YT YouTube2Text

YouTube2Text → Transcripts → He Sells AI Second Brains to Businesses (For $1000s)

He Sells AI Second Brains to Businesses (For $1000s) — Transcript

by Ryan Doser · 6,034 words · 932 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00Welcome back to another edition of AI
0:02rabbit holes. This is a video series
0:04where we go down the endless amount of
0:06rabbit holes in artificial intelligence.
0:08It's then happy to have back on Adam
0:10Sandler. Now Adam is the founder of the
0:12viable edge where he builds AI powered
0:15second brains and knowledge systems for
0:18founders, agencies and marketing
0:19leaders. So Adam, it's good to have you
0:21back once again as you were on the
0:23podcast actually once before.
0:25>> Right. Yep, Ryan. I really appreciate
0:27you having me back. Congrats on
0:29everything you've been doing.
0:31Uh love all your stuff.
0:33>> Same to you as well, bro. So today we're
0:35going to be walking through your AI
0:36second brain setup, right? Second
0:38brain's kind of that trendy term we see
0:40everywhere now. Um but Adam, it's been a
0:42little bit since our first conversation.
0:44I first want to know kind of what's
0:46changed in what you're building with AI
0:48since we last talked.
0:50>> Yeah, it's a great question and you
0:53know, the second brain slash knowledge
0:55base has sort of like permeated a lot of
0:58my world because
1:01in the client engagements that I've been
1:03working on over time and I'm sure you
1:05have seen this in a lot of people
1:07viewing this have encountered this where
1:10you're working with a client and you
1:11want to build a workflow or there's some
1:13kind of project, but frequently I would
1:16run into a issue where the information
1:19that was needed to power this workflow
1:21or whatever it was was not easily
1:25accessible, right? And so there's always
1:28this bottleneck that I would encounter
1:30and
1:31uh you know, eventually it occurred to
1:34me that this should just be the first
1:36step, right? As part of an offering. And
1:38what that first step ends up being, if
1:41you want to do it correctly, is to
1:43develop a knowledge base or a second
1:44brain. Um right? I think these terms are
1:47somewhat interchangeable.
1:49Um and there's so many benefits to doing
1:52this uh both as a practitioner and from
1:55a client standpoint, there's a ton of
1:57value in it. And essentially, what this
1:59is is just, you know, collecting and
2:01aggregating and structuring relevant
2:04information um that is durable and then
2:07can be uh scaled in certain ways.
2:12>> I I'm pretty sure you hear stuff like
2:13this all the time, or where you talk to
2:14a prospect or someone and they're like,
2:16"Why I just want an AI agent, or I want
2:18this entire system to do everything for
2:20me, like a fractional CMO and da da da."
2:23But, it's like that is completely
2:24useless without some sort of second
2:27brain or context like you're describing
2:29behind the scenes. So, this needs to be
2:31step 1A or step 1B before we even move
2:34into any of this actual workflow
2:36building stuff. And people need to
2:37understand this.
2:39>> Yeah, that's correct. And, you know,
2:41there's there's a lot of um intrinsic
2:44benefits that I think uh clients,
2:46especially, are not necessarily thinking
2:49about. And the one of the biggest, in my
2:51mind, is that if you hitch your wagon to
2:55a knowledge base and not to Anthropic's
2:58ecosystem or OpenAI's ecosystem, you're
3:00not stuck there if suddenly that service
3:03becomes, for one reason or another,
3:06untenable for your business.
3:08Right? So, at the very least, it sort of
3:10unshackles you from one provider's
3:13ecosystem and allows you to travel to
3:16different ecosystems. It makes you
3:18portable to a degree, right? So, I think
3:20there's um a degree of security there,
3:23longevity, um and uh a way to
3:28uh just shore up your future when it
3:31comes to working with AI from a right
3:34>> And I think something a lot of
3:35businesses don't realize is they're
3:36sitting on a lot more context than they
3:38really understand, right? We're talking
3:40about all sorts of meeting transcripts
3:42that can scan back from Hopefully,
3:44you're they're using like a
3:44transcription service or taking meeting
3:46notes where they can just have that, or
3:48maybe even email chains or call
3:51transcripts if you're doing anything
3:52sales related, right? There is so much
3:54context that businesses are just sitting
3:56on that they just do not know how to
3:58leverage it. And so using a second brain
4:00is something that really any business or
4:02individual can come in and implement and
4:04then use that to improve their AI
4:06outputs moving forward. That's what I
4:08tell a lot of people.
