Episode 26
Meta's AI Power Play: Llama 3.1 405b, Reel Search
Summary In this episode, Dalton Anderson discusses Meta's upcoming release of their 405 billion parameter model. He mentions that the model will be open source but the model weights may not…
Summary In this episode, Dalton Anderson discusses Meta's upcoming release of their 405 billion parameter model. He mentions that the model will be open source but the model weights may not be released. He also talks about Meta's new feature on their Meta AI website that allows users to search for reels. Dalton shares his experience using the feature for fashion inspiration and discusses the potential use cases for the feature. He also provides updates on the progress of his home remodeling project.
Episode content
Explore every layer of this episode.
Each article, guide, analysis, and field note has its own focused page and stays linked to this source conversation.
Articles & stories
Narrative and editorial pieces that carry the conversation forward.
Llama 3.1 Prelaunch and Release Timeline
A dated Llama 3.1 timeline separating Venture Step's prelaunch expectations from Meta's announcement, model card, license, weights, and later evidence.
What E026 Reveals About AI Claims Before Launch
Revisit Venture Step E026 before Llama 3.1 shipped, including uncertain model claims, a live Reel search test, and lessons for evaluating AI products.
How to Build an AI Feature Claim Ledger
Track changing AI features by claim, source, evidence state, version, account, region, observed test, public wording, correction, and refresh trigger.
Guides & how-tos
Practical ways to apply the episode's ideas.
How to Verify an AI Model Claim Before Release
A practical evidence ladder for checking AI model rumors, leaks, demos, previews, announcements, model cards, artifacts, evaluations, and corrections.
How to Evaluate Semantic Search for a Media Library
Build a representative semantic-search test for media using relevance judgments, baselines, retrieval metrics, privacy checks, latency, and failure analysis.
How to Design Permission Boundaries for an AI Smart Home
Map people, devices, data, actions, confirmations, logs, expiry, recovery, and manual control before an AI assistant can operate a connected home.
Research & analysis
Evidence-led work that tests and expands the claims in the conversation.
Semantic Media Search Evaluation Protocol
A semantic media-search evaluation starts with a frozen corpus, index, model, configuration, and representative query set. Each query needs an information need and graded
Search, Recommendation, and Generation System Map
Search begins with an expressed information need. The system interprets a query, retrieves candidates from an eligible corpus, and ranks them for that need.
Prelaunch AI Claim Evidence Ladder
Treat each model statement as an atomic claim. Give it one current evidence state: rumor, attributed report, controlled preview, publisher announcement, released document
Llama 3.1 Claim State Timeline Record
E026 was recorded before Meta released Llama 3.1 on July 23, 2024. The captions preserve uncertainty about the model's name, parameter count, availability, weights, licen
E026 Episode Recording Evidence Record
The public YouTube recording is the authoritative spoken-evidence source for E026. Its English auto-generated captions were recovered on July 28, 2026 and preserved both
AI Smart Home Permission Boundary Framework
Model household access across people, roles, devices, data, actions, locations, time windows, consequences, confirmation, logging, expiry, recovery, and manual control.
AI Feature Claim Ledger Framework
The ledger unit is one falsifiable claim, not one product page. Record the exact claim, product, feature, version, account, plan, platform, region, language, date, source
Field notes
Focused observations and durable ideas worth carrying into other work.
Full episode
Read the complete record.
The show notes, transcript, and source trail remain on this canonical episode page.
TranscriptRead the full conversation.
Meta's AI Power Play: Llama, Reel Search, and Home Update
Source fidelity
The timestamped transcript below is the exact English auto-generated SRT caption file recovered from the preserved YouTube episode on 2026-07-28. No spelling, punctuation, speaker, or wording corrections were applied. The original SRT is preserved beside this note as E26 - YouTube captions - J2I1fJW1sB4.en.srt.
Timestamped transcript
1 00:00:00,120 --> 00:00:01,680 welcome to V step podcast where we
2 00:00:01,680 --> 00:00:03,879 discuss entrepreneurship industry Trends
3 00:00:03,879 --> 00:00:06,160 and the occasional B review today we're
4 00:00:06,160 --> 00:00:08,760 going to be discussing meta's exciting
5 00:00:08,760 --> 00:00:11,040 release that doesn't happen to line up
6 00:00:11,040 --> 00:00:12,759 with the episode so I can't do a live
7 00:00:12,759 --> 00:00:15,320 demo but we'll be briefly discussing it
8 00:00:15,320 --> 00:00:18,840 it's their 48 billion parameter model
9 00:00:18,840 --> 00:00:21,920 that they were releasing later this week
10 00:00:21,920 --> 00:00:23,320 hopefully it doesn't get delayed I think
11 00:00:23,320 --> 00:00:24,880 there's been a lot of hype to it so I'm
12 00:00:24,880 --> 00:00:28,560 pretty sure that Mark will release the
13 00:00:28,560 --> 00:00:30,119 model and it's going to be open source
14 00:00:30,119 --> 00:00:31,640 which was confirmed I think that they're
15 00:00:31,640 --> 00:00:34,000 not releasing the model weights I think
16 00:00:34,000 --> 00:00:35,520 there was some general push back from
17 00:00:35,520 --> 00:00:38,000 the public like hey if it's open source
18 00:00:38,000 --> 00:00:40,719 let's make sure we release the model
19 00:00:40,719 --> 00:00:44,879 weights that and meta released a new
20 00:00:44,879 --> 00:00:47,160 feature on their meta AI website where
21 00:00:47,160 --> 00:00:49,360 you can search reals and I think that
22 00:00:49,360 --> 00:00:51,039 there's some interesting use cases
23 00:00:51,039 --> 00:00:53,079 potentially people might use them for
24 00:00:53,079 --> 00:00:55,920 and how it might be useful to yourself
25 00:00:55,920 --> 00:00:58,160 and what I was using it for as well and
26 00:00:58,160 --> 00:01:00,680 then some general updates on
27 00:01:00,680 --> 00:01:03,359 the success of the the jobs and the
28 00:01:03,359 --> 00:01:06,320 nightmares that I had to working on on
29 00:01:06,320 --> 00:01:08,040 Nana's home the last couple weeks I had
30 00:01:08,040 --> 00:01:10,600 some some scares and we got through them
31 00:01:10,600 --> 00:01:12,240 it's a general sense of what we will be
32 00:01:12,240 --> 00:01:15,240 T discussing today before we dive in of
33 00:01:15,240 --> 00:01:18,119 course my name's Dalton Anderson I'm
34 00:01:18,119 --> 00:01:20,240 your host my background is a bit of a
35 00:01:20,240 --> 00:01:21,840 mix of programming data science
36 00:01:21,840 --> 00:01:23,320 Insurance offline you can find me
37 00:01:23,320 --> 00:01:24,920 running building my side business or
38 00:01:24,920 --> 00:01:28,200 lost in a good book if you want to
