24:44orenmeetsworld · Talking Head
How to Be a 1-Person Marketing Machine in 2026
A 25-minute whiteboard argument that this summer is the last window to build a lean, compounding marketing operation before saturation closes the gap.
June 6thThe argument in one line.
AI has already crossed from a one-to-one chatbot into a one-to-many orchestration layer, and the people who thrive will be those who manage systems of autonomous agents rather than those who type better prompts.
Read if. Skip if.
READ IF YOU ARE…
- You use AI daily but still think of it as a smarter Google — you ask, it answers, you close the tab.
- You work in a mid-sized company and want to understand what AI adoption actually looks like at the Fortune 500 level before it reaches you.
- You are early-career and genuinely worried about whether your role will exist in five years.
- You are an entrepreneur or small-team operator curious about building AI agent workflows without a technical background.
- You have heard the term AI agents but have not yet built or used one.
SKIP IF…
- You are a seasoned AI practitioner already running multi-agent systems — the frameworks here will feel introductory.
- You want hard technical implementation detail; this is a conceptual and motivational conversation, not a tutorial.
The full version, fast.
Ali K. Miller argues that the dominant productivity framing of AI will trap most workers and companies in a dead end within five years. The real unlock is transformation: using AI to reinvent workflows, overcome fear, and build systems that work autonomously for hours. She tracks three capability benchmarks (coding, autonomy scale, self-learning), runs 34 named AI agents built as plain files in Claude Code, and gives a concrete framework for career-proofing: become a systems thinker who operates at level 4-5 ownership. The job displacement risk is real but uneven — high-liability roles are safer, and the winning move is shrinking your team while expanding your agent fleet.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Who's talking.
01:30guestAli K. Miller
01:08hostKate Mackz
Where the time goes.
00:00 – 01:30
01 · Intro and hook
Show cold open, subscribe CTA, guest intro.
01:30 – 03:01
02 · Can AI make you the best version of yourself?
The dual-use nature of AI: transformation tool or crutch. Start with knowing what you want.
03:01 – 05:46
03 · Ali's AI journey and when AI became inevitable
Fortune 500 advising, AI labs, posting daily for a decade. The tidal wave realization 11 years ago.
05:46 – 08:57
04 · The assumptions that keep changing and the 3 benchmarks to watch
8-year prediction spreadsheet analyzed in Claude Code. Coding (SWE-bench: 94), autonomy scale (METR), self-learning.
08:57 – 13:15
05 · What most people get wrong — and how to fix it
Still in chatbot mode. Mindset shift: from eliciting knowledge to chaining actions. Proactive task chains for a podcaster.
13:15 – 18:06
06 · Using AI to overcome fear and protect your privacy
The fear-busting tool. What you should not share. 30-day data retention reality. LM Studio for offline use.
18:06 – 24:00
07 · How different cities think about AI and whether trust affects innovation
SF vs. NYC vs. Sydney. US low trust globally. 3-year enterprise adoption lag even without further AI progress.
24:00 – 29:04
08 · Productivity vs. transformation
Central thesis: productivity framing traps companies. Benchmarks that matter: rewarding jobs, net-new workflows, human joy.
29:04 – 33:27
09 · Will AI take our jobs?
Overlap does not equal displacement. High-liability roles safer. GDP-val at 80%+. Small teams learn faster.
33:27 – 38:10
10 · How to future-proof your career
Proactivity pyramid levels 1-5. Hiring only at level 4+. Early-career advice: show trade-off thinking and ownership.
38:10 – 43:52
11 · The skills that matter most and the agent era
Artistic vision, adaptability, communication, tech intelligence — plus systems thinking. Job titles dissolving. Ali runs 34 agents named after The Office and Friends, all just files.
43:52 – 49:38
12 · Women and AI adoption
Women adopt 25% less but scientific research shows they are better at building AI agents. Verbal skills = prompting advantage.
49:38 – 55:16
13 · The wrong way to think about productivity
Transform the quality of your process, not just the volume. Review emails from 10 angles; dictate workout plans; live-coach yourself through meetings.
55:16 – 58:36
14 · Analog life and AI balance
Phone detox retreats with AI-industry friends. Walking while dictating to Claude. The more sci-fi the job, the more analog the free time.