4:09>> Yep, absolutely. And, you know, one
4:11thing that I think is a little bit of a
4:13a watch out in a trap is that
4:16you can have this huge body of
4:18information, this historical, you know,
4:21uh however you're storing it internally
4:23in your business,
4:24history of transcripts.
4:26A lot of that stuff is likely not
4:28relevant right now for what's happening
4:30in your business. So, part of the
4:32knowledge base or second brain is
4:35creating a flywheel-like system that can
4:38manage the information for you to keep
4:41it relevant, right? How is information
4:44being ingested? How is it being
4:46synthesized and processed? And how is
4:48older information or stale information
4:51being archived or pulled out of what's
4:54relevant right now, right? So, you know,
4:56there are a lot of dynamics to it, which
4:58I think is not something a lot of people
5:01are, you know, going to think about
5:03right away. It's not obvious.
5:05>> I love how you said flywheel system,
5:07right? So, in my case, this is why I try
5:09to create weekly content. It's not just
5:11so I can keep posting things out to
5:13YouTube and all the other platforms.
5:14There's a whole other strategy behind
5:16this, Adam. It's the context strategy,
5:18right? Where do you think I'm able to
5:19get all these social media posts and all
5:22this content on ryandozer.com? Well, a
5:24lot of that is taking all of this
5:26context from my past content on podcast
5:28episodes or maybe even podcast guest
5:30appearances that I show up on. So,
5:32there's a whole second strategy to just
5:34having a flywheel-type content system
5:37that's applicable to most businesses out
5:39there.
5:39>> Yeah, exactly. And there's a lot of
5:42things that this type of mechanism
5:45brings to a business that I don't think
5:47are necessarily standard or formal
5:51practices. For example, decision
5:53tracking, right? So, within a knowledge
5:56base, there's an opportunity to create a
5:58decision log. So, all of your high-level
6:01or high-impact decisions are captured
6:04and documented for reference, for
6:06history, for consistency, right? Uh in
6:09the a lot of the places that I've worked
6:11that I've, you know, I've been
6:13like been in the workforce for over two
6:15decades.
6:16Uh I've never really seen a business
6:18practicing the practicing that sort of
6:20thing with that with the level of
6:22specificity that you can enable with a
6:25knowledge base that's powered by an AI,
6:27right? So, you sort of you get all of
6:29the benefit that we talked about, plus
6:31you get this additional layer of
6:34detailed tracking and history, and you
6:37know, there's
6:39a much easier way for businesses to sort
6:41of reflect on past decisions, why those
6:44decisions were made, and understand how
6:47past decisions might apply to new
6:48business, right? So, there's that whole
6:50element of learning and progression that
6:53I think is somewhat unique.
6:56>> For sure. And you mentioned before that
6:57your context system runs on markdown
6:59files, right? I don't know if this might
7:01be a good opportunity to screen share
7:02and kind of show your second brain
7:04system, but while you're doing that, we
7:06hear these terms a lot like Obsidian and
7:09rag and vector databases and all these
7:11fancy terms, right? How can you like
7:14simplify this for the audience here? Do
7:16we need to even go down that path?
7:18>> The short answer is no,
7:20because markdown and for anyone who's
7:22watching who's not aware, it's just a
7:24another flavor of text file that has
7:27different formatting and works really
7:29well with AI, right?
7:31Um and
7:32you know, when you
7:34a knowledge base like at its most basic
7:36form is a folder of documents, right?
7:39But we need to put mechanisms in place
7:42for those documents to make sense in the
7:44context of each other, right? And that's
7:46where where some of the AI comes in and
7:48some of the systems and tooling comes
7:50in, right? But, you only really need
7:53markdown files because a lot of the
7:55businesses that a lot of small
7:56businesses simply don't have the amount
7:59of information that would necessarily
8:02require a sophisticated rag or vector
8:04database, right? So, I hope I'm
8:06answering your question, but like you
8:08don't really have to worry about the
8:10tech stack. Personally, I use Obsidian
8:12for my personal knowledge base because
8:15it's just it makes sense to me as an
8:18organizational system, right?