39 00:01:28,200 --> 00:01:30,400 listen to the podcast and video in audio
40 00:01:30,400 --> 00:01:32,799 format
41 00:01:32,799 --> 00:01:36,119 YouTube is the place to be if audio is
42 00:01:36,119 --> 00:01:38,119 more your thing you can find the podcast
43 00:01:38,119 --> 00:01:40,000 on Apple podcast
44 00:01:40,000 --> 00:01:43,960 Spotify or wherever else you get your
45 00:01:43,960 --> 00:01:48,079 podcast okay llama 3 48 billion
46 00:01:48,079 --> 00:01:51,719 parameter model so right now LL 3 they
47 00:01:51,719 --> 00:01:55,320 have a 70 billion in an seven or 8
48 00:01:55,320 --> 00:01:57,000 billion they have a seven they have an 8
49 00:01:57,000 --> 00:01:58,880 billion they have an 8 million parameter
50 00:01:58,880 --> 00:02:01,119 model a 70 billion parameter model and
51 00:02:01,119 --> 00:02:03,399 those seven and 8 billion parameter
52 00:02:03,399 --> 00:02:07,160 models performed very well against their
53 00:02:07,160 --> 00:02:10,840 counterparts and in some cases even they
54 00:02:10,840 --> 00:02:13,360 were lumped in so typically they'll
55 00:02:13,360 --> 00:02:15,760 classify the models like say like small
56 00:02:15,760 --> 00:02:18,080 medium large like a large would be like
57 00:02:18,080 --> 00:02:19,800 okay this 400 billion perameter model we
58 00:02:19,800 --> 00:02:22,519 compared with other large models they
59 00:02:22,519 --> 00:02:24,319 don't necessarily have the same
60 00:02:24,319 --> 00:02:27,360 parameters the amount of parameters so
61 00:02:27,360 --> 00:02:30,680 in certain cases like this llama 38
62 00:02:30,680 --> 00:02:32,680 billion parameter models compared to
63 00:02:32,680 --> 00:02:35,160 other models with 15 billion or 20
64 00:02:35,160 --> 00:02:35,959 billion
65 00:02:35,959 --> 00:02:37,879 parameters and There's 7 billion
66 00:02:37,879 --> 00:02:41,599 parameter model that's mainly made to
67 00:02:41,599 --> 00:02:43,760 work on wearables
68 00:02:43,760 --> 00:02:49,480 like the metag glasses or I don't I know
69 00:02:49,480 --> 00:02:52,920 the metag glasses uses their small model
70 00:02:52,920 --> 00:02:56,560 they're they're 8 billion but I'm unsure
71 00:02:56,560 --> 00:03:01,319 whether Quest meta Quest uses the8 or
72 00:03:01,319 --> 00:03:04,519 the 70 that I don't
73 00:03:04,519 --> 00:03:08,280 know but basically the models perform
74 00:03:08,280 --> 00:03:11,440 fairly well against their peers in that
75 00:03:11,440 --> 00:03:14,720 same category with most of the time
76 00:03:14,720 --> 00:03:18,200 those respective meta models having less
77 00:03:18,200 --> 00:03:19,799 parameters
78 00:03:19,799 --> 00:03:22,319 than their counterparts which is pretty
79 00:03:22,319 --> 00:03:24,920 cool and I was pretty I wasn't pretty
80 00:03:24,920 --> 00:03:29,000 impr I was impressed with how the models
81 00:03:29,000 --> 00:03:31,560 performed with the things I asked it it
82 00:03:31,560 --> 00:03:36,000 it seemed generally pretty legit for
83 00:03:36,000 --> 00:03:38,519 only 70 billion parameters because when
84 00:03:38,519 --> 00:03:41,120 you go to mea.com you're only using the
85 00:03:41,120 --> 00:03:43,599 70 billion parameter model because the
86 00:03:43,599 --> 00:03:47,400 408 hasn't been released yet but what is
87 00:03:47,400 --> 00:03:50,840 really cool about meta approach and
88 00:03:50,840 --> 00:03:53,920 similar to x.com or I guess otherwise
89 00:03:53,920 --> 00:03:57,000 known as Twitter old Twitter or X it's
90 00:03:57,000 --> 00:03:59,480 called X now X and meta have taken the
91 00:03:59,480 --> 00:04:00,280 approach
92 00:04:00,280 --> 00:04:02,519 of putting large sums of money to train
93 00:04:02,519 --> 00:04:04,840 these AI models and then open sourcing
94 00:04:04,840 --> 00:04:08,799 them and so if you go to mea.com there's
95 00:04:08,799 --> 00:04:11,760 a big button in bold to download the
96 00:04:11,760 --> 00:04:13,280 models all you got to do is fill out
97 00:04:13,280 --> 00:04:16,840 your name like who you are and an email
98 00:04:16,840 --> 00:04:18,320 and you can download the models that's
99 00:04:18,320 --> 00:04:20,320 it you just got to register yourself and
100 00:04:20,320 --> 00:04:22,280 they allow you to download the models
101 00:04:22,280 --> 00:04:25,400 apparently with the leaks that they're
102 00:04:25,400 --> 00:04:29,080 saying there was some leaks of the 48
103 00:04:29,080 --> 00:04:30,600 Brillion parameter
104 00:04:30,600 --> 00:04:35,919 model and it was a download payload
105 00:04:35,919 --> 00:04:38,720 of I think 800 gigs or something like
106 00:04:38,720 --> 00:04:42,759 that so it's almost terabyte of of stuff
107 00:04:42,759 --> 00:04:44,360 that you have to download to run the
108 00:04:44,360 --> 00:04:46,440 model and that's not even to run the
109 00:04:46,440 --> 00:04:49,080 model that's to get your M get the model
110 00:04:49,080 --> 00:04:51,000 on your machine for you to run it you
111 00:04:51,000 --> 00:04:53,880 probably need to either combine the
112 00:04:53,880 --> 00:04:57,280 resources of multiple computers or have
113 00:04:57,280 --> 00:04:58,639 a really
114 00:04:58,639 --> 00:05:02,280 powerful GPU like rig or something like
115 00:05:02,280 --> 00:05:07,120 that to to use for compute for the model
116 00:05:07,120 --> 00:05:08,360 I don't know I haven't really looked
117 00:05:08,360 --> 00:05:10,000 into it that much but I do know that
118 00:05:10,000 --> 00:05:12,360 it's too much to run on a computer one
119 00:05:12,360 --> 00:05:13,680 computer at least you could maybe
120 00:05:13,680 --> 00:05:16,680 combine two really good computers and be
121 00:05:16,680 --> 00:05:18,639 able to run a model but not to its full
122 00:05:18,639 --> 00:05:21,039 capabilities obviously that being said
123 00:05:21,039 --> 00:05:23,360 you could you could pick up and download
124 00:05:23,360 --> 00:05:25,600 and make your own little app for free
125 00:05:25,600 --> 00:05:27,800 the 70 billion or the 8 billion
126 00:05:27,800 --> 00:05:29,199 parameter model like that's not that
127 00:05:29,199 --> 00:05:31,400 much like a billion parameter models
128 00:05:31,400 --> 00:05:33,639 like made just be stored on watches
129 00:05:33,639 --> 00:05:36,280 glasses and stuff like that and that's
130 00:05:36,280 --> 00:05:38,280 what they I know for a fact that's what
131 00:05:38,280 --> 00:05:40,600 meta uses for their what what are they
132 00:05:40,600 --> 00:05:42,280 called meta Rayband
133 00:05:42,280 --> 00:05:44,520 glasses that's the model that they use
134 00:05:44,520 --> 00:05:47,039 and people have said that they find it
135 00:05:47,039 --> 00:05:49,240 useful and eventually they're going to
136 00:05:49,240 --> 00:05:53,880 get it to the point of being able to see
137 00:05:53,880 --> 00:05:55,000 like you'll have to give a permission
138 00:05:55,000 --> 00:05:59,240 like hey hey meta can you look at what
139 00:05:59,240 --> 00:06:02,880 I'm looking at can you tell me what year
140 00:06:02,880 --> 00:06:04,600 this car was produced or something like
