3 $1M+ AI startup ideas to build in 2025 (and how to grow them)

00:49:42
https://www.youtube.com/watch?v=BjxS-AQaDkE

Summary

TLDRIn this podcast episode, the shift from human teams to AI agents in business operations is discussed, emphasizing the opportunity in building businesses around AI rather than just improving AI technology itself. Arvid shares three AI-centered business ideas that can be implemented immediately, each solving tangible problems with potential for significant financial gain. One idea focuses on creating an AI co-founder tool that could perform tasks like marketing and sales autonomously. Arvid delves into strategic steps for implementation, emphasizing the importance of targeting a specific niche and the potential for high return on investment. Another idea is based on automating the constant optimization of code bases, potentially revolutionizing how software maintenance is conducted. The third idea involves aggregating and summarizing industry-specific data for users in various formats tailored to individual consumption preferences. The episode underscores the rapidly growing role of AI in startup ecosystems and discusses the philosophical and practical implications of having AI as a part of founding teams.

Takeaways

  • 🤖 AI agents are increasingly replacing traditional team roles in business.
  • 💡 The real opportunity lies in building businesses around existing AI tools.
  • 📈 Arvid suggests focusing on niche markets for AI-driven solutions.
  • 🤝 AI co-founders could handle tasks like marketing without requiring equity.
  • 🚀 AI can accelerate various aspects of business, from ideation to operations.
  • 🛠 Tools that automate codebase optimization could revolutionize software maintenance.
  • 🔄 Aggregating and summarizing industry-specific data can be a profitable venture.
  • 🎯 Startups should leverage AI to enhance every phase of their business strategy.
  • ⚗️ Experimentation with AI agents can lead to innovative business models.
  • 📊 Focusing pricing models on AI tools' ROI is crucial.
  • 🌍 Understanding your market is key to tailoring AI solutions effectively.
  • 🗣 There's potential for developing AI-driven communication strategies for businesses.

Timeline

  • 00:00:00 - 00:05:00

    Arvid discusses how AI agents are replacing traditional teams and emphasizes that the opportunity lies not in creating better AI, but in building businesses around AI agents. He plans to share three business ideas centered around AI agents that are practical and potentially lucrative, noting that they are straightforward concepts that reveal their value once recognized.

  • 00:05:00 - 00:10:00

    In a discussion with Arvid, he points out that his entrepreneurial ideas stem from personal needs as a Founder and solo-preneur. He emphasizes using AI and automation to perform tasks that individuals wish were offloaded. Arvid has a specific interest in how computers and AI can execute tasks traditionally completed by humans.

  • 00:10:00 - 00:15:00

    Arvid introduces the concept of an AI co-founder, a virtual entity capable of handling business tasks autonomously as well as collaboratively with the human founder. He envisions this AI as a reliable, 24/7 partner that integrates with personal and business data to execute tasks such as marketing and sales with high efficiency.

  • 00:15:00 - 00:20:00

    The discussion proceeds about how an AI co-founder can become smarter over time, learning business nuances and performing specialized tasks like marketing or sales, tailored to specific needs. Arvid imagines the AI co-founders forming networks, which could share insights and strategies across businesses.

  • 00:20:00 - 00:25:00

    Arvid contemplates the balance and potential limitations of having AI agents as co-founders, addressing the emotional elements of a co-founder relationship. He thinks alignment of personality in AI agents with founders can make them more relatable and effective. However, human creativity and partnership remain valuable.

  • 00:25:00 - 00:30:00

    The idea of an AI co-founder is compared to current technologies, illustrating potential benefits and drawbacks. Arvid discusses iterative development and potential startup pitfalls regarding pinpointing target markets and dealing with diverse feedback. He suggests iterating carefully to create valuable AI business tools.

  • 00:30:00 - 00:35:00

    Arvid introduces another business concept—a silent, constant refactoring service for code bases. He envisions a tool that autonomously optimizes and corrects code errors, learns, and simulates improvements. This idea taps into AI's potential to perform continuous, autonomous software development enhancements.

  • 00:35:00 - 00:40:00

    The technical and market aspects of building a continuous code refinement system are discussed. Arvid considers existing solutions and outlines how a SaaS offering could model this concept, highlighting his name suggestion for the platform and considering strategic market entry points such as targeting specific niches.

  • 00:40:00 - 00:49:42

    Arvid's final concept revolves around personalized data aggregation platforms. He envisions services that curate information from broad data streams into user-specific formats, leveraging AI to tailor content delivery to individual preferences, thus bridging informational gaps efficiently in various industries.

Show more

Mind Map

Video Q&A

  • What is the main focus of the podcast episode?

    The main focus is on building businesses around AI agents, with Arvid presenting three business ideas related to this.

  • Why does Arvid believe most founders are building in the wrong direction?

    Arvid believes most founders are focused on creating better AI, rather than building businesses around the existing AI agents, which is where he sees real opportunity.

  • What kind of business ideas does Arvid present?

    Arvid presents three business ideas that revolve around AI agents solving real-world problems, with potential to generate significant revenue.

  • Why are these AI business ideas described as not revolutionary?

    The ideas are described as not revolutionary because they seem obvious once explained, which is part of their appeal and effectiveness.

  • What is the role of AI co-founders according to the episode?

    AI co-founders could take on tasks such as marketing or sales, doing what human co-founders would do but without equity and constantly available.

  • How should one start building the AI co-founder tool according to Arvid?

    One should start by focusing on a specific niche or customer base already interested in such a tool, tailoring it to their needs.

  • How does Arvid suggest pricing AI-driven tools?

    Arvid suggests looking at the ROI of the tool, possibly using a subscription model based on the tool's utility and cost savings.

  • What does the third idea about data aggregation entail?

    The third idea involves creating a business that aggregates, summarizes, and personalizes information from multiple data sources tailored to individual preferences.

  • Why does Arvid emphasize start-ups using AI today?

    Arvid emphasizes it because AI is seen as a strong accelerator in ideation, organizing, creating processes, implementing, and operating a business.

  • What is Arvid's perspective on creating bonds with AI co-founders?

    Arvid believes AI co-founders might form practical bonds with founders by aligning with their operational needs, though lacking emotional human attributes.