8:20Um and I followed the Karpathy method of
8:23wiki linking everything, which is really
8:25cool. And it you know, there's a huge
8:27benefit to that. Um but you don't need
8:29Obsidian, right? Uh Claude or ChatGPT is
8:33smart enough to enable that system on
8:36its own just by working with it.
8:39Um
8:39>> I'll tell you my vault quick. I'm not
8:41going to screen share my stuff, but I
8:43have it set up where every single time
8:45I'm in a conversation with Claude code,
8:47it automatically saves it into a nodes
8:50vault internally that I back up to
8:51Google Drive on a daily basis. And so,
8:54it's just this mountain of vault of all
8:56these markdown files that not only
8:58document like the key points from that
9:00conversation, but there is some wiki
9:02stuff going on and going back and forth,
9:04connecting to the Claude MD and agents
9:06MD and all that fun stuff. That's kind
9:08of like my rundown of my personal
9:11context second brain right now. I'm
9:13always tinkering around with different
9:14things and what I should do to improve
9:16on that.
9:17>> Yeah, yeah. And And one thing one
9:18example I really like that helps um
9:20demonstrate the value is the transcript
9:23thing that you mentioned, right? Like
9:25we're
9:26I'm sure you're on as many calls as I am
9:28or more per week, right? Collecting
9:30transcripts, you know,
9:32uh probably hundreds a month at this
9:34point or something like that, right? And
9:36when you want to reference a phone call,
9:39the AI, you don't necessarily want it to
9:41go and search through transcripts and
9:43read transcripts and fill up context,
9:44right? So, a knowledge base, when it's
9:47well maintained, can have a summary of
9:49the transcript on top where and then
9:52where detail is accessible if needed,
9:55right? And what's implied here is that
9:58the AI knows how to navigate your
10:01information, right? And the transcript's
10:03a perfect example. Because if I have
10:06what what we call hot context, which is
10:08relevant context right now, might be
10:10summaries of calls that I've had over
10:11the last 2 weeks, right? And if I need
10:14to understand a specific detail, I don't
10:17need to tell the AI where to find it
10:19because the knowledge base, when it's
10:21set up and has the linking system uh
10:24implemented, it can just navigate its
10:27way properly, right? Um and in practice,
10:31it kind of feels like magic. Right? In
10:33practice, you know, when I begin my day
10:35and I start working with Claude or Her-
10:37I'm using Hermes a lot these days. Um
10:41I don't have to explain anything cuz
10:42every system I use is hooked up to my
10:45knowledge base.
10:46Um
10:47>> the benefit of a knowledge base, like
10:48you talked about before, right? It's
10:50transferable across any AI system versus
10:52you get locked into ChatGPT or
10:54Anthropic, like that's no man's land at
10:57some point if you ever want to divulge
10:59from something else. But Adam, I think
11:00this is a good time. Why don't we go
11:02ahead and screen share kind of your
11:03system so maybe people can just get a
11:06better understanding of how you're
11:07thinking about second brains when you're
11:09doing all this.
11:10>> Yeah, so my background is in marketing
11:13and so um a lot of my clients and
11:16business engagements are marketing
11:18flavored. Um and so based on a lot of
11:21the work that I've done over the past
11:23year and change, I developed a
11:26online-based knowledge base for brands.
11:29But the
11:30the takeaway here is just sort of how it
11:32works, right? So,
11:35when you onboard a brand or when you
11:37there's always like an ingestion period,
11:39right? Knowledge has to get into the
11:41knowledge base somehow. Um and there's a
11:43lot of different methods for doing that.
11:45The method that I use for this is
11:47through a chat, an onboarding chat um
11:51that will scan your website. It will
11:54uh look at competitors. It will take any
11:56information that you want to give it.
11:57This This is already set up. So, this is
12:00what it looks like after you go through
12:01that. And on the right here, we have a
12:04visual knowledge graph. And what this
12:07does is
12:09give you a visual of the different
12:10relationships between the different
12:12pieces of knowledge in my knowledge
12:14base, right? So, these red guys are
12:17competitors. So, if I click Jasper,
12:19I can see all of the links where Jasper
12:22might be relevant to other pieces of my
12:24knowledge base. And my site just gives
12:26you some [snorts] specific information
12:29uh on how it relates to you from a
12:31competitive standpoint. But again, the
12:33takeaway here is
12:34>> for the audience, quick out of This is
12:36your own Vibe Coded app, correct?
