To see a flow.

183 00:17:39.450 00:17:42.610 Samuel Roberts: like, Yeah, probably,

184 00:17:45.700 00:17:56.219 Samuel Roberts: Yeah, I think that probably is the best place. I was actually dealing with this yesterday, where I, like… the nice benefit of being able to, like, chat with ChatGPT or cursor is I can output mermaid diagrams.

185 00:17:57.030 00:18:00.079 Samuel Roberts: Which is, like, code-based diagram stuff.

186 00:18:00.560 00:18:06.639 Samuel Roberts: It doesn’t really play well with FigJam, though, so it’s not… doesn’t go… Both ways into it, but…

187 00:18:07.400 00:18:10.639 Samuel Roberts: Maybe we want to set something up where we, yeah, we diagram the current

188 00:18:11.950 00:18:16.639 Samuel Roberts: Current way it is, and figure out where we need to add some more steps to maybe…

189 00:18:17.740 00:18:20.149 Samuel Roberts: You know, classify the meetings better.

190 00:18:20.420 00:18:22.370 Samuel Roberts: You know, figure out the titles.

191 00:18:22.820 00:18:26.589 Samuel Roberts: Figure out how we’re gonna add the…

192 00:18:29.180 00:18:32.570 Samuel Roberts: Well, the action items and summary is kind of already there, but…

193 00:18:34.030 00:18:39.329 Samuel Roberts: And maybe this is not… maybe… I think I’m going a little bit too far. I think what you have for the week makes sense.

194 00:18:39.330 00:18:39.790 Gabriel Lam: No problem.

195 00:18:41.750 00:18:45.439 Gabriel Lam: Would anyone like to take up the timeout streaming issue?

196 00:18:49.320 00:18:54.240 Samuel Roberts: I’m… I would, I’m out at 1 o’clock, though.

197 00:18:54.240 00:18:55.090 Gabriel Lam: Okay.

198 00:18:55.620 00:18:57.949 Samuel Roberts: So I don’t know if I should, but I…

199 00:18:58.540 00:19:00.910 Gabriel Lam: Hi, how’s the rest of the day look for you guys?

200 00:19:00.910 00:19:02.100 Samuel Roberts: Casey Mustafa.

201 00:19:04.610 00:19:06.589 Mustafa Raja: Yeah, I have.

202 00:19:09.150 00:19:10.179 Casie Aviles: Yeah, I was…

203 00:19:10.180 00:19:15.249 Mustafa Raja: I don’t know, if I would be able to play today, but yeah. Okay, okay.

204 00:19:15.710 00:19:17.830 Mustafa Raja: By tomorrow, I definitely can.

205 00:19:18.550 00:19:25.870 Samuel Roberts: Okay, I can take a swing at it before 1, maybe, because I did it… for the… Stand-up assistant?

206 00:19:26.350 00:19:27.450 Samuel Roberts: So maybe I’ll…

207 00:19:27.450 00:19:28.220 Mustafa Raja: set it up.

208 00:19:29.100 00:19:29.520 Samuel Roberts: For the.

209 00:19:29.520 00:19:35.910 Mustafa Raja: Yeah, we just need… yeah, we just need to configure it, configure the webhook to, you know, stream.

210 00:19:37.150 00:19:39.120 Samuel Roberts: Oh, this is N-A-N, I’m sorry, you’re right, you’re right, okay.

211 00:19:39.120 00:19:40.439 Casie Aviles: Definitely not me, though.

212 00:19:40.440 00:19:46.860 Samuel Roberts: Sorry, I was thinking the way I did it in Monster for the stand-up agent. Yeah, okay, I’m definitely not the one to do that, then.

213 00:19:47.350 00:19:52.589 Casie Aviles: Yeah, I’ll… I can… I guess I can take a look. I’ll try to take a look at that today.

214 00:19:53.340 00:19:53.840 Mustafa Raja: Yeah.

215 00:19:53.840 00:19:58.009 Samuel Roberts: Because is that blocking being able to, like, work on the prompts for that, because it’s not returning?

216 00:19:58.520 00:19:59.130 Gabriel Lam: It’s not black.

217 00:19:59.130 00:20:01.699 Mustafa Raja: For shorter meetings, the prompt works.

218 00:20:02.120 00:20:02.750 Gabriel Lam: Yeah.

219 00:20:03.510 00:20:08.489 Samuel Roberts: Okay, so we can make some movement on that, and then get the rest of it streaming.

220 00:20:08.910 00:20:09.949 Samuel Roberts: Figure the rest.

221 00:20:10.270 00:20:10.760 Mustafa Raja: Yeah.

222 00:20:11.190 00:20:11.800 Samuel Roberts: Okay.

223 00:20:13.680 00:20:17.540 Samuel Roberts: Okay, what else was on the list?