View more video summaries

Get instant access to free YouTube video summaries powered by AI!
Subtitles
en
Auto Scroll:
  • 00:00:00
    AI agents are replacing what teams of
  • 00:00:02
    people used to do but most Founders are
  • 00:00:05
    building in the wrong direction the real
  • 00:00:08
    opportunity isn't creating better AI
  • 00:00:10
    it's building businesses around AI
  • 00:00:13
    agents think of it like this the tools
  • 00:00:16
    are already here the market wants it but
  • 00:00:19
    no one's connecting the dots in this
  • 00:00:21
    episode Arvid breaks down three business
  • 00:00:24
    ideas each one could be started tomorrow
  • 00:00:27
    and they're all around AI agents each
  • 00:00:31
    one solves real problems and each one
  • 00:00:34
    could be worth millions the interesting
  • 00:00:36
    part well these ideas aren't
  • 00:00:38
    revolutionary they're obvious once you
  • 00:00:40
    hear them that's exactly why they work
  • 00:00:44
    and I hope you enjoy this
  • 00:00:47
    [Music]
  • 00:00:54
    episode Arvid you one of my favorite
  • 00:00:58
    Indie hacker solo Piner
  • 00:01:01
    people and I want you to tell people
  • 00:01:03
    what sort of ideas will they get by
  • 00:01:06
    listening to this podcast episode uh
  • 00:01:09
    well thank you for the compliment I can
  • 00:01:10
    only return it because I I follow you
  • 00:01:12
    with uh similar Vigor and enthusiasm but
  • 00:01:15
    hey the ideas that I have usually come
  • 00:01:17
    from just my reality of being a Founder
  • 00:01:20
    being a software founder being a an
  • 00:01:22
    entrepreneur being somebody who has and
  • 00:01:24
    wants to run a solo business that being
  • 00:01:26
    solo preneurial yet still needs to do
  • 00:01:28
    all the things at the same time wear all
  • 00:01:30
    the different hats and all of that stuff
  • 00:01:32
    that usually triggers many of a I wish
  • 00:01:35
    that was kind of thoughts right and then
  • 00:01:38
    often I have the the opportunity to note
  • 00:01:41
    that down and then there's a list of 200
  • 00:01:42
    different things that I need and want
  • 00:01:45
    but never get to make because I'm
  • 00:01:46
    focusing on the thing that I'm currently
  • 00:01:48
    building always right there's always
  • 00:01:49
    that one thing you're focused on so my
  • 00:01:51
    ideas uh that I have are mostly focused
  • 00:01:54
    on things
  • 00:01:56
    that other human being should be doing
  • 00:02:00
    but now we kind of have technology that
  • 00:02:01
    might be able to do it for us so that's
  • 00:02:05
    that's kind of where all of this is
  • 00:02:06
    going so we talking about like Ai and
  • 00:02:09
    automation type ideas I think so mostly
  • 00:02:13
    like I'm I've always been a big fan of
  • 00:02:15
    anything machine learning anything AI
  • 00:02:17
    before it was AI right the the whole GPT
  • 00:02:20
    movement the llms and that stuff that
  • 00:02:22
    made it extremely accessible but even
  • 00:02:24
    before that I always wanted to see what
  • 00:02:28
    human skill based task can I give to a
  • 00:02:31
    machine and then the machine will do it
  • 00:02:33
    good enough so that in most cases the
  • 00:02:35
    reliable results that I get are things
  • 00:02:37
    that I can just use that's always been
  • 00:02:39
    my approach to computers they're great
  • 00:02:41
    and if we can just tune them just right
  • 00:02:43
    they can do things that other people
  • 00:02:45
    would have to do for us but we don't
  • 00:02:46
    have to like pay them or listen to them
  • 00:02:48
    or put them to bed or anything right
  • 00:02:50
    like we we have this this always on
  • 00:02:52
    thing that does work for us better than
  • 00:02:54
    we can do it so that's where most of
  • 00:02:56
    this is focused around yes AI has been
  • 00:02:58
    like I I don't think if if you start a
  • 00:03:00
    business
  • 00:03:01
    today you can avoid AI like or you
  • 00:03:04
    should avoid it you should never avoid
  • 00:03:06
    it either in the process of ideation the
  • 00:03:10
    process of organizing it creating
  • 00:03:12
    processes themselves implementing the
  • 00:03:14
    business operating the business having
  • 00:03:16
    parts of the business themselves be
  • 00:03:18
    features that are running on AI anything
  • 00:03:20
    AI is a very strong uh accelerator for
  • 00:03:24
    all of these things so you'll find it in
  • 00:03:26
    all of the ideas that I have to either
  • 00:03:28
    large degrees or smaller degrees
  • 00:03:30
    I love it I feel like uh it's
  • 00:03:33
    counterintuitive but the the laziest
  • 00:03:36
    people are the wealthiest people and and
  • 00:03:39
    what I mean by that is if you can figure
  • 00:03:41
    out how to automate a lot of what you're
  • 00:03:43
    doing um and if you can figure out how
  • 00:03:46
    to get good economics then you've got a
  • 00:03:51
    business on your hands so um I I I'm I'm
  • 00:03:57
    chomping at the bit I want to hear these
  • 00:03:58
    ideas what do you want to start with I
  • 00:04:01
    think I'm I'm going to go with like the
  • 00:04:03
    the biggest and maybe most complicated
  • 00:04:04
    one first cuz that's the one I wish I
  • 00:04:06
    had the most particularly because I'm a
  • 00:04:09
    solo founder like I said all those hats
  • 00:04:11
    and some of the hats that I'm wearing I
  • 00:04:13
    really enjoy but many I don't I'm not a
  • 00:04:16
    really good marketer I do that and I I
  • 00:04:18
    can also sell and I can also do bizde
  • 00:04:21
    and go through my financials and all of
  • 00:04:22
    that but I wish there was somebody else
  • 00:04:23
    to do this for me and most of the time
  • 00:04:26
    we have just businesses that kind of are
  • 00:04:30
    agents for us that like agencies that do
  • 00:04:32
    this work for us particularly with like
  • 00:04:34
    taxes or marketing design we kind of
  • 00:04:36
    source that out but what I want is an AI
  • 00:04:39
    co-founder I want a virtual always on
  • 00:04:42
    247 always available co-founder that
  • 00:04:44
    does that stuff for me that is not just
  • 00:04:46
    there for me to kind of bounce back and
  • 00:04:48
    forth ideas we can we already do this I
  • 00:04:50
    think if you look in a Founder Community
  • 00:04:52
    there are many many people who now use
  • 00:04:54
    jgpt or anthropics Claude in this
  • 00:04:57
    constant conversation about what should
  • 00:04:59
    I do next or how should I do this give
  • 00:05:01
    me 10 versions of this give like write
  • 00:05:03
    an email but write it with the tone of
  • 00:05:05
    that person and then you kind of like P
  • 00:05:07
    Meal all of these tasks that you would
  • 00:05:10
    otherwise have to do all by yourself
  • 00:05:11
    together with the help of an AI I think
  • 00:05:13
    that's like the initial stage we're
  • 00:05:14
    already at that most people very
  • 00:05:16
    clumsily do by themselves just trying to
  • 00:05:19
    figure things out right they might use
  • 00:05:21
    an API and then build their own tools to
  • 00:05:23
    do this for them or they go right to the
  • 00:05:25
    browser and do it on the web interface
  • 00:05:27
    that's what many people particularly
  • 00:05:29
    solo founder do right now what I want is
  • 00:05:32
    to have a co-founder with zero Equity is
  • 00:05:34
    a very selfish thing I guess but I that
  • 00:05:37
    is just constantly thinking about the
  • 00:05:39
    tasks that I give them and maybe in a
  • 00:05:42
    future version of that product or in a
  • 00:05:44
    version of that product actually execute
  • 00:05:47
    on those tasks right the easiest example
  • 00:05:50
    would be I want somebody who does
  • 00:05:52
    marketing for me but is smart about it
  • 00:05:54
    has my voice my tone but is also
  • 00:05:57
    intelligent enough to look into the
  • 00:05:58
    lives of the people that they're talking
  • 00:06:00
    to right if you I don't know if you
  • 00:06:02
    Source your your sales Outreach emails
  • 00:06:05
    or your your marketing channel you know
  • 00:06:08
    the the sources where you want to go
  • 00:06:09
    where you want to direct stuff add from
  • 00:06:12
    tools like Apollo you get emails you get
  • 00:06:14
    like the industry that people work in
  • 00:06:15
    maybe the businesses they work in well
  • 00:06:17
    why not have an AI that constantly
  • 00:06:19
    scrapes the web constantly scour all
  • 00:06:22
    potential data sources to make up an
  • 00:06:24
    internal representation of that person
  • 00:06:27
    and then figures out the best way to
  • 00:06:28
    talk to that person then gets a task for
  • 00:06:30
    me to sell this product for to this
  • 00:06:32
    person and then they start having a
  • 00:06:34
    conversation with them or they start
  • 00:06:36
    giving me the tools to have this
  • 00:06:38
    conversation myself right anywhere
  • 00:06:40
    between complete Independence and just
  • 00:06:44
    helping me do my thing like a VA would
  • 00:06:45
    do or maybe even an advanced uh
  • 00:06:47
    executive assistant that does jobs for
  • 00:06:49
    me I want an AI to do this and a
  • 00:06:52
    business that does this as a kind of a
  • 00:06:54
    scalable thing does not only offer me
  • 00:06:57
    like one AI they offer me like a
  • 00:06:59
    marketing AI they offer me a CTO AI or
  • 00:07:02
    like a sales AI somebody who virtually
  • 00:07:05
    like a virtual person that has a lot of
  • 00:07:08
    deep skill in that field looks at it
  • 00:07:11
    from certain perspectives that may also
  • 00:07:14
    change over time learns what my my
  • 00:07:16
    business is about and then constantly
  • 00:07:19
    churns behind the scenes new ideas old
  • 00:07:22
    ideas ideas from somebody else into this
  • 00:07:26
    presence that I as a Founder can
  • 00:07:28
    interact with and and the the actual
  • 00:07:30