12:38>> Correct. Mhm. Yep. This is on
12:41viableedge.com.
12:42Um it's a brand knowledge base for brand
12:45marketers who struggle with copying and
12:48pasting the same thing into their chats
12:51all the time and get inconsistent
12:53results when doing things like content
12:55creation.
12:56Um and one of the great things about
12:59this is that it's easily outputable,
13:02right? So, this all lives in the cloud
13:05and
13:06you know, there's a various methods
13:09which this information can be pulled out
13:11such as an MCP server, Claude
13:14instructions, an Obsidian vault even,
13:16even custom skills that could be loaded
13:18right into any chat as well as a Claude
13:21plugin, right? So,
13:23all of these different output methods
13:26ultimately um link back to the same body
13:30of information.
13:31And that information, I'll just
13:34my knowledge base is represented here in
13:36this list.
13:37So, if I pull up, for example, my voice
13:39and tone guide,
13:41here it is.
13:42And when I hook up my MCP to an external
13:45chat tool,
13:47it will just have this access.
13:49Right? So, if I were to open a a uh
13:51incognito chat uh hooked up to the
13:54server, and I could do that. I don't
13:56know if we want to spend the time on
13:57that. I don't have to give it any
13:59prompting at all. And I could also give
14:01it to um humans. I can give it to a
14:03freelancer, agency partner, and just
14:07like a knowledge base should, there is
14:10there are autonomous agents behind the
14:12scenes that are checking for things like
14:14brand drift. They're pulsing competitive
14:17updates and capturing um things that it
14:20might think are interesting or relevant
14:22for your brand, making recommendations.
14:24So, we talked about the flywheel. That
14:26flywheel is very much in effect here.
14:29You know, uh
14:30>> That's awesome. Yeah, you got the
14:31autonomous stuff going, too. So, anyone
14:33watching or listening to this that might
14:34be thinking to themselves, Adam, how do
14:36I make money from something like this,
14:38right? Is this something that someone
14:39could implement for businesses like you
14:41are? Or I guess what's kind of the the
14:43next game plan there?
14:45>> So, the the opportunity here, and I
14:48think this is one of the biggest
14:49opportunities in AI right now,
14:51especially for practitioners who are
14:53just getting going or trying to figure
14:55out what to do, um
14:57y- a lot of small businesses can benefit
14:59hugely from this knowledge base. So, if
15:01you know how to position it properly,
15:04you can really get a great foot in the
15:05door and set yourself up for an ongoing
15:09relationship, because once you have a
15:10knowledge base, building agents or tools
15:13on top of it,
15:15I don't want to say becomes easy, but it
15:16becomes much more straightforward.
15:18Right? So, you're setting yourself up
15:20for more business. So, when you engage a
15:23client or you find a small business that
15:25you want to work with,
15:27you know, a lot of businesses that I've
15:29worked with and and let me know if if
15:30this resonates. Like,
15:32most companies they're they're a mess.
15:33They're a chaotic mess. Their stuff's
15:35all over the
15:35>> and meager they are.
15:36>> Right? So, you can approach a situation
15:39like this or a prospect and say, "Not
15:41only am I going to get your knowledge in
15:43order, I'm going to clean up your
15:45information. I'm going to structure it
15:47for you. We're going to make you ready
15:50for AI. We're going to future-proof your
15:51business for AI readiness
15:54and you're not stuck with one provider."
15:57>> Yeah.
15:57>> Right? So, how do you scale this, right?
15:58Is this something where you just you use
16:00a Superbase and then you just kind of
16:01rinse and repeat different templates?
16:03>> Yeah, yeah. So, we talked about how you
16:05can just get started with Markdown.
16:07But, at a certain point, if you want to
16:09scale, Superbase or something like it is
16:12a great next step. Because I like
16:13Superbase because it's inexpensive.
16:16Obviously, it integrates really well
16:18with a ton of different platforms out
16:19there, good customer service,
16:22um and I've just had a great experience
16:25with it. There's other tools out there
16:26like Convex, which I really want to
16:28explore. Um
16:31so, with Markdown, when you're just for
16:33guess when you're first getting started
16:34with Markdown, it's very e- Well, again,
16:37I don't I hate using the word easy. It's
16:39straightforward to then migrate and step
16:41up to something like Superbase, right?
