1
00:00:00,000 --> 00:00:04,320
Welcome to the Deep Dive. We're plunging straight into a stack of sources today

2
00:00:04,320 --> 00:00:07,360
that introduces us to something pretty interesting in the software world.

3
00:00:07,360 --> 00:00:12,000
Live Helper Chat, or LHC. Our mission really is to unpack this thing.

4
00:00:12,000 --> 00:00:15,920
It's an open source solution, apparently doing some serious heavy lifting for big

5
00:00:15,920 --> 00:00:19,040
organizations. So for you, the learner, we want to focus

6
00:00:19,040 --> 00:00:23,600
on explaining what LHC actually is, how something free can be, well, battle

7
00:00:23,600 --> 00:00:28,800
tested for enterprise use, and why its design makes it sort of a

8
00:00:28,800 --> 00:00:32,640
central hub for almost every communication and AI tool out there today.

9
00:00:32,640 --> 00:00:36,080
We want to give you the key info, the strategic bits, without getting you totally

10
00:00:36,080 --> 00:00:39,360
bogged down in technical jargon. But first, a quick word from our supporter.

11
00:00:39,360 --> 00:00:42,960
This deep dive is supported by Safe Server. Safe Server handles the hosting

12
00:00:42,960 --> 00:00:45,760
for this kind of software and supports you in your digital

13
00:00:45,760 --> 00:00:50,480
transformation journey. You can find out more at www.safeserver.de

14
00:00:50,480 --> 00:00:53,600
You know, what's really fascinating here is the contradiction. I mean, live

15
00:00:53,600 --> 00:00:56,800
customer support that's usually locked down, proprietary stuff,

16
00:00:56,800 --> 00:01:01,120
yet here's LHC delivering it at scale, completely open source under an Apache

17
00:01:01,120 --> 00:01:04,880
2.0 license. Ah, the license. That's important, right?

18
00:01:04,880 --> 00:01:08,240
Absolutely. It guarantees transparency, long-term use.

19
00:01:08,240 --> 00:01:11,680
It's probably why you see these big security-focused clients actually

20
00:01:11,680 --> 00:01:14,960
trusting it. Okay, that makes sense. That's really the core of it. So

21
00:01:14,960 --> 00:01:18,400
let's unpack this. Start from the beginning. What is Live Helper Chat,

22
00:01:18,400 --> 00:01:21,840
basically? Okay, so simply put, it's an open-source

23
00:01:21,840 --> 00:01:26,000
application designed to add powerful live support to your website or app.

24
00:01:26,000 --> 00:01:30,320
And the key thing, it's completely free. Free live support, okay.

25
00:01:30,320 --> 00:01:33,760
But, and the sources really stress this, it's not just some basic text chat

26
00:01:33,760 --> 00:01:37,760
widget, it's way more. It supports web, mobile apps, proper voice

27
00:01:37,760 --> 00:01:40,880
and video calls, even screen share. Right out of the box.

28
00:01:40,880 --> 00:01:46,400
Wow. So enterprise features, but free. For someone just starting, maybe a small

29
00:01:46,400 --> 00:01:48,960
teen, what's the setup like? Is it complicated?

30
00:01:48,960 --> 00:01:52,000
No, actually that's another plus. It's designed to be incredibly accessible.

31
00:01:52,000 --> 00:01:56,960
The basic setup, dead simple. You just need PHP version 8.2 or newer,

32
00:01:56,960 --> 00:02:00,880
and MySQL. Just PHP and MySQL, that's it. That's pretty much all you need to get a

33
00:02:00,880 --> 00:02:03,840
basic chat service up and running. Very common stuff.

34
00:02:03,840 --> 00:02:07,280
Okay, but here's the bit I'm trying to connect for the learner.

35
00:02:07,280 --> 00:02:12,000
Simple setup, fine. But enterprise customization,

36
00:02:12,000 --> 00:02:15,520
how does it handle that without locking you in like

37
00:02:15,520 --> 00:02:18,960
proprietary software often does? Right, good question. That comes down to the

38
00:02:18,960 --> 00:02:22,240
architecture. It's built using a modern MVC structure.