    thing that I find makes this the most
  • 00:07:33
    potentially interesting idea is that
  • 00:07:35
    imagine you you run this business it's
  • 00:07:36
    kind of software service business people
  • 00:07:38
    sign up they train their own AIS on
  • 00:07:40
    their code base on the the documents
  • 00:07:43
    they have in their knowledge base on
  • 00:07:44
    their notion document that has all the
  • 00:07:45
    weird thoughts that they have about
  • 00:07:47
    their business they throw this all in
  • 00:07:49
    and train this one person this one AI
  • 00:07:51
    person sh don't tell anyone but I've got
  • 00:07:55
    30 plus startup ideas that could make
  • 00:07:57
    you Millions and I'm giving them away
  • 00:08:01
    for free these aren't just random
  • 00:08:04
    guesses they're validated Concepts from
  • 00:08:07
    entrepreneurs who built hundred million
  • 00:08:10
    plus businesses I've compiled them into
  • 00:08:14
    one simple
  • 00:08:16
    database compiled from hundreds of
  • 00:08:19
    conversations I've had on my podcast but
  • 00:08:22
    the main thing is most of these ideas
  • 00:08:25
    don't need a single investor some cost
  • 00:08:28
    nothing to start I'm pretty much handing
  • 00:08:31
    you a cheat sheet the idea bank is your
  • 00:08:34
    startup
  • 00:08:35
    shortcut just click below to get
  • 00:08:38
    access your next cash flowing business
  • 00:08:41
    is waiting for you but you as a sass
  • 00:08:44
    owner you operate many of these things
  • 00:08:46
    at the same time right you have the CTO
  • 00:08:48
    for that one company and the CTO for
  • 00:08:50
    that other company well how about they
  • 00:08:52
    start interacting they go to into
  • 00:08:54
    virtual AI on AI meetings and just kind
  • 00:08:57
    of bring ideas to each other
  • 00:08:59
    right like to kind of start exchanging
  • 00:09:02
    things that worked for this business
  • 00:09:04
    maybe it works for you kind of almost
  • 00:09:07
    like Mastermind groups of AI agents that
  • 00:09:10
    interact with each other and bring new
  • 00:09:12
    ideas into your own business like I'm
  • 00:09:14
    thinking of how can we give this and
  • 00:09:16
    this is a term that I've read over the
  • 00:09:18
    last couple weeks so so many times like
  • 00:09:20
    agentic approach like the idea of having
  • 00:09:22
    agents to do stuff for us how can we
  • 00:09:25
    bring this into the deliberation process
  • 00:09:28
    of a solo founder who has nobody to be
  • 00:09:30
    their agent unless they pay them on like
  • 00:09:32
    a you know kind of contractor basis
  • 00:09:35
    that's that's one of the things that I
  • 00:09:36
    would really really
  • 00:09:38
    want
  • 00:09:39
    so I don't think you're alone I think
  • 00:09:42
    that every founder wants
  • 00:09:45
    this uh although I'll I'll speak I know
  • 00:09:48
    someone is listening to this and be like
  • 00:09:50
    yeah
  • 00:09:51
    but Arvid uh my co-founder like there's
  • 00:09:55
    this you know emotional element that I
  • 00:09:58
    have with a co-founder your agent will
  • 00:10:00
    never be able to do something like that
  • 00:10:02
    what's what's your response to someone
  • 00:10:03
    who says that I I mean probably not
  • 00:10:06
    because they're not a human being but
  • 00:10:08
    maybe that is not the thing that you
  • 00:10:10
    need when you even think about getting
  • 00:10:12
    an AI co-founder right it's not supposed
  • 00:10:14
    to be your wifeu or anything like this
  • 00:10:16
    right this is this is not a kind of a an
  • 00:10:19
    emotional attachment it's not an AI
  • 00:10:20
    girlfriend right or maybe it is but on a
  • 00:10:22
    professional level here's the thing like
  • 00:10:24
    all these llms all these AI tools that
  • 00:10:26
    we have in the world right now they're
  • 00:10:28
    effectively gas lighting engines right
  • 00:10:30
    that's the world's biggest gaslighting
  • 00:10:32
    process that you can ever imagine is
  • 00:10:34
    like asking a question of an llm and
  • 00:10:36
    then seeing as it tries to convince you
  • 00:10:38
    that it has any idea what it's talking
  • 00:10:40
    about like that happens on chat GPT all
  • 00:10:42
    the time it's always trying to Gaslight
  • 00:10:43
    you into believing that it knows what
  • 00:10:45
    it's talking about and then you tell it
  • 00:10:47
    I don't think that's right and it says
  • 00:10:48
    oh so I'm so sorry I completely missed
  • 00:10:50
    that here's the actual right answer and
  • 00:10:52
    then you say no it's actually not right
  • 00:10:54
    as well because I check this and this
  • 00:10:56
    doesn't work oh I missed that too sorry
  • 00:10:58
    sorry here's the actual correct answer
  • 00:11:00
    this is like all what L&M do is to try
  • 00:11:02
    to convince you that they're right by
  • 00:11:04
    using phrasing that they think might
  • 00:11:06
    convince you so in a way that is also
  • 00:11:09
    what people do to each other to get
  • 00:11:10
    their ideas across the table and to get
  • 00:11:12
    things done right to get their their
  • 00:11:14
    interests dealt with so it may it might
  • 00:11:17
    be that we need to train these models in
  • 00:11:19
    a way that are aligned with the
  • 00:11:22
    personality of the founder that is using
  • 00:11:24
    the models right like like an AI
  • 00:11:26
    girlfriend has to be your type right you
  • 00:11:28
    have to train your AI girlfriend to be
  • 00:11:31
    as I don't I don't know like friendly or
  • 00:11:34
    funny or dismissive as you like people
  • 00:11:36
    have KS right so this will likely also
  • 00:11:39
    have to happen for any other virtual
  • 00:11:42
    person that is doing a job for you you
  • 00:11:45
    will have to train them they will have
  • 00:11:46
    to be kind of aligned and I don't mean
  • 00:11:48
    like actual model alignment the
  • 00:11:50
    technical term but they have to have a
  • 00:11:52
    personality that is trained into them
  • 00:11:54
    that is aligned with you maybe also
  • 00:11:56
    something you can pick maybe you can
  • 00:11:57
    pick your um the the kind of AI that you
  • 00:12:01
    want from these templates like I want to
  • 00:12:04
    have a Peter teal like person or an Elon
  • 00:12:06
    Musk like person or maybe just a
  • 00:12:08
    completely different person that the AI
  • 00:12:10
    system doesn't even know yet but then it
  • 00:12:12
    goes and scrapes their blog posts and it
  • 00:12:15
    scrapes their their social presence and
  • 00:12:17
    it integrates that into a a virtual
  • 00:12:19
    version of that person with that
  • 00:12:22
    particular kind of job I think we might
  • 00:12:25
    even create bonds between Founders and
  • 00:12:28
    AI agents
  • 00:12:29
    that are stronger than between Founders
  • 00:12:31
    and other Founders because all this
  • 00:12:34
    social tension between co-founders that
  • 00:12:37
    is often quite strong particularly when
  • 00:12:39
    it's crunch time or when Runway is
  • 00:12:41
    running out I think you will not get an
  • 00:12:44
    anxious AI unless you train it so if you
  • 00:12:48
    are like two months before your business
  • 00:12:50
    has to close because you have no more
  • 00:12:51
    money and your human co-founder would
  • 00:12:53
    start trying to look for a job because
  • 00:12:55
    they have a family to feed your AI
  • 00:12:57
    co-founder is like okay let's try every
  • 00:12:59
    single thing until the last second then
  • 00:13:01
    you can that you can afford the
  • 00:13:02
    subscription to the service and then
  • 00:13:04
    even then they might even give you a
  • 00:13:06
    week for free or whatever to try and
  • 00:13:08
    Salvage your business I think the I
  • 00:13:10
    don't want to remove Humanity from
  • 00:13:11
    co-founder relationships obviously that
  • 00:13:12
    is not something I'm looking for ideally
  • 00:13:15
    this co-founder gets me to a point where
  • 00:13:16
    I can actually find a real co-founder to
  • 00:13:19
    be a creative and like explorative human
  • 00:13:21
    being to work with me but for the first
  • 00:13:24
    couple months of any business having
  • 00:13:26
    access to a person that has the the
  • 00:13:29
    knowledge of the world combined into a
  • 00:13:32
    personality that can help you make
  • 00:13:33
    choices quickly I think that is worth
  • 00:13:35
    something it's not perfect nothing ever
  • 00:13:37
    is but it's definitely better than you
  • 00:13:39
    having to read every single book on
  • 00:13:41
    marketing on sales out there trying to
  • 00:13:43
    distill it into choices that you've
  • 00:13:44
    never made before right I feel it's a
  • 00:13:47
    crutch but we have crutches to do
  • 00:13:50
    crutches work right so for that I think
  • 00:13:53
    it it definitely is worth I would pay
  • 00:13:54
    for this I would pay 50 bucks 100 bucks
  • 00:13:57
    for an agent like this every single
  • 00:13:58
    month just to see if it gets me an Roi
  • 00:14:02
    of 50 or 100 bucks which it likely will
  • 00:14:04
    because it's always doing stuff right it
  • 00:14:06
    can always do Outreach on in my name it
  • 00:14:08
    can always do research in my name it can
  • 00:14:10
    like that's that's what I'm currently
  • 00:14:11
    building I'm building pod scan which is
  • 00:14:12
    a a podcast scanning app that gives
  • 00:14:15
    realtime mention notifications to people
  • 00:14:17
    who sign up for it and if I know that my
  • 00:14:20
    customers are talking about a certain
  • 00:14:22
    competitors's product then I can have
  • 00:14:24
    this agent listen to all the podcasts in
  • 00:14:26
    the world look at all the blogs in the
  • 00:14:28
    world look at all the things that happen
  • 00:14:30
    in real time and tell me hey this person
  • 00:14:32
    was talking about a competitor that you
  • 00:14:35