16:43So, with Markdown, the path to
16:45scalability is right in front of you
16:47already. So, that's why if you're not a
16:51a technical person or if you're just
16:53learning,
16:54you don't even need a stack to provide
16:57major value with something like a
16:59knowledge base.
17:00>> Here's another question I have, and I've
17:02often struggled with this part of it.
17:03So, I use Calendly for all my meetings,
17:06you know, call recordings, transcripts,
17:08etc. But, one thing Calendly still
17:10struggles with, they even have API and
17:12MCP, is they struggle with connecting to
17:15systems like Claude Code, where you can
17:16pull transcripts. Yeah, so you probably
17:18need been through this before, right?
17:20>> I use Calendly.
17:22So I use Calendly also.
17:24>> For people listening to this, I've done
17:25this before.
17:26Yeah.
17:27>> Sorry, I didn't mean to talk over you.
17:29Um
17:29it's funny cuz like Calendly, I set up a
17:31skill in Claude code
17:34that it um will go and use the browser
17:37control to pull down text files of all
17:41of the transcripts since the last time
17:43it captured transcripts. So basically
17:45like once a day I'll run this skill. It
17:47pulls all the transcripts down and has
17:49to use the browser control because
17:51there's no integration to pull
17:52transcripts from Calendly, which is a
17:54major frustration.
17:56Um and so like I think this is a
17:58stopgap. You know, maybe there's a if I
18:00have time, I might hunt for other
18:02solutions. Maybe Calendly will add that
18:04integration soon, who knows. But there,
18:07you know, sometimes you have to be
18:08creative and and work around certain
18:10things. And you know, with uh as AI gets
18:14better and if you've used um
18:17you know, 5.6 Sol's computer and browser
18:20use, it is phenomenal.
18:22So there's even some people out there
18:24who prefer browser use through chat GPT
18:29or computer use versus API because it's
18:32that fast and it's that good. Um so, you
18:36know, these are technically shortcuts,
18:38but you know, these are things that
18:40you're going to work around. Inevitably,
18:42there's going to be something that you
18:44have to work around.
18:45>> Right.
18:46>> And so getting a little bit scrappy,
18:48rolling up your sleeves and you know, uh
18:50having an arsenal to implement things
18:53like browser control and develop
18:55specialized skills
18:57um
18:58is definitely a skill that is going to
19:00help a ton for
19:03kicking up this kind of business.
19:05>> Yeah, so staying on this same topic of,
19:07you know, making money off of this AI,
19:09you know, second brain business, I guess
19:11what else should someone know? Are there
19:12certain types of industries that work
19:14better than others? How do you monetize
19:16this? Is it reoccurring? Is it a
19:17one-time AI second brain build? Like,
19:20talk about that a little bit.
19:21>> Yeah, so there's, you know, an when
19:25these engagements are beginning, there's
19:26a huge opportunity for an audit. And
19:30when when I mentioned how you can sort
19:32of clean up a company's knowledge for
19:35them and that's sort of like an added
19:36value, the audit can be focused on that,
19:39right? When you're um
19:41specking out a knowledge base, you do
19:44want to make sure that you are organized
19:46in terms of the kinds of schema that
19:48you're documenting, right? Uh because
19:50your client's not going to know that.
19:52And there are best practices and I do
19:55have some material that lays out
19:58um a baseline set of schema. There's
20:01seven different data points, basically.
20:04And this is usually enough to get going.
20:08One thing that I like to
20:10uh
20:11focus on with my clients is what is sort
20:13of like their core spine, as they say,
20:17right? This What is the one thing that
20:19everything else in the business sort of
20:21like ladders up to? And usually that is
20:23a goal or an objective, right? Something
20:26that is very high-level and impactful
20:27for the company.
20:29So, when you're kicking off this
20:30conversation, what you can say is like,
20:32"Okay, let's talk about your spine,"
20:34right? I wouldn't know if I would use
20:35that word, but, you know, how is what is
20:37the driving force of this company?
20:39Right?
20:40And that can be a conversation that uh
20:44generates an audit.
20:46And from this audit, you'll have a very
20:49clear scope of what the knowledge base
20:51looks like because the audit will have
20:53covered what's this company's knowledge,
20:56where is it, what's the spine, here are
20:58the recommended data points that go
21:00around it. So, you can use the
21:02opportunity for an audit to like do the
21:04map.