39
00:02:22,240 --> 00:02:25,840
MVC, model view controller. Exactly. If you're not a developer,

40
00:02:25,840 --> 00:02:29,840
just think of it as separating the app's brain, the logic, from how it looks, the

41
00:02:29,840 --> 00:02:33,440
appearance. The view part, the look and feel, uses

42
00:02:33,440 --> 00:02:36,240
the Bootstrap 5 CSS framework. Very standard,

43
00:02:36,240 --> 00:02:40,000
very reliable. Okay. So because the look and the core code are separate,

44
00:02:40,000 --> 00:02:43,040
you can customize the design fonts, colors, layout,

45
00:02:43,040 --> 00:02:47,040
everything without ever needing to touch the main operational code,

46
00:02:47,040 --> 00:02:51,120
the kernel. I see. So you can paint the car without messing with the engine.

47
00:02:51,120 --> 00:02:54,560
That's a good way to put it. And that flexibility is really key to why it's so

48
00:02:54,560 --> 00:02:58,080
widely adopted and constantly updated by the community.

49
00:02:58,080 --> 00:03:02,160
It's perfect sense. Okay. Customizable shell, reliable engine.

50
00:03:02,160 --> 00:03:07,600
Now, the elephant in the room, volume. If something's free and open source,

51
00:03:07,600 --> 00:03:12,480
people always worry. Can it scale? Can it handle the kind of traffic

52
00:03:12,480 --> 00:03:15,840
I don't know, big bank needs? It definitely can. Looking at the metrics

53
00:03:15,840 --> 00:03:19,920
they share, load isn't an issue. They have production environments handling over,

54
00:03:19,920 --> 00:03:25,120
get this, 10,000 chats per day. 10,000 a day.

55
00:03:25,120 --> 00:03:29,120
Yep. With dozens of operators working actively at the same time, all running

56
00:03:29,120 --> 00:03:33,760
smoothly. It's not just scalable. It's optimized specifically for those high

57
00:03:33,760 --> 00:03:36,400
load situations. Okay. That's impressive. But hold on. We just

58
00:03:36,400 --> 00:03:41,280
said free, open source. Are we really saying major banks, health sectors,

59
00:03:41,280 --> 00:03:44,800
places where data protection is everything, are trusting this over some

60
00:03:44,800 --> 00:03:48,160
expensive license thing? Why? What gives them the confidence?

61
00:03:48,160 --> 00:03:51,520
Well, I think it boils down to two things. Transparency and capability. First,

62
00:03:51,520 --> 00:03:55,280
because the code is open, that Apache 2.0 license, again, these big institutions

63
00:03:55,280 --> 00:03:58,720
can literally audit every single line. Ah, so they can check it themselves for

64
00:03:58,720 --> 00:04:01,280
security holes. Exactly. They're not just taking a

65
00:04:01,280 --> 00:04:05,760
vendor's word for it. They can verify the code meets their standards, ensures

66
00:04:05,760 --> 00:04:09,520
compliance. Second, to handle that kind of scale, it

67
00:04:09,520 --> 00:04:13,360
uses optional tools that are themselves industry standards for

68
00:04:13,360 --> 00:04:16,960
high performance. Like what? Well, for example, to get stats

69
00:04:16,960 --> 00:04:19,840
instantly across millions of old chats, they integrate

70
00:04:19,840 --> 00:04:23,200
Elasticsearch. Elasticsearch. Okay, for the learner,

71
00:04:23,200 --> 00:04:26,240
why is that needed? Why not just use the regular database?

72
00:04:26,240 --> 00:04:30,080
Good point. Elasticsearch is basically a super fast search engine

73
00:04:30,080 --> 00:04:34,160
built for huge amounts of data. If you tried running complex reports on, say,

74
00:04:34,160 --> 00:04:38,720
a year's worth of chats at 10,000 a day using just a standard MySQL database,

75
00:04:38,720 --> 00:04:42,240
it would likely slow everything down, maybe even grind to a halt.