    you slept until 6 they talked about it
  • 00:14:36
    at 4 like in the next 30 minutes you
  • 00:14:38
    still have the window to get them right
  • 00:14:40
    and and to send them a message and maybe
  • 00:14:42
    sell your product that's the kind of
  • 00:14:44
    stuff that I see in this this agent
  • 00:14:46
    world so sometimes I hear a really good
  • 00:14:49
    idea like this like I buy this idea 100%
  • 00:14:52
    yes it's got some weird social
  • 00:14:54
    implication we're going to we'll talk
  • 00:14:56
    that about that another time on another
  • 00:14:57
    pod you know you scratch to surface on
  • 00:14:59
    it but from like a business opportunity
  • 00:15:02
    perspective this will exist where I
  • 00:15:06
    think a lot of people go wrong with
  • 00:15:07
    startup ideas and building startups in
  • 00:15:09
    general is they they pick the right idea
  • 00:15:12
    but they pick the wrong order of
  • 00:15:14
    operations so when I hear this idea I'm
  • 00:15:17
    like okay great check this is like a
  • 00:15:19
    good idea but I think that if you
  • 00:15:23
    actually went in to build this product
  • 00:15:26
    it would be like boiling the ocean
  • 00:15:29
    there's almost too many diverse startups
  • 00:15:33
    and startup Founders in for you to
  • 00:15:35
    create something
  • 00:15:36
    that is going
  • 00:15:38
    to really really be Roi positive in my
  • 00:15:43
    opinion uh I think that you know you'd
  • 00:15:45
    be you'd build it and then you'd you'd
  • 00:15:48
    get such diverse feedback from such
  • 00:15:50
    different types of Founders that you'd
  • 00:15:52
    look at it and be like you know just
  • 00:15:55
    running around with your pants on fire
  • 00:15:57
    basically oh yeah so sure I think the
  • 00:16:00
    way to go about and I'm curious your
  • 00:16:03
    perspective on this CU you're kind of
  • 00:16:05
    doing it with pod scan like you could
  • 00:16:06
    have done you know alerts for everything
  • 00:16:10
    but you chose a niche you chose podcast
  • 00:16:13
    I think that if you were going to go and
  • 00:16:15
    build this idea which I think could
  • 00:16:17
    generate millions of dollars a year in
  • 00:16:19
    ARR you would need to pick a specific
  • 00:16:22
    type of founder in a specific Niche for
  • 00:16:25
    example YC founder going through why YC
  • 00:16:29
    SAS
  • 00:16:30
    founder yeah for sure what do you think
  • 00:16:33
    yes I I think like that's the thing with
  • 00:16:36
    these models right if you if you want to
  • 00:16:38
    like also it's quite expensive to train
  • 00:16:40
    a model to do a certain thing right you
  • 00:16:42
    can you can take like off the shelf
  • 00:16:44
    stuff apis like GPD 40 or whatever right
  • 00:16:47
    now integrate them give them some system
  • 00:16:49
    prompt or whatever and they will be
  • 00:16:50
    behave slightly differently but if you
  • 00:16:52
    really want to have a specific thing for
  • 00:16:55
    a specific Niche that talks the right
  • 00:16:57
    language that gives people the right
  • 00:16:59
    kind of motivation the right pushes and
  • 00:17:01
    that kind of stuff then it gets quite
  • 00:17:04
    expensive quickly so you really have to
  • 00:17:07
    well that's the thing right it kind of
  • 00:17:08
    have a chicken and egg situation do you
  • 00:17:09
    want to have a specific model for a
  • 00:17:11
    specific Niche then you have to spend uh
  • 00:17:13
    quite some money on the specific model
  • 00:17:15
    so you can you know monetize this
  • 00:17:17
    particular Niche and the broader you go
  • 00:17:19
    the more likely you can use the already
  • 00:17:21
    existing broad uh models that exist out
  • 00:17:24
    there I think this is a problem that all
  • 00:17:25
    those those players have that are
  • 00:17:27
    offering apis to LMS like open AI
  • 00:17:30
    anthropic all of these players they have
  • 00:17:32
    to have models available that can do
  • 00:17:34
    everything well so they can't really
  • 00:17:36
    have models that do one one thing
  • 00:17:38
    specifically specifically well right you
  • 00:17:40
    see this with coding right now you have
  • 00:17:43
    cursor which is a a tool that is so
  • 00:17:47
    focused on doing just the coding task
  • 00:17:49
    that the models that they train and the
  • 00:17:52
    the systems that they build are Miles
  • 00:17:54
    better than the general models even
  • 00:17:56
    though you can plug them into your your
  • 00:17:58
    vs code or your PHP storm that's what I
  • 00:18:01
    use you can you can get these big models
  • 00:18:04
    the the anthropic has the clot on it 3.5
  • 00:18:08
    or whatever now plug them in they're
  • 00:18:10
    great but then you use cursor and you
  • 00:18:12
    see oh this is what a good model can do
  • 00:18:14
    right and you you kind of probably want
  • 00:18:16
    to have a similar experience for this AI
  • 00:18:18
    co-founder service oh this is what a
  • 00:18:21
    good specifically trained model can do
  • 00:18:23
    for a Founder for doing this role but
  • 00:18:26
    that is that costs you money and for
  • 00:18:27
    that you probably might have to raise
  • 00:18:29
    enough to to afford like a cluster of
  • 00:18:31
    gpus to train this for even for your
  • 00:18:33
    first couple hundred or dozens of
  • 00:18:35
    customers right what I would what I
  • 00:18:37
    would do if I were to bootstrap this
  • 00:18:38
    whole thing would just be again I think
  • 00:18:41
    coner approach from the beginning try to
  • 00:18:43
    set this up for somebody who already is
  • 00:18:46
    kind of using it but make it a bit more
  • 00:18:48
    streamlined make it a bit more automated
  • 00:18:50
    right instead of them having to go into
  • 00:18:52
    their chat GPT and going into this
  • 00:18:54
    conversation that they had before and
  • 00:18:56
    then resuming there you kind of
  • 00:18:58
    facilitate that for them you kind of
  • 00:19:00
    pull all the information you you add all
  • 00:19:02
    the documents that they have
  • 00:19:03
    automatically so they can be used for R
  • 00:19:06
    for retrieval like and and and smart
  • 00:19:08
    kind of vectorization you embed all the
  • 00:19:10
    documents that they have you embed all
  • 00:19:11
    the conversations that they have and you
  • 00:19:13
    use different Technologies to make it
  • 00:19:15
    kind of unified so they have one place
  • 00:19:17
    where they can have this conversation
  • 00:19:19
    where before they would have to go
  • 00:19:20
    through all these different jump through
  • 00:19:21
    these hoops and then I guess that will
  • 00:19:24
    tell you what is needed and this is the
  • 00:19:26
    iterative approach to this if you were
  • 00:19:28
    to to raise and to just like throw money
  • 00:19:31
    at the problem you probably would skip
  • 00:19:35
    this this particular iteration and you
  • 00:19:36
    would just build the problem that you
  • 00:19:38
    would want to use yourself at least
  • 00:19:40
    that's my experience with like having a
  • 00:19:42
    lot of funding and having no funding at
  • 00:19:43
    all right if you have funding you can
  • 00:19:46
    assume that your assumptions are kind of
  • 00:19:48
    validated almost through having gotten
  • 00:19:50
    funding for it even though I guess this
  • 00:19:51
    warrants a whole other conversation
  • 00:19:54
    around like the the viability of ideas
  • 00:19:56
    and how you present them but if you were
  • 00:19:58
    to bootst this I think conar approach
  • 00:20:01
    getting somebody who is already talking
  • 00:20:03
    about this on Twitter or whatever where
  • 00:20:05
    they oh yeah I I am using chpt as my CMO
  • 00:20:10
    you ask them hey I see you use this I
  • 00:20:13
    want to build this as a product that
  • 00:20:14
    does all of this for you plus then the
  • 00:20:16
    other things that you promise in the
  • 00:20:17
    future do you want to try this with me
  • 00:20:20
    and then you go that way that's that's
  • 00:20:21
    kind of my Approach I I appreciate it um
  • 00:20:26
    I think there's a lot in that one
  • 00:20:29
    comment section on YouTube please let us
  • 00:20:31
    know what you think of idea number one
  • 00:20:33
    yeah please yeah there's so many layers
  • 00:20:36
    to this too right there's the the whole
  • 00:20:38
    the moralistic argument there's the the
  • 00:20:40
    operational one do I want a machine to
  • 00:20:42
    tell me what to do do I want a machine
  • 00:20:43
    to tell or to do what I what it thinks
  • 00:20:46
    it should do like there are so many
  • 00:20:49
    almost philosophical problems on this
  • 00:20:50
    level alone that this is worth uh having
  • 00:20:53
    a conversation but maybe maybe I can
  • 00:20:55
    segue to my second idea because it's
  • 00:20:56
    related it it it shares here's this this
  • 00:20:59
    agentic approach this an AI doing work
  • 00:21:03
    in the background to do things for me
  • 00:21:05
    but it becomes much more technical it's
  • 00:21:07
    not on the entrepreneurial level anymore
  • 00:21:09
    this is now on the codebase level I want
  • 00:21:10
    to have a silent constant refactoring as
  • 00:21:13
    a service that's what I want I want to
  • 00:21:15
    have my code base and I want to have it
  • 00:21:17
    in a repository I want to connect this
  • 00:21:19
    repository with a service and that
  • 00:21:22
    service does the following thing it all
  • 00:21:24
    it checks for errors in my code it lints
  • 00:21:27
    my code it analyzes my code code for
  • 00:21:30
    syntax errors or errors of logic and it
  • 00:21:33
    obviously suggests how to fix this we
  • 00:21:36
    already see this this already exists if
  • 00:21:38
    you go to Sentry which is one of these
  • 00:21:39
    error tracking tools that a lot of
  • 00:21:40
    developers use they already have a beta
  • 00:21:43
    feature right now where if there is an