141 00:06:04,600 --> 00:06:07,160 that maybe you're looking at an antique
142 00:06:07,160 --> 00:06:09,960 car or something I don't know good
143 00:06:09,960 --> 00:06:12,479 example but or how to make this and
144 00:06:12,479 --> 00:06:14,039 maybe you ask it how to make a certain
145 00:06:14,039 --> 00:06:15,960 type of dish like what what ingredients
146 00:06:15,960 --> 00:06:18,160 is the dis dish typically have or maybe
147 00:06:18,160 --> 00:06:19,400 you don't even know the dish and you're
148 00:06:19,400 --> 00:06:21,479 like what is this dish and maybe it
149 00:06:21,479 --> 00:06:24,720 finds out for you but it at one point
150 00:06:24,720 --> 00:06:27,919 you'll be able to talk to your meta and
151 00:06:27,919 --> 00:06:31,560 like hey wi's this play open up and then
152 00:06:31,560 --> 00:06:34,400 without going to your phone meta will
153 00:06:34,400 --> 00:06:37,560 search the not the internet because it's
154 00:06:37,560 --> 00:06:39,440 meta glasses this is a different
155 00:06:39,440 --> 00:06:41,479 situation eventually they would wanted
156 00:06:41,479 --> 00:06:42,919 to search the internet but I I don't
157 00:06:42,919 --> 00:06:44,360 know where that's going to be with the
158 00:06:44,360 --> 00:06:46,479 glasses U maybe it's connected to your
159 00:06:46,479 --> 00:06:49,639 phone via Bluetooth and somehow the
160 00:06:49,639 --> 00:06:51,639 Bluetooth corresponds and gives commands
161 00:06:51,639 --> 00:06:53,960 to your phone to search back gives meta
162 00:06:53,960 --> 00:06:56,599 glasses back the information and and
163 00:06:56,599 --> 00:06:58,879 then meta talks back to you that might
164 00:06:58,879 --> 00:07:00,360 take too long you might as well just
165 00:07:00,360 --> 00:07:02,360 look it up on your phone maybe not for
166 00:07:02,360 --> 00:07:06,560 that kind of stuff that is super duper
167 00:07:06,560 --> 00:07:09,280 Dynamic but things that might be set in
168 00:07:09,280 --> 00:07:12,360 stone for a while might be better but
169 00:07:12,360 --> 00:07:15,720 basically long story short eventually
170 00:07:15,720 --> 00:07:17,639 metag glasses will allow you to search
171 00:07:17,639 --> 00:07:21,360 for stuff on your face without you
172 00:07:21,360 --> 00:07:23,960 having to take out your phone which is
173 00:07:23,960 --> 00:07:26,560 nice if you're down for that I don't
174 00:07:26,560 --> 00:07:28,199 know I think it's pretty cool I don't
175 00:07:28,199 --> 00:07:30,479 know how everyone else thinks about it
176 00:07:30,479 --> 00:07:33,080 but I find it pretty interesting so I
177 00:07:33,080 --> 00:07:35,160 would definitely do it I would buy the
178 00:07:35,160 --> 00:07:37,919 glasses but I before I buy the glasses I
179 00:07:37,919 --> 00:07:42,319 would probably buy of meta what is it
180 00:07:42,319 --> 00:07:44,960 meta Quest or yeah meta Quest 3 I would
181 00:07:44,960 --> 00:07:46,840 buy that before I bought the glasses
182 00:07:46,840 --> 00:07:50,120 just so I could give cool demos I could
183 00:07:50,120 --> 00:07:51,560 talk about it in the
184 00:07:51,560 --> 00:07:53,960 podcast and who knows maybe I could just
185 00:07:53,960 --> 00:07:56,879 start writing this stuff off you know I
186 00:07:56,879 --> 00:07:58,319 needed I needed to talk about the
187 00:07:58,319 --> 00:08:00,879 podcast like why not
188 00:08:00,879 --> 00:08:03,240 why not make a make a business expense
189 00:08:03,240 --> 00:08:07,360 out of it but anyways
190 00:08:07,360 --> 00:08:11,240 so the 48 billion parameter model is
191 00:08:11,240 --> 00:08:14,319 more of a Enterprise level model this
192 00:08:14,319 --> 00:08:17,479 model would compete with the chat
193 00:08:17,479 --> 00:08:19,639 GPT
194 00:08:19,639 --> 00:08:21,800 4.0 chap GT
195 00:08:21,800 --> 00:08:25,280 40 the Gemini
196 00:08:25,280 --> 00:08:29,599 Advanced or the pro like 1.5 and so that
197 00:08:29,599 --> 00:08:33,399 that's where the 4 point or 48 billion
198 00:08:33,399 --> 00:08:37,000 parameter model is positioned when it
199 00:08:37,000 --> 00:08:39,519 gets released so it's supposed to be an
200 00:08:39,519 --> 00:08:41,839 open source
201 00:08:41,839 --> 00:08:44,080 competition or an open source model that
202 00:08:44,080 --> 00:08:47,880 competes with open Ai and
203 00:08:47,880 --> 00:08:51,440 Google and anthropic with their CLA will
204 00:08:51,440 --> 00:08:55,240 it do that I don't know it from the
205 00:08:55,240 --> 00:08:56,880 leaks it seems like it's going to be
206 00:08:56,880 --> 00:08:57,959 very
207 00:08:57,959 --> 00:09:00,399 good I'm not
208 00:09:00,399 --> 00:09:03,360 completely sure I mean there's a lot of
209 00:09:03,360 --> 00:09:05,839 hype in Ai and sometimes I've gotten
210 00:09:05,839 --> 00:09:08,360 caught I've gotten caught uh especially
211 00:09:08,360 --> 00:09:11,360 with Google's AI announcement where they
212 00:09:11,360 --> 00:09:14,279 staged it there's been some some staging
213 00:09:14,279 --> 00:09:16,920 like there's been that AI what Devon
214 00:09:16,920 --> 00:09:20,720 Devon is like this AI code codot that is
215 00:09:20,720 --> 00:09:22,200 supposed to be able to be super good at
216 00:09:22,200 --> 00:09:24,560 coding and replace a whole bunch of
217 00:09:24,560 --> 00:09:28,880 developers they staged that demo Gemini
218 00:09:28,880 --> 00:09:30,680 staged there
219 00:09:30,680 --> 00:09:36,160 demo yeah I don't know it's tough that's
220 00:09:36,160 --> 00:09:38,519 tough so I don't really trust things
221 00:09:38,519 --> 00:09:40,640 until I look at them myself nowadays
222 00:09:40,640 --> 00:09:43,480 unless I can get my hands on it where
223 00:09:43,480 --> 00:09:46,680 there's been some independent reviews of
224 00:09:46,680 --> 00:09:50,240 these releases very hesitant to to say
225 00:09:50,240 --> 00:09:52,920 so but we'll see you soon because
226 00:09:52,920 --> 00:09:54,000 hopefully it's going to be released in
227 00:09:54,000 --> 00:09:55,360 the coming days they didn't give an
228 00:09:55,360 --> 00:09:57,240 exact date I don't think they said next
229 00:09:57,240 --> 00:09:59,920 week which was last week it's got to be
230 00:09:59,920 --> 00:10:01,519 sometimes this week right if I doing the
231 00:10:01,519 --> 00:10:05,839 math right the potential of it is you
232 00:10:05,839 --> 00:10:09,880 allow researchers or
233 00:10:09,880 --> 00:10:12,600 students uh at a University Research
234 00:10:12,600 --> 00:10:15,680 groups to play around with these models
235 00:10:15,680 --> 00:10:19,920 for free and it gives people an idea an
236 00:10:19,920 --> 00:10:21,360 idea playground I would say that's how
237 00:10:21,360 --> 00:10:23,560 I'd phrase it like you have the ability
238 00:10:23,560 --> 00:10:26,640 to play around with the model both
239 00:10:26,640 --> 00:10:29,600 models or all three for free and you get
240 00:10:29,600 --> 00:10:31,000 to be able to see the weights you get to
241 00:10:31,000 --> 00:10:32,640 see how it was constructed you get to
242 00:10:32,640 --> 00:10:35,760 see everything which is not something
243 00:10:35,760 --> 00:10:40,040 you get to do with closed Source models