21:05And through the audit, you'll also
21:07surface a lot of disconnects, right?
21:10Because you have to be ready because
21:12knowledge is it's going to be messy.
21:13You're you're jumping into a messy pool
21:15and you got to clean it all up.
21:17>> Yeah.
21:17>> Um and the great thing is like it gives
21:19you an opportunity to say, "Hey client,
21:22did you know this and that?" Right? Like
21:24you might surface insights, you might
21:26surface connections that were not
21:29obvious, and you might surface things
21:31that are on fire, right? You might
21:33surface things that are outdated. Um
21:36>> [clears throat]
21:36>> with almost every client that I work
21:38with, it's like where the give me your
21:40brand guidelines. Oh, here's a brand
21:41guideline from 2020, 5 years ago. Here's
21:43another one from right? And so like
21:46the audit can sort of like help just
21:49scrub out all of the old stuff and have
21:51fresh stuff, right?
21:52>> Speaking of brand audit, you know, one
21:53thing I've actually found works better
21:55than most companies brand audits from 10
21:57years ago is you just either Firecrawl
21:59or Webfetch just scrape their website
22:02and just say, "Create me a detailed
22:03brand guide." And it's usually 10 times
22:05better than what they had 10 years ago.
22:07>> Yeah, and for for my clients, too, what
22:09I like to do now is I'll use Claude
22:11design and pull a style guide for them.
22:14And then I'll use that style guide to
22:17um
22:18you know, style and design all of the
22:20collateral for that relationship.
22:23>> Yeah.
22:23>> Um and when you get a little bit more
22:25experienced, you could build a UI for a
22:27knowledge base.
22:29And the and I don't want to like I I I
22:31hope I had no we're veering away from
22:33the question you asked a little bit. But
22:35with um Claude code in particular, and I
22:38I tend to work with a lot of clients who
22:40use Claude. Um and they just released a
22:44browser in Claude code desktop.
22:47And it's a great opportunity because if
22:50you are a practitioner and you're
22:51delivering workflows, including
22:53something like a knowledge base,
22:55there's an abstraction for people who
22:57are not as experienced with AI, right?
22:59You put a workflow go front of them and
23:00you you know, you put a chat box and
23:02you're just like, "Type {slash} this."
23:04Right? A lot of people are going to be
23:05confused because it's just not how
23:06people are used to working.
23:08>> Yeah.
23:08>> However, you can create a UI fairly
23:11quickly these days, right? And you can
23:13create a nice one with cloud design and
23:15one that's branded for your client. And
23:17think about how you might be able to
23:19make that into a workspace for a
23:21workflow.
23:23So, you can actually create a visual
23:25layer for the knowledge base.
23:27Effectively, it could be a markdown file
23:29browser.
23:30It could be a way for the client
23:32themselves to click through their
23:34documents, similar to how I showed the
23:36knowledge graph.
23:37There's a lot of free and open source
23:40packages out there that have knowledge
23:42graphs. So, you can just pull something
23:43down and implement it yourself into your
23:45own app, right? There's just so many
23:47ways that you can add value and leverage
23:50existing resources out there to really,
23:54um, you know, step up your delivery on
23:56this sort of thing.
23:57>> For sure. And the main takeaway with all
23:59of this is that most companies do not
24:02have a organized second brain, and
24:04that's what's going to differentiate
24:06people or organizations that the amount
24:08of context that not only is quality and
24:10you have a flywheel system, but you have
24:12an organized second brain behind that.
24:14And I want to transition now, Adam, into
24:16some rapid-fire AI questions.
24:18>> Yes. Let me just Can I just add one more
24:20thought to what you just said?
24:21>> Yeah.
24:21>> Um, because like we talk we're talking
24:23about the second brain and knowledge
24:25base like, uh, on the company scale, but
24:28I'm working with clients now where we're
24:29doing department-level knowledge bases,
24:32right? And then there's sort of like a
24:34hierarchy and these knowledge bases can
24:36also have integrations with each other,
24:38right? So, I just wanted to throw out
24:40there that, you know, you don't have to
24:41think about it on a company scale. It
24:43could be a department scale, right? It
24:45could be an individual scale or a team
24:46scale, right? So, for those who are
24:49looking for business, right? You don't
24:50have to, you know, bite off the entire
24:52piece of of the company right away. You
24:55can come in with a smaller engagement to
24:58get your feet wet, to develop trust,
25:00um and something that could be a little
25:01bit more like manageable. So, I just
25:03wanted to throw that out there.