  • 00:21:45
    error it tries to figure out why that
  • 00:21:48
    error exists and then tries to suggest a
  • 00:21:50
    way of fixing it and creates a poll
  • 00:21:52
    request for you to either accept or deny
  • 00:21:54
    and then the code either goes into your
  • 00:21:56
    code base and whatever Branch you might
  • 00:21:57
    be on hopefully the def branch and not
  • 00:21:59
    your production Branch because can't
  • 00:22:01
    trust automatically generated code by
  • 00:22:03
    another company just yet but you know
  • 00:22:05
    that's that's where we're at the that
  • 00:22:06
    stuff exists and it's really really cool
  • 00:22:09
    that they they have the funnel right
  • 00:22:10
    they have the error they have the full
  • 00:22:12
    access to the code base they have all
  • 00:22:13
    past errors they have all past errors
  • 00:22:16
    like this from other clients that they
  • 00:22:17
    have they have this massive database of
  • 00:22:20
    errors and then also they have access to
  • 00:22:23
    the code that was committed to fix the
  • 00:22:26
    error quick ad break let me tell you
  • 00:22:29
    about a business I invested in it's
  • 00:22:30
    called boring marketing.com so a few
  • 00:22:34
    years ago I met this group of people
  • 00:22:36
    that were some of the best SEO experts
  • 00:22:39
    in the world they were behind getting
  • 00:22:41
    some of the biggest companies found on
  • 00:22:43
    Google The Secret Sauce is they've got a
  • 00:22:46
    set of technology and AI that could help
  • 00:22:50
    you outrank your competition so for my
  • 00:22:53
    own businesses I wanted that I didn't
  • 00:22:55
    want to have to rely on Mark Zuckerberg
  • 00:22:57
    I didn't want to depend on ads to drive
  • 00:22:59
    customers to my businesses I wanted to
  • 00:23:02
    rank high in Google that's why I like
  • 00:23:04
    SEO and that's why I use boring
  • 00:23:06
    marketing.com and that's why I invested
  • 00:23:09
    in it they're so confident in their
  • 00:23:10
    approach that they offer a 30-day Sprint
  • 00:23:12
    with 100% money back guarantee who does
  • 00:23:16
    that nowadays so check it out highly
  • 00:23:18
    recommend boring marketing.com right
  • 00:23:20
    like Sentry is at such a great position
  • 00:23:22
    to have all of this but you don't need
  • 00:23:24
    this you can have a system that just
  • 00:23:26
    constantly scans the code for these
  • 00:23:29
    things you can connect it to tools like
  • 00:23:31
    aor tracking tools or your log tracking
  • 00:23:33
    tools for that matter and the the first
  • 00:23:35
    stage would be error checking error
  • 00:23:37
    tracking but what I really wanted to do
  • 00:23:39
    that's the the final stage is that it
  • 00:23:41
    constantly tries to improve my code it
  • 00:23:43
    constantly takes a module looks at it
  • 00:23:45
    it's like this could be better makes the
  • 00:23:47
    change and then simulates this simulates
  • 00:23:50
    how the app would work with this change
  • 00:23:52
    runs all the unit tests runs all the
  • 00:23:54
    integration tests in a simulated in a
  • 00:23:56
    container dock a container somewhere or
  • 00:23:58
    a a a fictional virtual machine that is
  • 00:24:00
    spun up for this and everything is set
  • 00:24:02
    up to be as lifelike as possible it runs
  • 00:24:04
    it it runs performance tests and then it
  • 00:24:07
    figures out hm I thought this would be
  • 00:24:09
    faster but it's actually 20% slower next
  • 00:24:12
    idea goes to the next idea implements a
  • 00:24:14
    different variety of this and does the
  • 00:24:16
    same thing runs all the tests runs the
  • 00:24:18
    performance analysis okay this is 2%
  • 00:24:20
    faster I'm going to keep doing this
  • 00:24:22
    optimiz this UND that and then after
  • 00:24:24
    maybe a day or two of constant code
  • 00:24:27
    optimization
  • 00:24:28
    performance checking simulation back to
  • 00:24:31
    code back to optimization back to
  • 00:24:33
    Performance checking Loops of this it
  • 00:24:35
    presents me with something that it can
  • 00:24:37
    tangibly truthfully say is going to make
  • 00:24:40
    my application more performant without
  • 00:24:42
    me having to do anything the thing is
  • 00:24:44
    just constantly scanning my code new
  • 00:24:46
    features that I put in Old features that
  • 00:24:48
    are already there and tells me this
  • 00:24:51
    could be improved this could be improved
  • 00:24:53
    and as The Little Dot on the eye the
  • 00:24:55
    little final thing on top I want this
  • 00:24:58
    tool to connect to my road map whatever
  • 00:25:01
    that might be the Fe the features that I
  • 00:25:03
    plan to have in the future and already
  • 00:25:05
    think about how I can improve my
  • 00:25:07
    database uh and my my codebase all of
  • 00:25:09
    the things really my D my data schema
  • 00:25:12
    the code base my documentation all of
  • 00:25:14
    this my models to make it easier to
  • 00:25:17
    implement the features that I haven't
  • 00:25:19
    even thought about yet in the future I
  • 00:25:21
    want like this thing to be an autonomous
  • 00:25:23
    developer that just constantly tries to
  • 00:25:25
    make things better and gives me choices
  • 00:25:28
    fully developed code fully tested code
  • 00:25:31
    fully commented code commented for my
  • 00:25:33
    sake and its own sake because it's
  • 00:25:35
    probably going to use that code to
  • 00:25:36
    suggest more code in the future to make
  • 00:25:39
    my my database more stable more reliable
  • 00:25:42
    and more performant that's my silent
  • 00:25:44
    refactoring as a service because I have
  • 00:25:46
    no better name for this well I have a
  • 00:25:48
    better name for it um and I checked to
  • 00:25:52
    see if it was
  • 00:25:53
    available um and it is and this is the
  • 00:25:57
    type of there you go audience that
  • 00:25:59
    whenever I say a name for
  • 00:26:02
    anything doain gone like I hit publish
  • 00:26:04
    and it's like see you later so I think
  • 00:26:07
    that you know I've talked about this on
  • 00:26:09
    the Pod before I have a whole naming
  • 00:26:11
    guide how to name your startup um around
  • 00:26:13
    coming up with like kind of viral names
  • 00:26:15
    I actually think that naming is such an
  • 00:26:17
    underrated way to get customers right
  • 00:26:19
    now to stand out and humor is a really
  • 00:26:22
    good way to do it so I think this code
  • 00:26:24
    couldbe better.com it's awesome yeah
  • 00:26:28
    that's a great name I hope you
  • 00:26:30
    registered it already because I have a
  • 00:26:32
    keyboard here I don't ever register I
  • 00:26:35
    give it to the people that's why people
  • 00:26:36
    come to this podcast they come for the
  • 00:26:38
    ideas I don't gatekeep anything it's
  • 00:26:40
    yours take it go and find it if you came
  • 00:26:43
    here first you deserve it something like
  • 00:26:46
    that every developer knows if you've
  • 00:26:49
    shipped code this code could be better I
  • 00:26:51
    don't care if you're Sam Alman you know
  • 00:26:54
    I don't care if you're uh the CTO of of
  • 00:26:58
    meta like you you know that um you
  • 00:27:02
    always have some constraint either it's
  • 00:27:05
    a time constraint or or whatever but I
  • 00:27:07
    think that this idea if you're trying to
  • 00:27:09
    create the
  • 00:27:10
    Sentry if you're trying to compete with
  • 00:27:12
    Sentry ultimately which I think is just
  • 00:27:14
    an interesting kind of like prompt right
  • 00:27:16
    Sentry a lot of people don't know this
  • 00:27:18
    but centry is like a three billion
  • 00:27:19
    dollar company I think they do like uh
  • 00:27:21
    nine figures and ARR like it's a big
  • 00:27:24
    business um it's probably going to look
  • 00:27:28
    more similar to this uh than like when
  • 00:27:31
    you log into Sentry you know so I think
  • 00:27:33
    that what's going to happen is um you
  • 00:27:37
    can create sort of an MVP of this code
  • 00:27:40
    could be better and then it could evolve
  • 00:27:42
    into something that looks more like a
  • 00:27:43
    century yeah or they acquire it might
  • 00:27:46
    just as well right if you've built all
  • 00:27:47
    the knowledge around how to set this up
  • 00:27:49
    and that to me is is why AI is so
  • 00:27:51
    interesting like a lot of companies they
  • 00:27:53
    have their staff and if they're lucky
  • 00:27:55
    they have people in there who understand
  • 00:27:57
    AI but the people who really dive into
  • 00:27:59
    AI who listen to every AI podcast who
  • 00:28:01
    are always just playing with the models
  • 00:28:04
    like they are very interesting for a
  • 00:28:05
    company to ACTA higher at some point so
  • 00:28:07
    if you're building an AI based business
  • 00:28:10
    you could potentially just kind of put
  • 00:28:12
    it on people's radar and that could that
  • 00:28:14
    that could be your job application if
  • 00:28:16
    you wanted to right if you're if you're
  • 00:28:17
    not into like building your own thing or
  • 00:28:19
    if you want to take this as a stair step
  • 00:28:21
    towards finding a place at this company
  • 00:28:24
    I could easily see somebody even just
  • 00:28:26
    trying to get this on the road with an
  • 00:28:28
    example project to be an interesting
  • 00:28:30
    potential hire or an intern or whatever
  • 00:28:33
    for a company like this right you just
  • 00:28:34
    show initiative in a field where you
  • 00:28:36
    know they're going to go but they're
  • 00:28:38
    kind of slower than you are because you
  • 00:28:40
    can do whatever you want because you
  • 00:28:41
    have a lot of free time and they have
  • 00:28:42
    their existing product to maintain and
  • 00:28:45
    slowly roll out beta features like this
  • 00:28:47
    right so it definitely this is a good