224 00:20:18.000 00:20:21.719 Gabriel Lam: So it’s just the streaming, it’s the prompt refinement, the summary action…

225 00:20:22.120 00:20:29.749 Gabriel Lam: step. I think that would just be, again, on an 8 and… Putting an agent onto the…

226 00:20:29.890 00:20:31.340 Gabriel Lam: The summary part.

227 00:20:32.200 00:20:32.730 Gabriel Lam: Yeah.

228 00:20:32.730 00:20:35.209 Samuel Roberts: Instead of the summarizer that’s there, yeah.

229 00:20:35.210 00:20:37.320 Gabriel Lam: Yeah, I think it’s, like, a built-in summarizer.

230 00:20:37.320 00:20:42.480 Samuel Roberts: Yeah, so for now, we’ll just make the summary a… In…

231 00:20:42.610 00:20:46.679 Samuel Roberts: Custom prompt, and also specify list action items or something?

232 00:20:46.680 00:20:47.380 Gabriel Lam: Yeah.

233 00:20:47.850 00:20:55.350 Samuel Roberts: Okay, I think that’s fine for now. I think at some point we might want to split that into, like, separate things that get stored onto the… like, a summary and an action items…

234 00:20:56.180 00:21:00.340 Samuel Roberts: Next field kind of thing, but for now, let’s work with that. Let’s work with what’s there.

235 00:21:00.800 00:21:01.420 Gabriel Lam: Okay.

236 00:21:05.580 00:21:12.590 Gabriel Lam: Alright, I’ll edit the tickets, and then we’ll see what we can get up by the end of the year.

237 00:21:12.590 00:21:19.829 Mustafa Raja: want to… Yeah. Something… Sorry, let me just get the…

238 00:21:21.650 00:21:24.730 Mustafa Raja: Workflow up, and then I’ll share my screen.

239 00:21:30.760 00:21:31.590 Mustafa Raja: Okay.

240 00:21:31.840 00:21:35.389 Mustafa Raja: Okay, we only talked about this agent, right?

241 00:21:35.390 00:21:36.340 Samuel Roberts: One sec.

242 00:21:36.340 00:21:37.560 Mustafa Raja: the ticket generally.

243 00:21:37.810 00:21:38.380 Samuel Roberts: Yep.

244 00:21:39.780 00:21:42.169 Mustafa Raja: Yeah, so we only talked about this agent, right?

245 00:21:42.190 00:22:01.640 Mustafa Raja: That is linked with this, ticket generator prompt. So if, if any of these tickets… if any of the tickets generated by this agent belongs to any of these clients, it’ll, that particular ticket will go up to this agent for more context.

246 00:22:01.880 00:22:07.549 Mustafa Raja: So, what we need to do is also update the ticket groomer prompt.

247 00:22:07.820 00:22:15.200 Mustafa Raja: Which should be… this one, generated tickets groomer. So this should also then, know the…

248 00:22:15.500 00:22:17.950 Mustafa Raja: Format that we need.

249 00:22:18.150 00:22:19.249 Mustafa Raja: Yeah, that’s all.

250 00:22:19.640 00:22:20.070 Gabriel Lam: Okay.

251 00:22:20.070 00:22:21.489 Mustafa Raja: I just wanted to add that to…

252 00:22:21.490 00:22:22.670 Samuel Roberts: To that prompt.

253 00:22:23.060 00:22:23.700 Gabriel Lam: Yeah.

254 00:22:23.700 00:22:26.030 Samuel Roberts: Yeah, yeah. Okay. That’s all, thank you.

255 00:22:26.850 00:22:27.440 Gabriel Lam: No problem.

256 00:22:27.440 00:22:29.439 Samuel Roberts: That makes sense. Good, good catch there.

257 00:22:29.440 00:22:36.410 Gabriel Lam: Okay, I just have two things to end on. Sure. The first thing is… Okay.

258 00:22:36.670 00:22:40.029 Gabriel Lam: I… I’m able to… So, for my…

259 00:22:40.780 00:22:43.100 Gabriel Lam: My first question is N8N, and just…

260 00:22:43.880 00:22:48.990 Gabriel Lam: am I going to… should… I don’t have an account, actually. I thought I did.

261 00:22:48.990 00:22:49.780 Samuel Roberts: Oh, okay.

262 00:22:49.780 00:22:52.649 Gabriel Lam: would I have to make one and start the free trial, and then get…

263 00:22:53.520 00:22:55.119 Samuel Roberts: I think you probably just getting invited, is that?

264 00:22:55.120 00:22:56.000 Gabriel Lam: get invited.

265 00:22:56.000 00:22:58.899 Mustafa Raja: Yeah, let me see if I can invite…

266 00:22:59.040 00:23:02.830 Mustafa Raja: Or if… or if I’ll have to ask Rico for that. Yeah, I can invite.

267 00:23:02.960 00:23:04.580 Mustafa Raja: Okay. Meanwhile, too.