25:04>> Yeah, no, that's a really good point cuz
25:06like an SEO second brain is different
25:08than an accounting second brain.
25:09Although they could be intertwined
25:10somehow, but yeah, there's a big
25:12distinction there.
25:13>> And the magic for clients is when you
25:15connect different sources of information
25:17and discover new insights. That's where
25:19that the magic really comes.
25:21>> For sure. So, I want to just get into
25:23some quick rapid-fire questions here to
25:24conclude this. So, we heard the term
25:26prompt engineering, right? And now we're
25:28kind of moving towards loop engineering,
25:30but I've also heard the term context
25:32engineering, right? Is that a skill that
25:34actually matters?
25:36>> Oh, 100%.
25:37100%. Um and I think it's everything we
25:40talked about, right? I think a lot of it
25:42is managing the flywheel, right? It's an
25:44organism. It grows, right?
25:47Um and I think a lot of the uh skill it
25:51comes in in
25:53that piece of
25:55>> Next question I have for you. What is
25:56the one type of context that most people
25:58forget to include or companies?
26:01>> It's kind of like the opposite problem
26:02where like they provide too much, right?
26:04Where there's overload. That's That's
26:06the one of the biggest challenges that I
26:08have is that when I'm working with
26:10clients, they're giving me so much
26:11information that I sometimes have to
26:14clarify with them like, "Okay, we're
26:16overloading here. Not all of this can be
26:18true at the same time, right?" So, I
26:20know I'm not it's not a direct answer to
26:22your question, but it's really like the
26:25opposite in my experience that is the
26:26challenge. It's It's the overload of
26:28information, not the lack thereof.
26:30>> When you're fetching this context, are
26:32there certain AI models that work better
26:34than others? Like for me in my instance,
26:36why I'm asking this is that Gemini
26:37models historically have had a massive
26:39context window, right? Where sometimes
26:41you have the Sonnet, the Opus, the GPT
26:43models. Yes, the context windows are
26:45improving, but not as good as Gemini in
26:47certain areas of scanning like PDFs or
26:50long video transcripts. Like what AI
26:52models have you seen work best for this?
26:55>> Yeah, you know, to be quite honest, I
26:57will personally my preference is Claude,
27:00the Claude ecosystem. I am a fan of
27:03uh Soul in the brief time that I've been
27:04able to use it. But, [clears throat]
27:07the thing that I think really makes the
27:08difference is the orchestration,
27:11right? Because
27:12in the way that context is
27:14operationalized, I think that's where
27:17the uh real difference takes place
27:20because if you think about orchestration
27:23methodologies and in in a Claude
27:26scenario and in a Claude workflow, I'll
27:28have a orchestrator agent that might
27:32leverage sub agents for different jobs,
27:35right? And the sub and and I probably
27:38over index on this a little bit, but
27:40there's a method to the madness because
27:42I find that the sub agents who do the
27:44best works are the ones who have very
27:46limited access to the things that they
27:48don't need, right? So, I think what it
27:51boils down to is to get the best like to
27:55to operationalize the context in the
27:57best way for quality, efficiency, and
27:59consistency,
28:00it's about having a really good
28:01orchestration framework that knows how
28:03to use it.
28:05>> Yeah, that's a great point. Second brain
28:07as a service, do you think this is a
28:08real business with long-term potential
28:10or just a fad that dies in the next 6 to
28:1212 months?
28:14>> I think I think the um discipline is
28:17here to stay in one form or another. I
28:20do think that in some form it is going
28:23to be um durable as a service ongoing. I
28:27think, you know, who knows what the um
28:30you know, big players in the industry
28:32will introduce over time,
28:34>> Yeah.
28:35>> but again, you may not want to be
28:37hitched to one of their wagons to your
28:39detriment, right? So, I do think that in
28:41some form
28:43this is going to be a need. There's
28:46going to be a lot of value and a lot of
28:47impact that can be delivered with a
28:50second brain or knowledge base as a
28:52service, no doubt.