244 00:10:40,040 --> 00:10:43,000 and I think the general consensus
245 00:10:43,000 --> 00:10:46,320 is hey open source things are better
246 00:10:46,320 --> 00:10:47,639 there's a lot more good people in the
247 00:10:47,639 --> 00:10:51,120 world than there are bad people and if
248 00:10:51,120 --> 00:10:53,720 you open source things there's a lot
249 00:10:53,720 --> 00:10:56,839 more eyes and people can see what's
250 00:10:56,839 --> 00:11:00,399 going on in real with security
251 00:11:00,399 --> 00:11:03,079 defects like security leaks or or just
252 00:11:03,079 --> 00:11:05,639 different defects of of the code base
253 00:11:05,639 --> 00:11:07,440 can be fixed a lot faster because
254 00:11:07,440 --> 00:11:09,000 there's a lot more people seeing it
255 00:11:09,000 --> 00:11:11,680 there a lot people using it and so then
256 00:11:11,680 --> 00:11:13,839 when you have things open sourced it
257 00:11:13,839 --> 00:11:16,639 just makes the maintenance of the code
258 00:11:16,639 --> 00:11:19,200 base a little bit easier because the
259 00:11:19,200 --> 00:11:23,120 public has a interest in it all right
260 00:11:23,120 --> 00:11:25,320 not really the general public but the
261 00:11:25,320 --> 00:11:27,160 coding public I
262 00:11:27,160 --> 00:11:30,440 guess the active developer group
263 00:11:30,440 --> 00:11:32,959 community so it's just overall
264 00:11:32,959 --> 00:11:36,120 better and so you know open source
265 00:11:36,120 --> 00:11:38,519 allows things to move quicker people can
266 00:11:38,519 --> 00:11:42,040 provide feedback ideas and so hopefully
267 00:11:42,040 --> 00:11:44,760 that gives it a little bit more velocity
268 00:11:44,760 --> 00:11:45,639 so we'll
269 00:11:45,639 --> 00:11:50,200 see for you can use llama the license I
270 00:11:50,200 --> 00:11:51,920 I think the license is not for
271 00:11:51,920 --> 00:11:54,760 everything except non-commercial use so
272 00:11:54,760 --> 00:11:57,279 you can't start a business for it but
273 00:11:57,279 --> 00:11:59,440 you can
274 00:11:59,440 --> 00:12:01,200 and I think that and it hasn't been
275 00:12:01,200 --> 00:12:03,240 finalized I think I think it's going to
276 00:12:03,240 --> 00:12:08,639 be more relatable when the four 108
277 00:12:08,639 --> 00:12:10,800 billion parameter model comes out
278 00:12:10,800 --> 00:12:12,959 because the first two they're great but
279 00:12:12,959 --> 00:12:14,519 they're not oh my
280 00:12:14,519 --> 00:12:17,360 goodness so once we get to that level I
281 00:12:17,360 --> 00:12:18,079 think
282 00:12:18,079 --> 00:12:20,920 they'll provide some clear guidelines I
283 00:12:20,920 --> 00:12:22,199 know that Mark was talking about like
284 00:12:22,199 --> 00:12:24,839 hey like I don't want like a large
285 00:12:24,839 --> 00:12:27,160 massive company to be using this model
286 00:12:27,160 --> 00:12:29,240 for free like if they're big like I'm
287 00:12:29,240 --> 00:12:30,800 going to make them
288 00:12:30,800 --> 00:12:33,880 pay but if like some small shop is
289 00:12:33,880 --> 00:12:35,360 trying to get started and want to use
290 00:12:35,360 --> 00:12:38,000 our model to build their product then we
291 00:12:38,000 --> 00:12:40,000 could let them do that for you know if
292 00:12:40,000 --> 00:12:41,760 their revenue is under like 10 million
293 00:12:41,760 --> 00:12:43,600 or something like that so I think it
294 00:12:43,600 --> 00:12:44,959 depends on the size of the company
295 00:12:44,959 --> 00:12:46,440 whether it's free or not I think that's
296 00:12:46,440 --> 00:12:48,480 the structure of the license but those
297 00:12:48,480 --> 00:12:50,240 things are subject to change all the
298 00:12:50,240 --> 00:12:52,079 time so you have to be
299 00:12:52,079 --> 00:12:53,800 careful
300 00:12:53,800 --> 00:12:55,680 okay
301 00:12:55,680 --> 00:12:59,519 so that demo is coming next week real
302 00:12:59,519 --> 00:13:02,680 search so with meta aai if you go to the
303 00:13:02,680 --> 00:13:07,279 meta AI website you can now search
304 00:13:07,279 --> 00:13:12,560 for reals and so I was using it at first
305 00:13:12,560 --> 00:13:14,160 I was using it
306 00:13:14,160 --> 00:13:17,600 for fashion I got to get my fashion up I
307 00:13:17,600 --> 00:13:20,040 I may have a lot of things going on for
308 00:13:20,040 --> 00:13:22,680 myself and in a posit that are positive
309 00:13:22,680 --> 00:13:25,480 one thing I don't really have going on
310 00:13:25,480 --> 00:13:28,639 is some fashion so I think my fashion
311 00:13:28,639 --> 00:13:33,800 sense and my ability to properly
312 00:13:33,800 --> 00:13:36,880 structure outfits with
313 00:13:36,880 --> 00:13:40,959 a defined stylistic
314 00:13:40,959 --> 00:13:44,279 view is poor so I'm trying to become
315 00:13:44,279 --> 00:13:46,399 more fashionable which is a difficult
316 00:13:46,399 --> 00:13:49,160 thing when you're starting from zero I
317 00:13:49,160 --> 00:13:50,839 guess because you don't really know who
318 00:13:50,839 --> 00:13:53,320 you are what what what you like so you
319 00:13:53,320 --> 00:13:54,800 had to go through this whole thing where
320 00:13:54,800 --> 00:13:56,800 you like have to look up different type
321 00:13:56,800 --> 00:13:59,360 of styles and you have to look up okay
322 00:13:59,360 --> 00:14:02,320 of your styles that you like what is the
323 00:14:02,320 --> 00:14:05,759 defining color palette of that style and
324 00:14:05,759 --> 00:14:07,240 then you have to figure out okay well if
325 00:14:07,240 --> 00:14:09,720 I like this style I like this style you
326 00:14:09,720 --> 00:14:11,160 have to just go through all these
327 00:14:11,160 --> 00:14:13,040 different styles and figure out okay if
328 00:14:13,040 --> 00:14:16,079 I combine all these different styles can
329 00:14:16,079 --> 00:14:17,839 I make my personal style because you
330 00:14:17,839 --> 00:14:20,440 don't want to just copy copy paste other
331 00:14:20,440 --> 00:14:23,440 people's stuff you want to make your own
332 00:14:23,440 --> 00:14:25,920 identity and so you could use Pinterest
333 00:14:25,920 --> 00:14:27,519 to do that or you could
334 00:14:27,519 --> 00:14:30,199 use or you could use
335 00:14:30,199 --> 00:14:33,360 use this uh Med AI where you could
336 00:14:33,360 --> 00:14:36,360 search for the style you want which is
337 00:14:36,360 --> 00:14:40,000 pretty cool I think I still like
338 00:14:40,000 --> 00:14:42,880 Pinterest more because you can save it
339 00:14:42,880 --> 00:14:44,680 but you can get just general General
340 00:14:44,680 --> 00:14:48,279 reels and stuff from this this search
341 00:14:48,279 --> 00:14:50,759 ability or you could you could ask it
342 00:14:50,759 --> 00:14:52,920 okay like what are some home decorating
343 00:14:52,920 --> 00:14:55,279 ideas and then bam then it pulls up
344 00:14:55,279 --> 00:14:57,839 reels popular reels and then you can get
345 00:14:57,839 --> 00:14:59,720 get there from there there you can you
346 00:14:59,720 --> 00:15:02,240 can find the suggested posts that are
347 00:15:02,240 --> 00:15:03,600 related to that