  • 00:28:50
    time to build these things to see where
  • 00:28:51
    it goes they might buy it they might buy
  • 00:28:54
    it to just remove competition for their
  • 00:28:55
    own right there's all there's already a
  • 00:28:57
    lot of potential in just even following
  • 00:28:59
    this idea through so I hope somebody is
  • 00:29:01
    registering that domain I'm going to
  • 00:29:03
    check like if that domain is not
  • 00:29:04
    registered like 20 minutes after this
  • 00:29:06
    this episode airs that somebody's really
  • 00:29:07
    missing out on an opportunity and I'm
  • 00:29:09
    gonna get it so there you go exactly
  • 00:29:12
    that we do we do 20 minute grace period
  • 00:29:15
    but after 20 minutes it's yours I mean
  • 00:29:18
    look at this this could be this could be
  • 00:29:19
    a business idea all in itself like you
  • 00:29:21
    you you constantly poan would be the
  • 00:29:23
    perfect data source for this like the
  • 00:29:25
    moment an episode comes out I try to
  • 00:29:26
    transcribe it as fast as I can the idea
  • 00:29:28
    behind podan often takes me 3 to 20
  • 00:29:30
    minutes depending on you know how
  • 00:29:32
    important it is and then I I push the
  • 00:29:34
    full transcript of the episode uh in in
  • 00:29:36
    a fire host web hooks to all my um all
  • 00:29:38
    my users if they want to so they could
  • 00:29:40
    have access to every single podcast's
  • 00:29:42
    transcript with all the URLs in it
  • 00:29:45
    immediately after they come out somebody
  • 00:29:47
    could build a service to automatically
  • 00:29:48
    register every single URL that is
  • 00:29:50
    mentioned on any podcast out there
  • 00:29:52
    immediately within minutes might be
  • 00:29:54
    expensive but might also create a lot of
  • 00:29:56
    opportunity to then resell it right
  • 00:29:58
    that's the the kind of stuff building on
  • 00:29:59
    top of of uh data layers data apis that
  • 00:30:03
    today is such a such an interesting
  • 00:30:04
    potential for all of this but that's a
  • 00:30:06
    different idea that I just came up with
  • 00:30:07
    so let's just ignore this for a well I
  • 00:30:09
    just want to build on that idea there's
  • 00:30:10
    a business called
  • 00:30:13
    ungr uh UNG grab. Co I think is the
  • 00:30:15
    website where I think it's every day
  • 00:30:18
    they find like these UNG grab domains
  • 00:30:20
    and they send two emails a day and they
  • 00:30:22
    sell it for between like a hundred to
  • 00:30:24
    let's say $500 so they're buying it for
  • 00:30:26
    $9 let's say and they're selling it for
  • 00:30:28
    a few hundred so this idea is like built
  • 00:30:33
    on top of pod scan like y I love it I
  • 00:30:35
    think it's a great idea going back to
  • 00:30:37
    your original idea uh and and before we
  • 00:30:40
    move to your next idea yes
  • 00:30:43
    um if you were building this the Sentry
  • 00:30:46
    competitor let's call it
  • 00:30:48
    um how would you think about pricing
  • 00:30:52
    model and and you know yeah how would
  • 00:30:54
    you think about pricing model how would
  • 00:30:55
    you think about
  • 00:30:58
    had a charge for it yeah I I've been
  • 00:31:00
    thinking about this too because like the
  • 00:31:02
    moment you want to sell this to somebody
  • 00:31:03
    like an Enterprise customer it it
  • 00:31:05
    probably becomes less of a SAS where
  • 00:31:07
    they allow you to look into their GitHub
  • 00:31:10
    repository and it becomes something like
  • 00:31:12
    you have to sell an un premise version
  • 00:31:14
    of the thing that you offer like they
  • 00:31:16
    have to have their own GPU accelerated
  • 00:31:19
    servers they have to have their own
  • 00:31:21
    h100s or whatever GPU and you need to do
  • 00:31:23
    your AI work in the background and it
  • 00:31:25
    becomes more complicated so maybe that's
  • 00:31:27
    that's a future stage of the business
  • 00:31:28
    that we're not going to look at for this
  • 00:31:30
    but for the for the beginning I would do
  • 00:31:32
    this on a almost a per uh per connected
  • 00:31:35
    repository basis and then the question
  • 00:31:37
    is well how much cost does running this
  • 00:31:41
    agent whatever that may look like incur
  • 00:31:44
    right on the on the levels that I kind
  • 00:31:45
    of pointed out it starts with doing just
  • 00:31:48
    text analysis which can use static
  • 00:31:51
    analysis tools that already exist in all
  • 00:31:53
    kinds of languages programming languages
  • 00:31:55
    and then might take the output of those
  • 00:31:58
    tools into an llm to come up with
  • 00:32:00
    potential ideas and then takes the full
  • 00:32:03
    context of your code base plus those
  • 00:32:05
    ideas to create new code and that is
  • 00:32:07
    already a lot of computation right
  • 00:32:09
    that's a lot of GPU that goes into all
  • 00:32:11
    these many steps and the moment you go
  • 00:32:13
    into creating the code running up
  • 00:32:16
    spinning running it spinning up
  • 00:32:17
    instances on AWS or something it gets
  • 00:32:20
    very expensive so you might want to
  • 00:32:22
    shift the cost of actually testing these
  • 00:32:25
    things of of running like the the non
  • 00:32:27
    GPU accelerated stuff to the customer by
  • 00:32:30
    having them create stuff you create
  • 00:32:33
    instances or have um their their
  • 00:32:36
    AWS Am keys or something in the tool I
  • 00:32:39
    don't really know I would I would
  • 00:32:41
    definitely try to remove the
  • 00:32:43
    computational cost to the people who
  • 00:32:46
    already have um the the budget to spin
  • 00:32:51
    up testing instances and staging systems
  • 00:32:53
    and stuff like that what the business
  • 00:32:55
    itself would do would be the the GPU
  • 00:32:57
    accelerated calculation like the AI work
  • 00:33:00
    right the prompts the the context
  • 00:33:02
    building maybe the embedding all of
  • 00:33:04
    these things that's what the SAS itself
  • 00:33:06
    would do that's what they would pay for
  • 00:33:07
    on a either a per repository business or
  • 00:33:10
    maybe on a Cadence uh sorry per
  • 00:33:13
    repository price or maybe on a on a like
  • 00:33:16
    how often a day do I try to run new
  • 00:33:18
    experiments kind of cadence so you can
  • 00:33:21
    have a once a day I'm looking for the
  • 00:33:23
    biggest problem going to try and solve
  • 00:33:24
    it and send a pull request right it's
  • 00:33:26
    probably going to cost me I don't know
  • 00:33:28
    $20 in expense so over a month that's uh
  • 00:33:32
    $600 is going to cost you like 1,000
  • 00:33:35
    bucks right that that could be uh for
  • 00:33:37
    for a company that wants to have this
  • 00:33:41
    that that could be a pretty good price
  • 00:33:43
    for somebody who's constantly working on
  • 00:33:45
    this or once a day and that kind of
  • 00:33:47
    scales up depends depends really on the
  • 00:33:50
    I mean there's there's a couple ways
  • 00:33:51
    about doing you know doing pricing just
  • 00:33:52
    listening to that one is you you look at
  • 00:33:55
    the costs and then you're like okay
  • 00:33:56
    here's the cost let me go
  • 00:33:58
    and uh add 60% margin or whatever to it
  • 00:34:02
    to create a sustainable business and the
  • 00:34:04
    second way which you actually you talked
  • 00:34:06
    about in your first idea I think which
  • 00:34:08
    is what is the ROI of this thing and
  • 00:34:11
    then how do you just anchor to the ROI I
  • 00:34:14
    mean if if it runs all the time you
  • 00:34:16
    literally have three full-time
  • 00:34:18
    developers working on your code base so
  • 00:34:21
    three shifts of 8 hours of full-time
  • 00:34:23
    developer
  • 00:34:24
    24/7 um so you are replacing somewhere
  • 00:34:28
    north of $200,000 in uh just fees that
  • 00:34:33
    this person would cost you in a year
  • 00:34:34
    right so you could also anchor it on on
  • 00:34:36
    the the the job to be done value like
  • 00:34:39
    the job to be done would be to have
  • 00:34:40
    three people work on your code base all
  • 00:34:42
    day long if that's what the tool does
  • 00:34:44
    then you know that's that's what the
  • 00:34:46
    price might be although I guess 300K a
  • 00:34:48
    year for a single tool that is quite the
  • 00:34:52
    expense right so it would really have to
  • 00:34:55
    have a higher impact than what three
  • 00:34:56
    developers could do with that time but
  • 00:35:00
    this this also depends on the the uh the
  • 00:35:03
    potential um capabilities of that agent
  • 00:35:07
    right if that agent really can try
  • 00:35:10
    wildly different things not just write
  • 00:35:12
    new code and maybe run it but maybe spin
  • 00:35:15
    up a different kind of tool right you're
  • 00:35:17
    you're using I don't know rabbit mq as a
  • 00:35:19
    message queue between your servers but
  • 00:35:21
    the thing is trying to experiment with
  • 00:35:23
    with a an AWS Kafka Q or something and
  • 00:35:27
    just changes some part of your
  • 00:35:28
    infrastructure just to see if that would
  • 00:35:31
    increase performance but also reduce
  • 00:35:32
    cost like if that tool could have the
  • 00:35:35
    agency to do this this could save you
  • 00:35:37
    millions in a year so you know like that
  • 00:35:40
    the question is what is the potential
  • 00:35:42
    impact what's the potential expense and
  • 00:35:44
    how can you find a price that people are
  • 00:35:46
    willing to pay and for this you probably
  • 00:35:48
    want to start small enough with you know
  • 00:35:50
    just static analysis error analysis
  • 00:35:52
    linting that kind of thing and for that
  • 00:35:54
    you probably want to be in the like $50
  • 00:35:56
    to $100 a month kind of bucket but the
  • 00:35:59
    more capabilities you add you add to
  • 00:36:01