76
00:04:42,240 --> 00:04:45,280
Right. Elasticsearch lets LHC pull statistics on

77
00:04:45,280 --> 00:04:48,640
millions of chats in like seconds. It justifies itself pretty quickly in

78
00:04:48,640 --> 00:04:52,480
those really high volume places. They also use things like Node.js and

79
00:04:52,480 --> 00:04:55,920
background workers to handle heavy tasks away from the main application.

80
00:04:55,920 --> 00:04:58,480
Offloading the work. Yeah, keeps things snappy.

81
00:04:58,480 --> 00:05:01,360
And it even supports Amazon S3 for file storage.

82
00:05:01,360 --> 00:05:04,480
S3. So basically infinite storage for uploads.

83
00:05:04,480 --> 00:05:08,080
Pretty much. You never really have to worry about running out of space for

84
00:05:08,080 --> 00:05:12,480
documents or images shared in chats. Wow, okay. So it can handle the load.

85
00:05:12,480 --> 00:05:18,180
But volume's one thing, intelligence is another. To handle 10k chats and connect to

86
00:05:18,180 --> 00:05:18,640
everything.

87
00:05:18,640 --> 00:05:23,360
That architecture must be pretty flexible. You mentioned it acts like a hub.

88
00:05:23,360 --> 00:05:27,280
Exactly. That seems to be the core design philosophy. Make it easy to plug anything

89
00:05:27,280 --> 00:05:27,600
in.

90
00:05:27,600 --> 00:05:32,390
Especially AI. The sheer range of AI integrations is probably its biggest strategic

91
00:05:32,390 --> 00:05:33,120
advantage.

92
00:05:33,120 --> 00:05:35,120
We're not just talking about hooking up one chat button.

93
00:05:35,120 --> 00:05:38,720
Okay, so what specific AIs are we talking about? What are the sources mentioned?

94
00:05:38,720 --> 00:05:42,780
Well, pretty much all the main generative ones. It supports chat GPT and not just

95
00:05:42,780 --> 00:05:43,440
basic chat,

96
00:05:43,440 --> 00:05:47,650
but the more powerful stuff like chat completion, generating responses, even the

97
00:05:47,650 --> 00:05:48,480
assistant API.

98
00:05:48,480 --> 00:05:52,800
But beyond that, it connects to other big platforms. Rasa AI,

99
00:05:52,800 --> 00:05:57,120
Alama AI for maybe self-hosted models, and Google's Gemini too.

100
00:05:57,120 --> 00:06:01,680
So flexibility. A company can pick the AI that fits their needs or budget.

101
00:06:01,680 --> 00:06:04,800
Precisely. And manage it all through the one LHC interface.

102
00:06:04,800 --> 00:06:09,620
And you said it uses AI for more than just chatting back to customers, like for

103
00:06:09,620 --> 00:06:10,800
management tasks.

104
00:06:10,800 --> 00:06:14,320
Yes. And that's really clever. It uses AI for practical stuff.

105
00:06:14,320 --> 00:06:18,160
For instance, it uses Deepavlov for sentiment analysis.

106
00:06:18,160 --> 00:06:23,840
Sentiment analysis. So figuring out if a customer is happy or angry in real time.

107
00:06:23,840 --> 00:06:27,920
Kind of, yeah. Gaging the emotional tone of the conversation as it's happening.

108
00:06:27,920 --> 00:06:31,280
And even more practically, it includes insult detection.

109
00:06:31,280 --> 00:06:35,040
Insult detection. Wow. What's the point of that? I mean, beyond the obvious.

110
00:06:35,040 --> 00:06:38,000
Well, think about agent well-being and management oversight.