28:54>> For sure. Last question I have for you,
28:55I'm going to ask you to get your crystal
28:56ball out and make a prediction. Let's
28:58say 12 months from now, what is the
29:00number one way that you think AI second
29:02brains could change knowledge work that
29:04no one's talking about right now?
29:07>> Yeah, that's a great question. Really
29:09hard to think about. My instinctive
29:13thought goes to like the hidden insight
29:15piece, right? Where
29:17you know, connecting different parts of
29:19the company that are not typically
29:22connected can be synthesized for new
29:25information,
29:26more strategic intelligence, and
29:29unlocking doors that, you know, people
29:32might not have thought were unlocked.
29:34And I think that, you know, could happen
29:35today.
29:36But perhaps like 12 months from now,
29:39there's even further adoption of AI. Who
29:42knows what models will be working with
29:44and how advanced they are, but they're
29:46all going to need context, they're all
29:48going to need information.
29:50I can't think of that changing in any
29:52way. Um and I think, you know,
29:56there's likely going to be a lot of
29:58services. I We're already seeing There's
30:01already a lot of open-source and premium
30:03solutions out there that do some flavor
30:05of this.
30:06A lot of the context engineering
30:08solutions, which I think is
30:09in the same lane that I see out there,
30:12are very much targeting engineers and
30:15software development, not so much
30:16knowledge work. So I think there's
30:19um right now a
30:21opportunity and a gap for this delivery
30:24for non-software related
30:27practices.
30:29>> I liked your point about insights.
30:30Remember I made that example of the SEO
30:32and accounting thing before? You know,
30:34one thing in my brain that just kind of
30:36clicked is I know there's like
30:37proprietary information in accounting
30:39data. Don't get me wrong, but in order
30:41to rank effectively in AI search and
30:43traditional search results, you need
30:45unique context and data. So, I mean, if
30:47you're pulling in accounting data and
30:48maybe just generalizing it and finding
30:50trends and creating infographics to put
30:53in that content, right? These are just
30:55some ways my brain is spinning thinking
30:56about this because there are so many
30:58insights in someone's financial data or
31:01maybe internal data with usage or how
31:04people are using your platform if you're
31:06a product that you can incorporate into
31:08your content somehow as a unique
31:10>> Yeah.
31:11I have [clears throat] a a really good
31:12real-world example of this. I work with
31:13a client who they have a call center.
31:16And their call center is, you know,
31:18fielding calls from prospects and
31:20customers all around the country.
31:22And [clears throat] the call center
31:23typically operates in somewhat of a
31:25silo.
31:26Right? They're not really harvesting any
31:28insights or data from the transcripts.
31:30They use them for coaching, for Q&A, and
31:32that's really about it.
31:34And I'm working with them on a workflow
31:36to operationalize
31:38how they're managing their transcripts
31:40and and creating workflows to
31:42scale up how they can manage QA and
31:45coaching and all that kind of stuff.
31:47However,
31:48we're also deliberately focusing on what
31:51can marketing learn from this body of
31:53work, from all of these customer calls,
31:55from the language and the words they
31:56use, right? The you know, the obvious
31:59stuff like what are the trending topics,
32:01right? Is there seasonality, right?
32:02There's just so much that can be
32:04harvested now from
32:07>> [clears throat]
32:07>> this call center that the marketing team
32:10can then leverage, right? So, you know,
32:12when I say it out loud, it sounds a
32:13little obvious, but
32:15it's, [clears throat] you know, that's
32:16one of that's a real-world connection
32:17where those that connectivity was not
32:20there for this company in the past.
32:22>> All right, Adam, so this has been fun.
32:23Where can people follow you online to
32:25learn more about what you're doing with
32:26Second Brains and just all the other
32:28stuff that you're putting out in the
32:29internet?
32:30>> Yeah, you can find me at viableedge.com
32:33uh and viableedge on all of the
32:35platforms, YouTube, X, um Adam Sandler
32:39on LinkedIn, feel free to reach out.
32:41That's where I'm most active.
32:43>> Awesome. Well, I appreciate everyone
32:45who's made it this far into this edition
32:47of AI Rabbit Holes. Be sure to tune in
32:50again next time.

About this transcript

This page contains the full transcript of He Sells AI Second Brains to Businesses (For $1000s) by Ryan Doser, generated from the public captions YouTube serves with the video. The transcript has 6,034 words across 932 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.