268 00:23:04.870 00:23:10.759 Gabriel Lam: Awesome, thank you. The second question is, we talked about moving things onto Mastra.

269 00:23:12.230 00:23:16.330 Gabriel Lam: At least that first part, like… the… the first…

270 00:23:16.570 00:23:22.050 Gabriel Lam: Half before it gets split into client hubs, or before it gets…

271 00:23:22.920 00:23:25.019 Gabriel Lam: Yeah, I guess split into client hubs.

272 00:23:26.190 00:23:32.319 Gabriel Lam: Is that… it sounded like something that we can’t take on for this week, and maybe we should just…

273 00:23:32.500 00:23:35.800 Gabriel Lam: leave for… The full migration?

274 00:23:36.240 00:23:37.010 Samuel Roberts: I think so.

275 00:23:37.170 00:23:38.020 Casie Aviles: Yeah. Okay.

276 00:23:38.490 00:23:39.920 Casie Aviles: The whole sprint.

277 00:23:40.740 00:23:42.799 Samuel Roberts: Yeah, won’t take a week to do that.

278 00:23:43.580 00:23:44.510 Gabriel Lam: Yeah. Yeah.

279 00:23:47.030 00:23:48.530 Gabriel Lam: Awesome. Okay.

280 00:23:48.530 00:23:49.070 Samuel Roberts: Okay.

281 00:23:51.560 00:23:54.120 Gabriel Lam: I think… I think we’ve got it.

282 00:23:54.120 00:23:54.530 Samuel Roberts: Okay.

283 00:23:54.530 00:23:54.870 Gabriel Lam: Thank you.

284 00:23:54.870 00:23:55.949 Samuel Roberts: Yeah,

285 00:23:56.400 00:24:02.039 Samuel Roberts: like I said, I’m out at 1, and I have that, like, this is the last of those classes that I was doing.

286 00:24:02.450 00:24:05.370 Samuel Roberts: But ping me if you need something, I’ll have Slack, so…

287 00:24:06.130 00:24:08.109 Samuel Roberts: I just might not respond immediately.

288 00:24:08.260 00:24:08.910 Gabriel Lam: All good.

289 00:24:09.860 00:24:10.880 Samuel Roberts: Alrighty.

290 00:24:11.530 00:24:13.640 Mustafa Raja: I’ve invited you to any dimension.

291 00:24:13.910 00:24:15.150 Gabriel Lam: Awesome, thank you.

292 00:24:15.850 00:24:18.189 Samuel Roberts: But I think once we can get the streaming working.

293 00:24:19.130 00:24:23.150 Samuel Roberts: You should be able to just test by changing in length using running on the forge.

294 00:24:23.740 00:24:25.149 Samuel Roberts: Yeah, I think it runs…

295 00:24:25.150 00:24:29.990 Gabriel Lam: I can see it on Langviews, it’s more just…

296 00:24:31.960 00:24:33.869 Gabriel Lam: I’d like to see it on the forge, which…

297 00:24:34.070 00:24:36.499 Gabriel Lam: Is probably where everyone else will be using it, so…

298 00:24:36.500 00:24:37.989 Samuel Roberts: Yeah, I would say…

299 00:24:38.190 00:24:47.139 Samuel Roberts: when you make the change in length views, it should populate the NAT immediately, so, like, any change there should go, but that might also be tedious and time-consuming.

300 00:24:48.520 00:24:53.759 Samuel Roberts: Because of how long it takes to run. So, there is a way to, like, pin data in N8N,

301 00:24:53.990 00:24:56.639 Samuel Roberts: That you can rerun nodes.

302 00:25:00.920 00:25:07.789 Samuel Roberts: So, like, you could… Pin the data coming in from a given meeting.

303 00:25:08.950 00:25:14.239 Samuel Roberts: update the LangFuse, and then run the, like, LangFuse poll and the agent and see the output.

304 00:25:14.990 00:25:15.560 Gabriel Lam: Yeah.

305 00:25:17.210 00:25:18.789 Samuel Roberts: So, that might be a good way to do it.

306 00:25:18.790 00:25:20.410 Gabriel Lam: and run on parallel, yeah.

307 00:25:20.410 00:25:21.980 Samuel Roberts: Yeah, yeah, exactly. Okay, cool.

308 00:25:21.980 00:25:23.480 Gabriel Lam: Okay, awesome.

309 00:25:23.710 00:25:26.250 Samuel Roberts: Just wanted to make sure you had that knowledge there. Yes. Okay, cool.

310 00:25:26.650 00:25:27.500 Samuel Roberts: Alright.

311 00:25:28.110 00:25:30.129 Samuel Roberts: Are we good then?

312 00:25:30.130 00:25:32.839 Gabriel Lam: I think we’re good. Okay, cool. Thanks, guys.

313 00:25:33.030 00:25:38.850 Samuel Roberts: Alrighty I’ll talk to you guys later. Talk to you guys later. Bye-bye. Bye.