348 00:15:03,600 --> 00:15:06,759 post and so it might it might find you
349 00:15:06,759 --> 00:15:09,440 an area where you might want to start
350 00:15:09,440 --> 00:15:11,839 but you don't necessarily know where to
351 00:15:11,839 --> 00:15:15,360 find that first reel or that first video
352 00:15:15,360 --> 00:15:18,720 and so you could search via met AI like
353 00:15:18,720 --> 00:15:19,839 hey I want to
354 00:15:19,839 --> 00:15:22,519 see stuff relating I said like I want to
355 00:15:22,519 --> 00:15:24,800 see stuff relating to home decor and it
356 00:15:24,800 --> 00:15:27,240 shows you reals and let me see this I
357 00:15:27,240 --> 00:15:30,279 share my share this tab here
358 00:15:30,279 --> 00:15:31,880 screen this
359 00:15:31,880 --> 00:15:34,079 one
360 00:15:34,079 --> 00:15:39,639 okay so I am sharing this tab with
361 00:15:39,639 --> 00:15:42,440 everyone so basically if you're not
362 00:15:42,440 --> 00:15:45,720 watching on video I am sharing my screen
363 00:15:45,720 --> 00:15:50,279 and I'm sharing of this meta AI search
364 00:15:50,279 --> 00:15:53,680 feature so I asked meta AI um can you
365 00:15:53,680 --> 00:15:54,800 show
366 00:15:54,800 --> 00:15:57,759 me uh oh I have some major typos in here
367 00:15:57,759 --> 00:16:00,560 but it says there should be of oh I'm
368 00:16:00,560 --> 00:16:02,160 not sure there should be reals of
369 00:16:02,160 --> 00:16:04,279 creatives for products in the home space
370 00:16:04,279 --> 00:16:07,319 so creative is just like a video ad and
371 00:16:07,319 --> 00:16:10,440 I it sent me some stuff related to
372 00:16:10,440 --> 00:16:12,639 different ideas like Decor inspiration
373 00:16:12,639 --> 00:16:15,279 home home organization Smart Home Tech
374 00:16:15,279 --> 00:16:17,560 whatever and then I said please show me
375 00:16:17,560 --> 00:16:20,800 reals and so it it shows me reals of
376 00:16:20,800 --> 00:16:23,360 people you know cooking with stuff doing
377 00:16:23,360 --> 00:16:27,519 some organization things and all of
378 00:16:27,519 --> 00:16:29,040 these
379 00:16:29,040 --> 00:16:31,880 reals they might not be exactly
380 00:16:31,880 --> 00:16:35,480 ads but they are advertising a product
381 00:16:35,480 --> 00:16:38,639 with just like this like sneaky sneaky
382 00:16:38,639 --> 00:16:40,680 advertising with their
383 00:16:40,680 --> 00:16:43,079 creatives and then I said show me reals
384 00:16:43,079 --> 00:16:43,880 for
385 00:16:43,880 --> 00:16:47,480 couples and I got one for like shopping
386 00:16:47,480 --> 00:16:49,000 this was less successful I got one for
387 00:16:49,000 --> 00:16:52,360 shopping like I know this family bought
388 00:16:52,360 --> 00:16:53,079 a
389 00:16:53,079 --> 00:16:55,639 Dyson I don't know hair curler or
390 00:16:55,639 --> 00:16:58,199 something like that and the husband was
391 00:16:58,199 --> 00:16:59,399 complaining
392 00:16:59,399 --> 00:17:00,519 the wife was like it's not that
393 00:17:00,519 --> 00:17:02,759 expensive he looks it up it's like ah
394 00:17:02,759 --> 00:17:05,520 it's like 600 bucks and then she's like
395 00:17:05,520 --> 00:17:09,160 babe I got it on sale I got I got a deal
396 00:17:09,160 --> 00:17:11,240 and he's like how and she's like this is
397 00:17:11,240 --> 00:17:12,720 the website you
398 00:17:12,720 --> 00:17:15,160 know so it was a cool little story that
399 00:17:15,160 --> 00:17:19,679 they sold um I I looked at some cooking
400 00:17:19,679 --> 00:17:23,760 stuff that wasn't as successful and then
401 00:17:23,760 --> 00:17:26,679 I looked at some more cooking stuff that
402 00:17:26,679 --> 00:17:29,679 was more successful and I did as or
403 00:17:29,679 --> 00:17:31,640 couples and the as or couples are pretty
404 00:17:31,640 --> 00:17:36,039 good to where it is showing you things
405 00:17:36,039 --> 00:17:38,360 that you want to buy for your partner or
406 00:17:38,360 --> 00:17:39,919 you want to buy for yourself to like
407 00:17:39,919 --> 00:17:42,880 enhance your experience your partner and
408 00:17:42,880 --> 00:17:45,880 it it it seemed like most of the time
409 00:17:45,880 --> 00:17:51,640 that these reals were related in
410 00:17:52,120 --> 00:17:54,919 a sexual Manner and it was more related
411 00:17:54,919 --> 00:17:56,000 to
412 00:17:56,000 --> 00:17:59,159 empathy being intimate with your partner
413 00:17:59,159 --> 00:18:01,799 um so it was all about sex and and toys
414 00:18:01,799 --> 00:18:03,919 and and and different things that oh I
415 00:18:03,919 --> 00:18:05,880 didn't mean to click on that different
416 00:18:05,880 --> 00:18:07,799 things that you could be doing for your
417 00:18:07,799 --> 00:18:08,960 partner I
418 00:18:08,960 --> 00:18:11,400 guess that's what I was
419 00:18:11,400 --> 00:18:14,880 seeing and hopefully no one's dogging me
420 00:18:14,880 --> 00:18:16,559 in the comments about this like oh this
421 00:18:16,559 --> 00:18:17,760 is just your
422 00:18:17,760 --> 00:18:20,559 algorithm because it's not so I'm not
423 00:18:20,559 --> 00:18:22,440 logged in it like it just doesn't know
424 00:18:22,440 --> 00:18:23,760 I'm logged in so it doesn't know my
425 00:18:23,760 --> 00:18:26,760 algorithm unless that and that'd be
426 00:18:26,760 --> 00:18:28,600 crazy U but I don't see that kind of
427 00:18:28,600 --> 00:18:31,120 stuff on my reals on my my Instagram my
428 00:18:31,120 --> 00:18:34,919 stuff's more like gym motivation running
429 00:18:34,919 --> 00:18:37,919 and motiv like motivational quotes and
430 00:18:37,919 --> 00:18:40,679 like books and stuff so I don't get much
431 00:18:40,679 --> 00:18:44,760 of that kind of content but I do watch
432 00:18:44,760 --> 00:18:47,600 reals and make sure I keep it in the the
433 00:18:47,600 --> 00:18:50,159 fitness area because it can get it can
434 00:18:50,159 --> 00:18:52,440 get pretty bad pretty quick depending on
435 00:18:52,440 --> 00:18:55,559 what you like or
436 00:18:55,559 --> 00:19:00,520 watch but I think that when
437 00:19:00,520 --> 00:19:03,280 and the models become more advanced and
438 00:19:03,280 --> 00:19:05,120 when I keep I when I'm saying more
439 00:19:05,120 --> 00:19:07,039 advanced I'm not talking about that far
440 00:19:07,039 --> 00:19:09,480 away like maybe this this 48 billion
441 00:19:09,480 --> 00:19:11,720 parameter model when I ask it these
442 00:19:11,720 --> 00:19:15,159 things it knows exactly what I'm looking
443 00:19:15,159 --> 00:19:18,480 for scour the comments scours the the
444 00:19:18,480 --> 00:19:22,000 posts on the Instagram knows like how
445 00:19:22,000 --> 00:19:24,000 one thing's an ad the other one's not an
446 00:19:24,000 --> 00:19:26,320 ad but really that is an ad even though
447 00:19:26,320 --> 00:19:28,400 it's not a paid ad it's like it's just
448 00:19:28,400 --> 00:19:29,720 organ organic
449 00:19:29,720 --> 00:19:32,799 marketing and then it populates that
450 00:19:32,799 --> 00:19:34,559 stuff to
451 00:19:34,559 --> 00:19:39,039 me because that would save some time I
452 00:19:39,039 --> 00:19:40,240 don't think it would save time on