111
00:06:38,000 --> 00:06:42,390
In a busy contact center, the system can automatically flag chats where staff might

112
00:06:42,390 --> 00:06:43,200
be getting abuse.

113
00:06:43,200 --> 00:06:45,360
So a supervisor can step in.

114
00:06:45,360 --> 00:06:50,160
Right. Immediately. Protect operators from burnout, maintain service quality, stop

115
00:06:50,160 --> 00:06:51,280
things escalating.

116
00:06:51,280 --> 00:06:55,440
It makes the platform proactive, not just reactive. It's a safeguard.

117
00:06:55,440 --> 00:06:58,720
That's actually a really powerful use case. Okay, let's shift the customer

118
00:06:58,720 --> 00:06:59,360
connection points.

119
00:06:59,360 --> 00:07:03,040
You said it connects to loads of channels, more than just the website widget.

120
00:07:03,040 --> 00:07:07,380
Oh, yeah. It genuinely seems to connect to almost every major channel people use

121
00:07:07,380 --> 00:07:07,920
daily.

122
00:07:07,920 --> 00:07:12,000
We're talking Telegram, which you can also use as a mobile app for the operators

123
00:07:12,000 --> 00:07:12,400
themselves,

124
00:07:12,400 --> 00:07:13,440
which is huge.

125
00:07:13,440 --> 00:07:15,840
Oh, interesting. Operators can use Telegram.

126
00:07:15,840 --> 00:07:20,290
Yeah, as their client. Plus Facebook Messenger, Facebook WhatsApp, Twilio for both

127
00:07:20,290 --> 00:07:21,280
SMS and direct

128
00:07:21,280 --> 00:07:26,160
WhatsApp, Viber, Discord, even Mattermost for internal or community chats.

129
00:07:26,160 --> 00:07:30,160
They have a lot. One interface manages conversations coming from all those

130
00:07:30,160 --> 00:07:30,800
different places.

131
00:07:30,800 --> 00:07:34,560
Exactly. Massive consolidation saves so much time and effort.

132
00:07:34,560 --> 00:07:38,160
And you mentioned voice and video earlier. That's often a big jump from just text

133
00:07:38,160 --> 00:07:38,560
chat.

134
00:07:38,560 --> 00:07:43,530
It is. But LHD handles it. You get Jitsi integration for free right out of the box.

135
00:07:43,530 --> 00:07:43,840
Voice,

136
00:07:43,840 --> 00:07:46,400
video, screen sharing, easy win.

137
00:07:46,400 --> 00:07:48,480
Jitsi. OK, free and open source too.

138
00:07:48,480 --> 00:07:54,000
Correct. But if you're a bigger enterprise needing, you know, guaranteed uptime top

139
00:07:54,000 --> 00:07:58,210
performance for really critical calls, there's also an option to integrate with the

140
00:07:58,210 --> 00:07:58,480
paid

141
00:07:58,480 --> 00:07:59,600
Agora platform.

142
00:07:59,600 --> 00:08:00,320
Agora.

143
00:08:00,320 --> 00:08:04,800
Yeah, it's a high performance enterprise grade provider for voice, video and screen

144
00:08:04,800 --> 00:08:05,360
share.

145
00:08:05,360 --> 00:08:07,680
So you have choices depending on your needs.

146
00:08:07,680 --> 00:08:12,000
That clarifies things. Free option with Jitsi, paid enterprise stability with Agora.

147
00:08:12,000 --> 00:08:17,200
It really brings up that question again, doesn't it? Why spend time and money

148
00:08:17,200 --> 00:08:17,760
building separate

149
00:08:17,760 --> 00:08:23,200
systems for chat, video, AI messaging, when this one open source tool seems to

150
00:08:23,200 --> 00:08:24,640
connect them all.

151
00:08:24,640 --> 00:08:26,240
And you can audit the code yourself.

152
00:08:26,240 --> 00:08:30,880
It really leans into centralization and efficiency, makes a strong case.