    this the more expensive the computation
  • 00:36:04
    and maybe also the more risky the
  • 00:36:06
    experiments get the more you have to
  • 00:36:08
    charge because somebody has to reain
  • 00:36:09
    this in I mean that's always the problem
  • 00:36:11
    with like super intelligence right you
  • 00:36:13
    never know is that thing that I'm
  • 00:36:14
    building here this this silent
  • 00:36:16
    background processing tool trying to
  • 00:36:18
    implement a back door into the system so
  • 00:36:21
    it can take it over when the AI
  • 00:36:22
    Revolution is happening I don't know
  • 00:36:24
    right you have to do that kind of work
  • 00:36:26
    as well you have to do like human peer
  • 00:36:28
    reviews and quality checks and Sanity
  • 00:36:31
    checks with that kind of stuff but I
  • 00:36:33
    would start with like 50 bucks a month
  • 00:36:34
    for some error tracking or common error
  • 00:36:38
    retrieval and automatic pull request
  • 00:36:41
    saving and go from there again
  • 00:36:43
    bootstrapping right iterative design
  • 00:36:45
    right fair enough yeah all right I think
  • 00:36:48
    we have time for one last idea oh man
  • 00:36:51
    I've let me let me take
  • 00:36:56
    uh okay I'm you talked about gatekeeping
  • 00:36:58
    earlier I think that's that's something
  • 00:37:00
    that I had on my mind as well like you
  • 00:37:02
    said you don't gatekeep and I love this
  • 00:37:04
    because you know entrepreneurs should be
  • 00:37:06
    very empowering to each other and not
  • 00:37:07
    prevent each other from doing stuff the
  • 00:37:09
    other way around we should help each
  • 00:37:10
    other to do this but I think gatekeeping
  • 00:37:12
    is something that has been on my mind a
  • 00:37:14
    lot particularly with how Twitter has
  • 00:37:16
    changed over the last couple years do
  • 00:37:18
    you remember when you actually had
  • 00:37:19
    agency over your Twitter feed like where
  • 00:37:22
    you followed people and you would
  • 00:37:23
    actually see what they tweeted the the
  • 00:37:25
    good old days of Twitter like I think
  • 00:37:26
    this happened on every single social
  • 00:37:28
    media Network you kind of switched from
  • 00:37:30
    the historical timeline to the
  • 00:37:31
    algorithmic timeline where now what you
  • 00:37:34
    see is not what you want to see but it's
  • 00:37:36
    what the platform wants you to see for
  • 00:37:38
    many reasons right for advertising
  • 00:37:39
    reasons for engagement reasons and all
  • 00:37:41
    of this so you cannot trust the
  • 00:37:43
    platforms that you operate on and you
  • 00:37:46
    you cannot trust that even what
  • 00:37:47
    aggregators give you is the thing that
  • 00:37:50
    you need you only get what people think
  • 00:37:52
    you need so what I would suggest or what
  • 00:37:56
    I would want to exist and I see this
  • 00:37:57
    already sorry for bringing up hotan all
  • 00:38:00
    the time because it's it's my baby it's
  • 00:38:01
    my project right now and I see in the I
  • 00:38:04
    have a couple users who use potan in one
  • 00:38:06
    particular way and this is kind of where
  • 00:38:08
    I see this going they have chosen a
  • 00:38:10
    niche that they operate in like like any
  • 00:38:13
    particular industry I have a couple
  • 00:38:14
    people who are in one specific sub
  • 00:38:16
    industry of the medical world they they
  • 00:38:19
    uh have a particular kind of doctor that
  • 00:38:22
    they want to serve and what they offer
  • 00:38:24
    them is a data aggregation and
  • 00:38:26
    summarization for every new thing that
  • 00:38:29
    happens in the industry like doctors
  • 00:38:31
    don't have the time to listen to 20
  • 00:38:32
    podcasts every single day where other
  • 00:38:34
    doctors like them talk about the things
  • 00:38:36
    that they did they just cannot have this
  • 00:38:37
    time right they are in the operating
  • 00:38:39
    room they are in in the the clinic or
  • 00:38:41
    whatever they have to do doctor stuff
  • 00:38:43
    and they want to know what's going on in
  • 00:38:44
    the world so they can figure out what
  • 00:38:45
    new things to learn so what I found is
  • 00:38:48
    people who are themselves overlapping
  • 00:38:51
    experts like they are experts in that
  • 00:38:53
    Niche and also entrepreneurial they
  • 00:38:55
    start aggregating the Building Systems
  • 00:38:57
    to to aggregate information from that
  • 00:38:59
    Niche and present it to less technical
  • 00:39:01
    people in that niche as a summary as a
  • 00:39:03
    newsletter as a podcast as a a blog post
  • 00:39:07
    or whatever so take this idea of this
  • 00:39:10
    person doing this with my podcasts and
  • 00:39:12
    expand it into any other industry and
  • 00:39:15
    any other medium I think that's the
  • 00:39:16
    important part like you can scrape and
  • 00:39:18
    collect and extract as much data as you
  • 00:39:20
    can for people in your Niche from
  • 00:39:21
    sources that are relevant to them you
  • 00:39:24
    have all these real-time data streams
  • 00:39:25
    API news sites and social feed
  • 00:39:27
    influencers what they are talking about
  • 00:39:29
    also interesting yeah you can parse
  • 00:39:31
    newsletters you can even like take
  • 00:39:32
    photos of magazines and OCR them like
  • 00:39:34
    the old school stuff like there's horse
  • 00:39:36
    magazine there's a lot of like horse
  • 00:39:39
    Farmers the area where I'm at right
  • 00:39:41
    because live in the countryside and
  • 00:39:43
    people have horses like they also want
  • 00:39:44
    to know what's going on in horse world
  • 00:39:46
    like in big horse what is big horse
  • 00:39:48
    doing today so you can grab this
  • 00:39:51
    information aggregate it and push it out
  • 00:39:52
    to them you can summarize it and this is
  • 00:39:54
    the Twist on the per subscriber level
  • 00:39:57
    you know that like person a really wants
  • 00:39:59
    really dense really uh
  • 00:40:02
    just overview bullet points kind of
  • 00:40:05
    things to figure out what's going on
  • 00:40:06
    with a link to the thing to do their own
  • 00:40:08
    research person B might want to have a
  • 00:40:10
    couple paragraphs per thing that is
  • 00:40:12
    interesting and person C has a lot of
  • 00:40:14
    time just not enough time to listen to
  • 00:40:16
    the stuff but enough time to read like a
  • 00:40:17
    2,000-word email so for every single
  • 00:40:20
    person that has a priority in how they
  • 00:40:22
    consume things you can summarize it to
  • 00:40:25
    the exact level of what they need and
  • 00:40:26
    present it to to them and maybe in the
  • 00:40:28
    future you can even deliver it in the
  • 00:40:30
    medium that they most prefer like they
  • 00:40:32
    want to have a fake podcast conversation
  • 00:40:35
    between two experts in the field that
  • 00:40:36
    just do like a oh and this happened in
  • 00:40:38
    the world today what do you think of
  • 00:40:39
    this and that and then they do this
  • 00:40:41
    weird Spiel where they talk like they
  • 00:40:44
    are super infused but they're both AIS
  • 00:40:46
    talking to each other some people like
  • 00:40:47
    this I've seen this uh like what is it
  • 00:40:50
    like with the Google thing that that was
  • 00:40:52
    recently notbook LM yeah like people
  • 00:40:55
    people like fake
  • 00:40:58
    conversations because people like
  • 00:40:59
    scripted conversations in the first
  • 00:41:01
    place that's that's all of movies and TV
  • 00:41:03
    really right like every reality show is
  • 00:41:05
    this and people enjoy it so why not
  • 00:41:07
    bring this Spirit into the the medium
  • 00:41:09
    through which people consume information
  • 00:41:11
    or you send a really short newsletter de
  • 00:41:13
    James Clear style three bullet points
  • 00:41:15
    that's it or what else um you could have
  • 00:41:18
    a news story like podcast where somebody
  • 00:41:20
    acts like they're reading the news and
  • 00:41:22
    it's just the stuff that you care about
  • 00:41:24
    in a tone and a voice that you like um
  • 00:41:26
    yeah a video you can make an automated
  • 00:41:28
    video an AI generated video with screen
  • 00:41:29
    grabs of the things like the articles
  • 00:41:31
    that people have posted or pictures that
  • 00:41:33
    they shared on social media and just
  • 00:41:35
    make that a three minute YouTube style
  • 00:41:37
    hyper action Mr Beast video right you
  • 00:41:39
    could do all of these things in the
  • 00:41:40
    medium for the person consuming it and
  • 00:41:43
    all you need to do is to take data
  • 00:41:44
    agregate it summarize it then shape it
  • 00:41:46
    into something that people want to use I
  • 00:41:48
    see this happen on so many platforms
  • 00:41:50
    that people take the whole big world of
  • 00:41:53
    data that nobody has the capacity to
  • 00:41:55
    understand and just crunch it down into
  • 00:41:57
    bite-sized whatever and present it to an
  • 00:41:59
    audience that is really thirsty for not
  • 00:42:02
    having to do the work but getting all
  • 00:42:04
    the good results so that's idea number
  • 00:42:06
    three trust the data platform I think
  • 00:42:09
    uh if I was like 21 and I'm trying to
  • 00:42:13
    create a startup and make $10,000 a
  • 00:42:16
    month I think I do an idea like this um
  • 00:42:19
    I don't think it's particularly
  • 00:42:21
    hard um of course it's hard to like find
  • 00:42:24
    pick the niche right like horses like
  • 00:42:27
    that actually might be a great Niche
  • 00:42:29
    people right um so I think picking the
  • 00:42:32
    niche as long as you can pick the niche
  • 00:42:34
    and you can reliably create the you know
  • 00:42:36
    curate the content um in a way I think
  • 00:42:40
    that yes you can give away 99% of it and
  • 00:42:43