453 00:19:40,240 --> 00:19:42,480 product Discovery if you want to do
454 00:19:42,480 --> 00:19:44,280 organic product Discovery and just
455 00:19:44,280 --> 00:19:46,559 scroll through make a burner account and
456 00:19:46,559 --> 00:19:48,320 scroll through and find products that
457 00:19:48,320 --> 00:19:50,760 would sell to your demographic that you
458 00:19:50,760 --> 00:19:53,280 want to sell to and help build a
459 00:19:53,280 --> 00:19:57,280 brand that's an option to start out but
460 00:19:57,280 --> 00:19:59,360 then you're like okay I want to get I
461 00:19:59,360 --> 00:20:01,520 want I don't want us to like copy
462 00:20:01,520 --> 00:20:05,559 creatives from my product space I want
463 00:20:05,559 --> 00:20:07,720 to do something completely different and
464 00:20:07,720 --> 00:20:10,240 so you could ask meta hey like can you
465 00:20:10,240 --> 00:20:13,039 can you provide me creatives that are in
466 00:20:13,039 --> 00:20:14,280 this
467 00:20:14,280 --> 00:20:17,320 space that are labeled as ads or seem
468 00:20:17,320 --> 00:20:18,440 like they're
469 00:20:18,440 --> 00:20:21,360 ads in the meadow just be like yeah sure
470 00:20:21,360 --> 00:20:24,280 thing I can do that for you and then you
471 00:20:24,280 --> 00:20:26,280 can just watch all these these ads and
472 00:20:26,280 --> 00:20:28,360 say okay like I like what these people
473 00:20:28,360 --> 00:20:30,880 did like the way that they presented
474 00:20:30,880 --> 00:20:33,280 this is very
475 00:20:33,280 --> 00:20:35,120 nice
476 00:20:35,120 --> 00:20:37,960 um there's a lot of motion going on it's
477 00:20:37,960 --> 00:20:42,120 clear what what this product does for me
478 00:20:42,120 --> 00:20:45,640 it it makes me want to use it uh let's
479 00:20:45,640 --> 00:20:49,000 see this um share this tab instead you
480 00:20:49,000 --> 00:20:53,360 should see this this reel this reel is
481 00:20:53,360 --> 00:20:55,320 really not that good but it's got 60
482 00:20:55,320 --> 00:20:57,799 likes it's literally just them washing
483 00:20:57,799 --> 00:20:59,640 glass with this little thing that they
484 00:20:59,640 --> 00:21:01,640 put on the side of their sink you click
485 00:21:01,640 --> 00:21:04,200 it and it washes
486 00:21:04,200 --> 00:21:06,760 it it's pretty bad
487 00:21:06,760 --> 00:21:09,400 honestly I'm not a I'm not a creative
488 00:21:09,400 --> 00:21:10,799 expert but I could definitely tell you
489 00:21:10,799 --> 00:21:13,159 that that reel is not
490 00:21:13,159 --> 00:21:16,240 good okay share this tab
491 00:21:16,240 --> 00:21:21,080 Instead This is not an ad this is just a
492 00:21:21,279 --> 00:21:24,679 recipe let's see let's just get one
493 00:21:24,679 --> 00:21:28,360 more okay let's do this one share this
494 00:21:28,360 --> 00:21:29,960 tab
495 00:21:29,960 --> 00:21:34,360 instead okay Link in the comments so
496 00:21:34,360 --> 00:21:37,480 she's got eight likes in a comment she's
497 00:21:37,480 --> 00:21:39,960 selling
498 00:21:40,960 --> 00:21:44,240 some I guess detachable cookware where
499 00:21:44,240 --> 00:21:45,919 you could detach the handle put it on a
500 00:21:45,919 --> 00:21:49,520 large pot small pot a
501 00:21:50,080 --> 00:21:52,760 skillet
502 00:21:52,760 --> 00:21:58,799 so really odd video angle though
503 00:21:59,559 --> 00:22:05,080 but anyways so you can you can use this
504 00:22:05,080 --> 00:22:09,679 feature to save you time on discovering
505 00:22:09,679 --> 00:22:11,799 what kind of creatives you want or you
506 00:22:11,799 --> 00:22:14,799 just see like you can you can populate
507 00:22:14,799 --> 00:22:16,720 yourself with good or bad ads and and
508 00:22:16,720 --> 00:22:19,159 you can see what people are doing and
509 00:22:19,159 --> 00:22:21,400 what's selling what's not
510 00:22:21,400 --> 00:22:23,919 selling versus you having to figure that
511 00:22:23,919 --> 00:22:25,520 out where it makes it just a little bit
512 00:22:25,520 --> 00:22:26,880 easier to figure out what creatives you
513 00:22:26,880 --> 00:22:29,240 want get new ideas
514 00:22:29,240 --> 00:22:32,320 these things are forever changing so I
515 00:22:32,320 --> 00:22:33,640 know less about it because I haven't
516 00:22:33,640 --> 00:22:37,320 been doing it in a bit but yeah I think
517 00:22:37,320 --> 00:22:38,600 it's pretty
518 00:22:38,600 --> 00:22:41,080 cool and going to be useful not only
519 00:22:41,080 --> 00:22:43,120 just for product Discovery but for other
520 00:22:43,120 --> 00:22:46,279 things as well as I I mentioned like a
521 00:22:46,279 --> 00:22:48,600 personal personal example like to get my
522 00:22:48,600 --> 00:22:51,679 fashion better those things you can also
523 00:22:51,679 --> 00:22:56,120 use it for maybe May birthday ideas or
524 00:22:56,120 --> 00:22:59,279 you know places to travel those kind of
525 00:22:59,279 --> 00:23:01,559 things and you can see actual videos of
526 00:23:01,559 --> 00:23:02,799 where people are going what they're
527 00:23:02,799 --> 00:23:05,679 doing best restaurants in Tampa I did
528 00:23:05,679 --> 00:23:10,080 that best places to me in
529 00:23:10,080 --> 00:23:13,360 Florida reals and let's see what they
530 00:23:13,360 --> 00:23:15,640 say I will say that to search reals to
531 00:23:15,640 --> 00:23:17,600 query reels it does take a while like it
532 00:23:17,600 --> 00:23:20,440 might take 40 seconds to pull back
533 00:23:20,440 --> 00:23:22,440 information that you
534 00:23:22,440 --> 00:23:26,080 requested all right so I just said best
535 00:23:26,080 --> 00:23:29,080 place to eat in
536 00:23:29,080 --> 00:23:32,120 Florida it's got stuff related to Miami
537 00:23:32,120 --> 00:23:35,720 smaller places but it's just general
538 00:23:35,720 --> 00:23:37,840 General I mean you could do like where
539 00:23:37,840 --> 00:23:39,880 you live but I just did Florida cuz it's
540 00:23:39,880 --> 00:23:41,760 easier cuz then if I do where I live I
541 00:23:41,760 --> 00:23:42,799 was thinking about it I was like I'm
542 00:23:42,799 --> 00:23:44,880 going to get two involved and I'm just
543 00:23:44,880 --> 00:23:46,840 going to watch them so I could look that
544 00:23:46,840 --> 00:23:48,960 up later but it gives it gives an
545 00:23:48,960 --> 00:23:51,679 ability for people
546 00:23:51,679 --> 00:23:54,600 to find content that they're really
547 00:23:54,600 --> 00:23:55,600 searching
548 00:23:55,600 --> 00:23:58,840 for instead of scouring for it and so
549 00:23:58,840 --> 00:24:00,600 maybe they don't have to watch as much
550 00:24:00,600 --> 00:24:02,760 stuff anymore where they could just find
551 00:24:02,760 --> 00:24:05,039 exactly what they want because I don't
552 00:24:05,039 --> 00:24:09,039 need to scroll through 12 reals or 12
553 00:24:09,039 --> 00:24:11,880 Tik toks to get to what I need to all I
554 00:24:11,880 --> 00:24:14,720 need to do is find the one that I like
555 00:24:14,720 --> 00:24:17,559 so you know I'm I'm looking at this is
556 00:24:17,559 --> 00:24:20,400 is oh I didn't mean this is one two
557 00:24:20,400 --> 00:24:22,200 three