153
00:08:30,880 --> 00:08:34,800
Okay, let's pivot slightly. Let's talk about the features that make life easier for

154
00:08:34,800 --> 00:08:34,960
the

155
00:08:34,960 --> 00:08:39,280
operators day to day, and maybe improve the customer's experience directly.

156
00:08:39,280 --> 00:08:42,400
Right. For the agents, there's some pretty sophisticated stuff, good performance

157
00:08:42,400 --> 00:08:42,880
tracking,

158
00:08:42,880 --> 00:08:48,080
priority queues. So important chats get handled first. And this one's surprising.

159
00:08:48,080 --> 00:08:50,880
The ability to see what users are typing before they hit send.

160
00:08:50,880 --> 00:08:53,520
Whoa, really? Like real-time typing preview?

161
00:08:53,520 --> 00:08:56,720
Yeah. Imagine the advantage you can anticipate the question,

162
00:08:56,720 --> 00:09:00,540
maybe even have an answer ready just as they finish typing. Big time saver, better

163
00:09:00,540 --> 00:09:01,040
service.

164
00:09:01,040 --> 00:09:04,240
That sounds slightly creepy, but I can see the efficiency gain.

165
00:09:04,240 --> 00:09:05,760
And what about canned messages?

166
00:09:05,760 --> 00:09:09,040
Ah, yes. Delayed canned messages. That's smart, too.

167
00:09:09,680 --> 00:09:13,680
It lets operators use standard pre-written answers for common questions,

168
00:09:13,680 --> 00:09:16,480
but the system sends them with a slight delay.

169
00:09:16,480 --> 00:09:17,920
So it looks like they're typing it out.

170
00:09:17,920 --> 00:09:22,480
Exactly. It feels more personal, less robotic, maintains that human touch,

171
00:09:22,480 --> 00:09:26,400
but still massively cuts down how long it takes to handle each chat.

172
00:09:26,400 --> 00:09:28,800
Efficiency without sacrificing the experience.

173
00:09:28,800 --> 00:09:34,400
Clever. Okay, now flip side. The customer experience and tracking.

174
00:09:34,400 --> 00:09:35,840
You mentioned tracking was detailed.

175
00:09:35,840 --> 00:09:38,960
Oh, very. It gives operators a ton of context.

176
00:09:38,960 --> 00:09:42,160
Real-time tracking of who's online, where they are geographically

177
00:09:42,160 --> 00:09:46,400
uses three different sources for accuracy, tracks every page they view,

178
00:09:46,400 --> 00:09:50,960
how long they spend on the site overall, and, really useful for marketing,

179
00:09:50,960 --> 00:09:52,960
it tracks the first time a user ever visited.

180
00:09:52,960 --> 00:09:56,400
The first visit. So it builds a history even for anonymous visitors.

181
00:09:56,400 --> 00:10:00,240
Seems so. Even if they never chat, the system knows their journey through the site

182
00:10:00,240 --> 00:10:00,880
over time.

183
00:10:00,880 --> 00:10:04,160
And that tracking helps route chats intelligently, you said.

184
00:10:04,160 --> 00:10:08,780
Absolutely. It's very flexible. You can set rules to send visitors to specific

185
00:10:08,780 --> 00:10:09,440
operators

186
00:10:09,440 --> 00:10:13,200
based on location or language. The source specifically mentions

187
00:10:13,200 --> 00:10:18,160
routing, say, a visitor from Lithuania directly to an operator who speaks Lithuanian.

188
00:10:18,160 --> 00:10:21,920
Makes sense. Immediate connection, better service, builds trust faster.

189
00:10:21,920 --> 00:10:23,840
Definitely. Big impact on quality.

190
00:10:23,840 --> 00:10:27,840
Okay, rounding out the features. What about security beyond the code auditing,

191
00:10:27,840 --> 00:10:30,320
and how do operators actually access this?

192
00:10:30,320 --> 00:10:34,480
For access, it's got dedicated native apps, Windows, Linux,

193
00:10:34,480 --> 00:10:38,000
Mac, so operators can use it easily on whatever system they have.