    you can also put a pay well for $9 a
  • 00:42:45
    month and start charging for it I uh I
  • 00:42:48
    was just thinking about this today this
  • 00:42:50
    this kind of similar idea
  • 00:42:53
    because someone on my
  • 00:42:55
    team he
  • 00:42:58
    like a big part of his job is to curate
  • 00:43:01
    all the most interesting news in
  • 00:43:04
    Innovation AI just like new technologies
  • 00:43:07
    and he writes a report and he posts it
  • 00:43:09
    to our slack to LCA which is our
  • 00:43:11
    Innovation design firm he basically says
  • 00:43:14
    these are the five things that you you
  • 00:43:17
    you need to do today you need to read
  • 00:43:18
    today um this is what matters just to
  • 00:43:21
    keep the team like up to date um and
  • 00:43:25
    it's probably one of the most valuable
  • 00:43:26
    pieces is a Content I read on a daily
  • 00:43:28
    basis if you read that like you're good
  • 00:43:31
    to go but what he's presenting to
  • 00:43:35
    designer should be different than what
  • 00:43:36
    he's presenting to Engineers
  • 00:43:37
    realistically exactly right yeah it has
  • 00:43:40
    the same Source though right it's the
  • 00:43:41
    exact same Source material it just has
  • 00:43:44
    to understand what the other person
  • 00:43:46
    needs how they speak how they ingest
  • 00:43:48
    information if you if you tell a story
  • 00:43:50
    to somebody who's a writer they will
  • 00:43:52
    take it in so differently than to
  • 00:43:54
    somebody who is super technical and does
  • 00:43:57
    not have like an inch of imagination
  • 00:44:00
    like they they just they need like the
  • 00:44:02
    the facts or whatever and the writer can
  • 00:44:04
    deal with pros and deal with like
  • 00:44:05
    abstract constructs constructs and all
  • 00:44:08
    that knowing who you talk to and how to
  • 00:44:10
    talk to them that is the actual magic
  • 00:44:12
    here which is why I say this works best
  • 00:44:15
    for people that are at the intersection
  • 00:44:17
    of already being either a dabbler or an
  • 00:44:19
    expert in that field like you want
  • 00:44:21
    somebody who has been grown up around
  • 00:44:23
    horses to build the horse information
  • 00:44:25
    channel right somebody who actually
  • 00:44:27
    knows how to horse like they will speak
  • 00:44:31
    the language of horse people to other
  • 00:44:33
    horse people and be able to prompt AIS
  • 00:44:35
    better like write better copy for their
  • 00:44:37
    own marketing have something relatable
  • 00:44:40
    for a sales conversation have something
  • 00:44:42
    relatable for building that thing in
  • 00:44:43
    public in front of their audience right
  • 00:44:45
    hey I'm building this other horse people
  • 00:44:47
    that I've already connected with because
  • 00:44:49
    I am a horse person right obviously
  • 00:44:51
    there's a massive benefit in being part
  • 00:44:53
    of the niche that you Target so I guess
  • 00:44:55
    that's that's the trick the trick is to
  • 00:44:57
    understand what niches you already are a
  • 00:44:59
    part of and how you can leverage your
  • 00:45:01
    existing amateur or expert status either
  • 00:45:03
    way it's better than being a complete
  • 00:45:05
    novice right like in most niches that
  • 00:45:07
    you've not been part of before and turn
  • 00:45:09
    that into a business that can help other
  • 00:45:11
    people save time it's really what this
  • 00:45:12
    is another reason I really like this
  • 00:45:14
    idea and people who listen to the Pod
  • 00:45:15
    know this about me is it's a it's a
  • 00:45:18
    wedge that could evolve into something
  • 00:45:20
    else so if you build the the horse
  • 00:45:24
    publication personal publication
  • 00:45:27
    step one and it starts taking off step
  • 00:45:29
    two could easily be be building sass for
  • 00:45:31
    that Community oh for sure there's a lot
  • 00:45:33
    of Logistics in horses right whenever
  • 00:45:36
    people go to fairs or whatever their
  • 00:45:37
    horses have to go and the trailer needs
  • 00:45:39
    to be there or new horses or vets need
  • 00:45:41
    to be in contact with them or like they
  • 00:45:44
    they need to get their medicine like I'm
  • 00:45:45
    I'm exposed to this stuff and whenever I
  • 00:45:47
    see people dealing with with this kind
  • 00:45:49
    of livestock there are thousands of
  • 00:45:51
    little things that need to be dealt with
  • 00:45:54
    that all kind of would be completely
  • 00:45:58
    invisible to somebody outside of the
  • 00:45:59
    industry so the moment you have that's
  • 00:46:02
    the thing like if you aggregate all this
  • 00:46:03
    information about horses you could have
  • 00:46:05
    easily an AI that just checks for
  • 00:46:08
    problems in the horse right you can have
  • 00:46:10
    ai that runs in the background as an
  • 00:46:12
    agent that just checks every single
  • 00:46:14
    newscast that it finds for did somebody
  • 00:46:16
    mention a problem that they might have
  • 00:46:18
    do I have a potential idea on how I
  • 00:46:20
    could solve this and present that to you
  • 00:46:22
    or to your customers like here's today's
  • 00:46:24
    five biggest problem in big horse and
  • 00:46:27
    then you know just just what you're
  • 00:46:29
    doing here really here are ideas on how
  • 00:46:31
    I can make these things and turn them
  • 00:46:33
    into a business I I think there's
  • 00:46:36
    there's a lot to be like the in between
  • 00:46:37
    a transmission between a lot of data and
  • 00:46:39
    people with a need right and that's
  • 00:46:41
    that's what this business is I'm glad
  • 00:46:43
    you like the idea dude I would tell you
  • 00:46:45
    if I didn't like it you know on the pot
  • 00:46:47
    I say do I you know do I sip the idea or
  • 00:46:50
    do I spit the idea sip is you like it
  • 00:46:52
    spit is you don't like it and I
  • 00:46:54
    definitely SI that idea so thank you I S
  • 00:46:57
    I actually sipped all your ideas you
  • 00:46:59
    brought the fire I'm I'm feeling pretty
  • 00:47:01
    like my creative juices are flowing I'm
  • 00:47:03
    feeling pretty good right now um if you
  • 00:47:05
    made it to this part of the podcast good
  • 00:47:08
    for you yeah
  • 00:47:10
    um that's awesome um but I have a a
  • 00:47:14
    quick ask go and like this on YouTube
  • 00:47:17
    because then more people will see this
  • 00:47:19
    video um that's right don't gatekeep
  • 00:47:21
    this video people like bookmark my
  • 00:47:23
    tweets they don't like my tweets they
  • 00:47:25
    they don't they don't want to like it on
  • 00:47:27
    YouTube because they're afraid the ideas
  • 00:47:29
    are going to get out there I'm asking
  • 00:47:30
    you please like comment subscribe I
  • 00:47:33
    appreciate it and Arvid where could
  • 00:47:34
    people get to know you more and and what
  • 00:47:38
    you're up to I appreciate you asking um
  • 00:47:41
    people can find me on Twitter because
  • 00:47:42
    that's where I hang out all day long uh
  • 00:47:44
    it used to be called Twitter you
  • 00:47:45
    probably know what it's called now you
  • 00:47:47
    can find me there at avitar a rvid Kahl
  • 00:47:50
    and then I guess podan is the big thing
  • 00:47:52
    that I'm working on I'm building this in
  • 00:47:53
    public too like every day I post like
  • 00:47:55
    how how much I suck as a developer veler
  • 00:47:57
    in trying to build this this product and
  • 00:47:59
    I post about what I'm trying to do how I
  • 00:48:01
    get my customers how I become profitable
  • 00:48:04
    that's at podc scan. FM and if if you
  • 00:48:06
    want to build something on top of what
  • 00:48:08
    is now probably 15 million transcribed
  • 00:48:11
    podcast episodes or um any business
  • 00:48:15
    right obviously on top of this data I
  • 00:48:17
    talked about the fire host the API or if
  • 00:48:19
    you just are a person that wants to
  • 00:48:21
    track mentions of your name or your
  • 00:48:23
    brand on 2.5 million English speaking
  • 00:48:26
    podcasts I scan every single day which
  • 00:48:28
    is kind of crazy still can't believe
  • 00:48:30
    that that's actually a thing that I was
  • 00:48:31
    able to build yeah podan FM and yeah
  • 00:48:34
    check me out on Twitter I think that's
  • 00:48:35
    that's where you can reach me my DMs are
  • 00:48:36
    open so you can annoy me or enlighten me
  • 00:48:39
    with any ideas that you might have any
  • 00:48:41
    criticism or cheering on is always
  • 00:48:43
    welcome appreciate it it really is an
  • 00:48:45
    interesting data set like the podcast
  • 00:48:47
    data set is really interesting like I'm
  • 00:48:48
    going to think about some ideas uh that
  • 00:48:51
    I could be building um and by the way in
  • 00:48:53
    the comment section comment because you
  • 00:48:56
    might find your your human co-founder
  • 00:48:59
    that's right not that you might need one
  • 00:49:01
    but I noticed that some people in the
  • 00:49:02
    YouTube comments have been like chatting
  • 00:49:04
    and saying Hey I want to go build up and
  • 00:49:06
    build something together so um yeah
  • 00:49:09
    maybe maybe you go build something on
  • 00:49:10
    top of
  • 00:49:11
    podcasts um Arvid you're I want you to
  • 00:49:15
    come back on you know on the pod in 2025
  • 00:49:18
    this has been this has been amazing
  • 00:49:20
    appreciate it and I'll see you on the
  • 00:49:22
    internet well thanks so much yeah I'll
  • 00:49:24
    be back I have a couple more ideas
  • 00:49:25
    probably then even even even more than
  • 00:49:27
    now thanks so much man see you all right
  • 00:49:30
    take care
  • 00:49:32
    [Music]
Tags
  • AI agents
  • business ideas
  • AI co-founders
  • startups
  • automation
  • market opportunity
  • data aggregation
  • software optimization
  • AI impact
  • entrepreneurship