558 00:24:22,200 --> 00:24:26,720 4 5 six 7even 8 N I think 10 reals per
559 00:24:26,720 --> 00:24:29,480 query so you could look through and say
560 00:24:29,480 --> 00:24:30,640 okay like I don't want to look at that
561 00:24:30,640 --> 00:24:32,840 one I don't want to look at that one so
562 00:24:32,840 --> 00:24:34,200 I think the implications of for
563 00:24:34,200 --> 00:24:36,440 creatives is like or
564 00:24:36,440 --> 00:24:39,000 creators you need to
565 00:24:39,000 --> 00:24:41,200 have in the future you're going to need
566 00:24:41,200 --> 00:24:44,840 to have a really good
567 00:24:44,840 --> 00:24:47,279 selling frame like your your your
568 00:24:47,279 --> 00:24:50,919 starting frame for your Tik Tok or real
569 00:24:50,919 --> 00:24:54,520 needs to be good if people start using
570 00:24:54,520 --> 00:24:57,720 this feature because a lot of them I'm
571 00:24:57,720 --> 00:24:59,279 just not going to click on them I'm not
572 00:24:59,279 --> 00:25:01,200 even going to see it because doesn't
573 00:25:01,200 --> 00:25:03,720 look interesting all right
574 00:25:03,720 --> 00:25:08,880 so house update I am almost finished
575 00:25:08,880 --> 00:25:13,720 with helping remodel my it's home so we
576 00:25:13,720 --> 00:25:16,399 had some help from someone to work on
577 00:25:16,399 --> 00:25:19,520 the house and he did the electrical
578 00:25:19,520 --> 00:25:21,440 which I don't know how to do I did the
579 00:25:21,440 --> 00:25:24,200 closets which sadly aren't fully done
580 00:25:24,200 --> 00:25:25,720 yet but they're getting there they're so
581 00:25:25,720 --> 00:25:27,559 close I just need to do a little little
582 00:25:27,559 --> 00:25:29,799 tweaking to m
583 00:25:29,799 --> 00:25:32,480 and finish the little baby closets
584 00:25:32,480 --> 00:25:33,919 they're not baby like they're not for
585 00:25:33,919 --> 00:25:36,399 babies they're just smaller closets for
586 00:25:36,399 --> 00:25:39,640 the corner units just got to put up the
587 00:25:39,640 --> 00:25:42,080 the rails and stuff for hanging clothes
588 00:25:42,080 --> 00:25:44,760 and other things and shelving but the
589 00:25:44,760 --> 00:25:46,440 other closets are done CU I did the
590 00:25:46,440 --> 00:25:49,919 pantry closet hallway closet Nana's
591 00:25:49,919 --> 00:25:52,279 closet my
592 00:25:52,279 --> 00:25:55,159 closet so I did in total four closets I
593 00:25:55,159 --> 00:25:57,159 feel like that's a good amount in a week
594 00:25:57,159 --> 00:25:59,159 it takes a long time
595 00:25:59,159 --> 00:26:01,520 so the guy working on the house finished
596 00:26:01,520 --> 00:26:04,919 the high hats which makes me glow I'm
597 00:26:04,919 --> 00:26:07,399 glowing in this podcast video because
598 00:26:07,399 --> 00:26:10,760 there's so much light now he also helped
599 00:26:10,760 --> 00:26:12,720 fix my
600 00:26:12,720 --> 00:26:15,960 attic my attic didn't have as much
601 00:26:15,960 --> 00:26:18,880 instulation as it needed and then there
602 00:26:18,880 --> 00:26:23,039 were no V uh I said Vance I meant to say
603 00:26:23,039 --> 00:26:26,080 vents there were no gable vents so that
604 00:26:26,080 --> 00:26:27,600 we did have gable vents but for some
605 00:26:27,600 --> 00:26:29,760 reason someone close them and basically
606 00:26:29,760 --> 00:26:31,399 the Gable is
607 00:26:31,399 --> 00:26:35,480 when the two Slants on a house meet and
608 00:26:35,480 --> 00:26:36,880 it makes a little
609 00:26:36,880 --> 00:26:39,200 triangle there should normally be like a
610 00:26:39,200 --> 00:26:42,000 vent in that area to allow there should
611 00:26:42,000 --> 00:26:45,039 be more than that but just rule of thumb
612 00:26:45,039 --> 00:26:47,600 you should at least have two hopefully
613 00:26:47,600 --> 00:26:48,960 especially if you live in Florida to
614 00:26:48,960 --> 00:26:51,200 allow the hot air to release from your
615 00:26:51,200 --> 00:26:53,120 attic and basically you know hot air
616 00:26:53,120 --> 00:26:55,840 rises it tries to find Cold Air Cold Air
617 00:26:55,840 --> 00:26:58,919 tries to find hot air so
618 00:26:58,919 --> 00:27:00,799 the hot air rises in the Attic goes to
619 00:27:00,799 --> 00:27:03,640 the vent the vent then circulates the
620 00:27:03,640 --> 00:27:05,279 cold air in
621 00:27:05,279 --> 00:27:08,399 passively and it cools Down The House of
622 00:27:08,399 --> 00:27:10,520 the attic the attic of the house we for
623 00:27:10,520 --> 00:27:13,640 some reason had someone close our Vents
624 00:27:13,640 --> 00:27:18,720 and then cover them up so Lewis
625 00:27:18,720 --> 00:27:23,520 helped cut like like cut out the framing
626 00:27:23,520 --> 00:27:26,960 of the house cut out a hole through the
627 00:27:26,960 --> 00:27:30,360 siding like a square picture frame it
628 00:27:30,360 --> 00:27:32,640 put a vent out there waterproof it all
629 00:27:32,640 --> 00:27:37,440 that stuff which is great for me because
630 00:27:37,440 --> 00:27:38,919 the house is a lot cooler than it
631 00:27:38,919 --> 00:27:41,519 normally is so it's awesome so I'm
632 00:27:41,519 --> 00:27:44,559 loving life but we did have a scare we
633 00:27:44,559 --> 00:27:46,399 had a plumbing scare so
634 00:27:46,399 --> 00:27:50,200 basically we were replacing our trim for
635 00:27:50,200 --> 00:27:52,559 the house and so the
636 00:27:52,559 --> 00:27:55,039 trim is like I guess like the shower
637 00:27:55,039 --> 00:27:58,279 trim or the shower set is basically the
638 00:27:58,279 --> 00:28:00,159 valve the shower valve that's inside the
639 00:28:00,159 --> 00:28:03,000 wall behind your toilet not your toilet
640 00:28:03,000 --> 00:28:05,640 behind your shower handle and then your
641 00:28:05,640 --> 00:28:07,360 shower arm which is the thing that
642 00:28:07,360 --> 00:28:09,240 connects to your shower head and then
643 00:28:09,240 --> 00:28:11,799 your shower head and then your drain
644 00:28:11,799 --> 00:28:14,640 overflow valve if you have a shower and
645 00:28:14,640 --> 00:28:16,960 tub and then your
646 00:28:16,960 --> 00:28:20,799 drain if you have a shower and tub so my
647 00:28:20,799 --> 00:28:23,880 bathroom is a shower and tub so I got
648 00:28:23,880 --> 00:28:27,039 this drain that looked great it matched
649 00:28:27,039 --> 00:28:29,320 the set that I I
650 00:28:29,320 --> 00:28:30,840 had
651 00:28:30,840 --> 00:28:33,519 and Lewis took out the drain but then
652 00:28:33,519 --> 00:28:36,840 the original drain broke upon removal
653 00:28:36,840 --> 00:28:38,760 which is pretty common for older drains
654 00:28:38,760 --> 00:28:40,200 like my drain has probably been in there
655 00:28:40,200 --> 00:28:43,159 for 50 years or something I'm not sure
656 00:28:43,159 --> 00:28:45,200 if it had been remod before maybe maybe
657 00:28:45,200 --> 00:28:46,760 40 years I don't know it's been in there
658 00:28:46,760 --> 00:28:52,480 for a long time and the drain broke and
659 00:28:52,480 --> 00:28:55,600 my drain didn't fit other drains didn't
660 00:28:55,600 --> 00:28:58,799 fit and so I was in this position to