194
00:10:38,000 --> 00:10:40,960
Good cross-platform. Yep. Security-wise, it supports

195
00:10:40,960 --> 00:10:44,400
two-factor authentication, 2FA, using authenticator apps.

196
00:10:44,400 --> 00:10:48,800
And of course, HTTPS encryption is standard, essential stuff.

197
00:10:48,800 --> 00:10:52,160
And importantly, they seem to have tackled the spam problem

198
00:10:52,160 --> 00:10:55,600
quite elegantly. Spambots in chat. Always annoying.

199
00:10:55,600 --> 00:10:59,120
Right. It uses advanced CAPTCHA techniques, but apparently without making the

200
00:10:59,120 --> 00:10:59,680
actual

201
00:10:59,680 --> 00:11:03,200
human users type in codes. No annoying click the traffic lights thing.

202
00:11:03,200 --> 00:11:07,040
Seems not. So it blocks the bots without adding friction for real visitors.

203
00:11:07,040 --> 00:11:09,520
It's all about efficiency and safety working together.

204
00:11:09,520 --> 00:11:12,640
Okay, let's try and wrap this up. What does this all mean? We've seen live

205
00:11:12,640 --> 00:11:16,560
helper chat is completely free, open source, under

206
00:11:16,560 --> 00:11:19,440
that Apache 2.0 license. You can check the code.

207
00:11:19,440 --> 00:11:23,760
Yet it's robust enough for banks, universities,

208
00:11:23,760 --> 00:11:27,600
handling, what was it, 10,000-plus chats a day? Yeah, that's the figure they quote

209
00:11:27,600 --> 00:11:31,840
for high load environments. And its real power seems to be acting as this

210
00:11:31,840 --> 00:11:36,480
central hub, pulling in AI like ChatGPT and Gemini,

211
00:11:36,480 --> 00:11:39,600
connecting all the messaging apps like WhatsApp, Telegram,

212
00:11:39,600 --> 00:11:43,280
all into one place. Right, I think the core takeaway for you, the learner,

213
00:11:43,280 --> 00:11:46,640
is that you don't always need massive license fees for sophisticated digital

214
00:11:46,640 --> 00:11:49,360
infrastructure. This demonstrates open source can deliver

215
00:11:49,360 --> 00:11:52,400
that complexity, even niche things like AI for insult

216
00:11:52,400 --> 00:11:55,680
detection or sentiment analysis, while still being fundamentally simple

217
00:11:55,680 --> 00:11:58,000
to set up and trustworthy because you can see the code.

218
00:11:58,000 --> 00:12:01,200
A final thank you again to our supporter SafeServer.

219
00:12:01,200 --> 00:12:04,400
They provide hosting for this kind of advanced software and assist with your

220
00:12:04,400 --> 00:12:10,960
digital transformation. You can find more info at www.safeserver.de.

221
00:12:10,960 --> 00:12:14,320
Now here's something to think about, a final thought for you to mull over.

222
00:12:14,320 --> 00:12:17,680
We heard LHC tracks not just visits, but the first visit,

223
00:12:17,680 --> 00:12:21,440
how long someone stays on a page and can even apparently grab a screenshot of

224
00:12:21,440 --> 00:12:24,400
the user's page. How does having that level of granular

225
00:12:24,400 --> 00:12:27,680
detail on anonymous visitors fundamentally change the relationship

226
00:12:27,680 --> 00:12:31,120
between a business and the people just browsing its site? Does it shift

227
00:12:31,120 --> 00:12:33,520
this from just a support tool into something more like a

228
00:12:33,520 --> 00:12:36,800
well a marketing intelligence or even an espionage engine?

229
00:12:36,800 --> 00:12:40,080
That's something to maybe explore on your own. Thank you for joining us for

230
00:12:40,080 --> 00:12:43,600
the Deep Dive, we'll catch you next time.