661 00:28:58,799 --> 00:29:01,880 where if I couldn't find a drain I
662 00:29:01,880 --> 00:29:04,559 couldn't use my shower and I would have
663 00:29:04,559 --> 00:29:06,440 to pay
664 00:29:06,440 --> 00:29:09,000 for the person working on the house
665 00:29:09,000 --> 00:29:13,120 either to chip underneath the the tub to
666 00:29:13,120 --> 00:29:15,440 disconnect the pipe to drain pipe to
667 00:29:15,440 --> 00:29:19,080 connect a new pipe or have him rip out
668 00:29:19,080 --> 00:29:21,919 everything and basically redo it both of
669 00:29:21,919 --> 00:29:24,679 those things were out of scope for the
670 00:29:24,679 --> 00:29:28,360 project so was in like dire need to find
671 00:29:28,360 --> 00:29:30,120 a drain and
672 00:29:30,120 --> 00:29:33,559 luckily I found a drain that fit my
673 00:29:33,559 --> 00:29:36,240 dimensions for my drain which was
674 00:29:36,240 --> 00:29:38,440 awesome
675 00:29:38,440 --> 00:29:41,240 because it saved a lot of money that I
676 00:29:41,240 --> 00:29:42,760 didn't want to spend and it was it was
677 00:29:42,760 --> 00:29:45,399 really cool because I was able to just
678 00:29:45,399 --> 00:29:48,159 put it in and just boom drop it right in
679 00:29:48,159 --> 00:29:49,559 there I mean really not but like you
680 00:29:49,559 --> 00:29:52,080 have to do PL plumber's putty put it in
681 00:29:52,080 --> 00:29:53,279 there tighten it
682 00:29:53,279 --> 00:29:56,399 down all good but if I couldn't have
683 00:29:56,399 --> 00:29:58,240 found the drain cuz our the dra drain
684 00:29:58,240 --> 00:30:00,679 was so old like the drain type I was
685 00:30:00,679 --> 00:30:03,039 having a hard time finding the the right
686 00:30:03,039 --> 00:30:05,440 not the right length but the right
687 00:30:05,440 --> 00:30:07,240 thread width that the threads wouldn't
688 00:30:07,240 --> 00:30:09,600 fit and so a lot of times everything was
689 00:30:09,600 --> 00:30:12,399 too big and if it's too small you could
690 00:30:12,399 --> 00:30:14,200 put on an attachment and basically
691 00:30:14,200 --> 00:30:16,000 rethread it called like a bushing I
692 00:30:16,000 --> 00:30:19,960 think but for my case mine were too like
693 00:30:19,960 --> 00:30:21,840 the new drains were too big so I
694 00:30:21,840 --> 00:30:24,320 couldn't really make it smaller so I
695 00:30:24,320 --> 00:30:28,000 eventually found one it worked out and
696 00:30:28,000 --> 00:30:31,440 everything is a happy happy
697 00:30:31,440 --> 00:30:35,840 story and so I'm excited you're excited
698 00:30:35,840 --> 00:30:38,120 I know you're thrilled to be hearing
699 00:30:38,120 --> 00:30:39,519 about that you're like yes you're
700 00:30:39,519 --> 00:30:42,120 jumping for joy like yeah that's awesome
701 00:30:42,120 --> 00:30:45,120 next week we're going to be
702 00:30:45,120 --> 00:30:48,120 discussing the 48 billion parameter
703 00:30:48,120 --> 00:30:52,159 model hopefully we'll have a live demo
704 00:30:52,159 --> 00:30:55,519 of the episode not the episode live demo
705 00:30:55,519 --> 00:30:57,240 of the model during the episode which
706 00:30:57,240 --> 00:30:58,679 would be great
707 00:30:58,679 --> 00:31:02,240 and of course before we end it
708 00:31:02,240 --> 00:31:06,320 out have a great day a good morning good
709 00:31:06,320 --> 00:31:09,440 afternoon wherever you are in this world
710 00:31:09,440 --> 00:31:11,639 and I hope that you're listening in next
711 00:31:11,639 --> 00:31:13,960 week have a great day thank you for
712 00:31:13,960 --> 00:31:16,480 listening
SourcesFollow the source trail.
E026 Sources
Preserved episode evidence
[[E26 - Transcript - Google Drive recovered]] is immutable but appears to be a production outline rather than a verbatim recording transcript. It includes planned segments about an anticipated Llama release, Reel search, and smart-home use.
[[E26 - Meta's AI Power Play - Open Models and Reel Search]] is the retained legacy article and provides the fuller source-era editorial artifact. It cannot reconstruct exact spoken words.
[[E26 - Transcript - YouTube auto captions]] is the exact English auto-generated caption file recovered from the public YouTube recording. Its SHA-256 is 71CF3DDAE025252DDB498C6403E0611B405B8B1EDAD374677C25A7B541667026.
[[E26 - Transcript - YouTube auto captions]] preserves the complete timestamped SRT body without correction. Its SHA-256 is 0F2A7F26174475E7853A55ACC6D9F779EF7E92A0FC7FAB2136BE9D1C8F9CDE01.
The captions are the final recording evidence for paraphrase. Exact quotations require an audio check. The recording covers pre-release model expectations, live Reel-search use, and a conventional home-remodeling update. It does not contain an AI smart-home implementation.
Existing public identity
daltonanderson.ghost.io/metas-ai-power-play-llama-3-smart-reel-search
open.spotify.com/episode/5xmE0hYheRvBOoqaQCyUok
Release and risk evidence
ai.meta.com/blog/meta-llama-3-1
ai.meta.com/research/publications/the-llama-3-herd-of-models
github.com/meta-llama/llama-models/blob/main/models/llama3_1/MODEL_CARD.md
github.com/meta-llama/llama-models
github.com/meta-llama/llama-models/blob/main/models/llama3_1/LICENSE
github.com/meta-llama/llama-models/blob/main/models/llama3_1/USE_POLICY.md
These sources establish the later Llama 3.1 release and provide a basis for comparing anticipation with the actual artifact.
csrc.nist.gov/pubs/ir/8425/final
nist.gov/blogs/taking-measure/7-tips-keep-your-smart-home-safer-and-more-private-nist-cybersecurity
NIST provides product-level consumer IoT security and privacy outcomes. It does not approve a particular home configuration.
ETSI EN 303 645 provides an outcome-focused consumer IoT security baseline. It does not define household consent or authorize an assistant integration.
Discovery-system and evaluation evidence
tensorflow.org/recommenders/examples/basic_retrieval
tensorflow.org/recommenders/api_docs/python/tfrs/tasks/Retrieval
tensorflow.org/resources/recommendation-systems
TensorFlow Recommenders supports the distinction between candidate retrieval, ranking, and possible post-ranking stages. It does not establish Meta's implementation.
elastic.co/docs/reference/elasticsearch/rest-apis/search-rank-eval
Elastic documents representative queries, relevance judgments, and information-retrieval metrics. It is an implementation reference, not independent proof that one retrieval product is useful.
AI evaluation evidence
nist.gov/itl/ai-risk-management-framework
NIST supports explicit testing, evaluation, verification, validation, and risk-management records. It does not verify a particular release claim.
Editorial boundary
E026's Episode Story may use the recovered machine captions with an audio-review gate for exact quotations. Evergreen work may use the recording as an idea source and direct observation record, the outline as a planning artifact, the legacy article as source-era commentary, and primary sources for model releases, discovery systems, evaluation, and consumer IoT controls. Do not claim that Meta's underlying Reel-search architecture was observed. Do not claim that E026 demonstrated an AI smart home.