58:36 – 1:02:36
15 · AI video and what the next 5 years look like
AI video length doubling yearly — short films within 5 years. Viscosity problem still detectable. Software faster than expected; hardware slower.
1:02:36 – 1:07:35
16 · Final advice and wrap
Apply existing life advice to AI: set goals, add full context, run a 3-hour planning session. Where to follow Ali.
Lines worth screenshotting.
- AI autonomy is doubling every five months — by early next year, systems will reliably complete a full day of human work unsupported.
- Only 6% of employees have used an AI agent, meaning anyone who has is already ahead of 94% of the workforce.
- Productivity-only AI adoption is a trap: companies that only ask people to write five blog posts instead of one will have a retention and morale crisis within three years.
- A 34-agent digital workforce is just 34 files on a laptop — personality, tools, memory, and delegation chains.
- Women outperform men on verbal and written communication in all 75 countries studied, which is the primary skill AI systems reward — yet women adopt AI 25% less.
- All inputs sent to third-party AI providers live on external servers for a minimum of 30 days; opting out of data sharing is a one-toggle change most people have never made.
- Job overlap with AI does not equal job displacement — liability density is the real moat.
- The METR autonomy benchmark shows AI now scores 80% reliability on tasks that take a human over twelve hours; that number doubles every five months.
- The best AI sessions happen on a one-hour walk dictating into Claude, not staring at a screen.
- Claude scored 94 on the SWE-bench software benchmark; two years ago that number was 17.
- Level 4 ownership means identifying a problem, proposing solutions with trade-offs, and committing to one before being asked.
- Asking AI the next three things you were going to do with its answer and delegating those three things is the single mindset shift from chatbot to agent mode.
- Software timelines are consistently underestimated; hardware timelines are consistently overestimated — AI video at feature-film length is within five years, robotics is decades out.
- The future of job titles is dissolution into broad buckets — hyper-specialized roles like SEO specialist for Mid-America are already disappearing.
The shift from asking AI to running AI.
WHAT TO LEARN
The gap between people who get real leverage from AI and those who do not comes down to one transition: from treating AI as a smarter search engine to treating it as a system that owns and executes tasks on your behalf.
02Can AI make you the best version of yourself?
- Start every AI session with your outcome, not your question — then ask what it would take to get there.
- Transformation requires knowing what you want first; without a north star, AI just accelerates directionlessness.
04The 3 benchmarks to watch
- AI autonomy is doubling every five months — any workflow built on the assumption that humans must check every step will be over-engineered within a year.
- Coding capability is not just a developer metric — it is how AI talks to other systems and chains actions together.
05What most people get wrong — and how to fix it
- The next three things you were going to do after getting an answer are the real prompt: delegate those steps to the system before you close the tab.
- Connecting AI to Gmail and Calendar is a two-second toggle, not a technical project — the main blocker is knowing it exists.
06Using AI to overcome fear and protect your privacy
- Data you send to any third-party AI provider lives on their servers for at least 30 days — a single toggle opts you out of data-sharing.
- For sensitive work, running a local model via LM Studio keeps everything on your machine with no cloud exposure.
08Productivity vs. transformation
- Productivity framing is the wrong goal; transformation framing — my system is now fundamentally different — is the right one.
- Companies that only optimize for headcount reduction via AI will face a morale and retention crisis as workers recognize they have been given less meaningful jobs.
09Will AI take our jobs?
- High-liability roles carry natural AI moats; roles that overlap heavily with AI output but have no liability floor are the most exposed.
- Small teams learn and adapt faster — being on a small team now is training for the agentic era.
10How to future-proof your career
- The highest-value skill in the AI age is level-4 ownership: finding the problem yourself, analyzing options with trade-offs, committing to one, and executing without being asked.
- Job titles are dissolving into broad functional buckets; the workers who thrive will own entire domains and govern teams of AI sub-agents.
11The skills that matter most and the agent era
- A team of AI agents is just a folder of files describing role, tools, memory, and delegation; anyone can build one without coding.
- The way to start is not 'build one agent that does one task' — it is stepping back to your goals and asking AI to design the organizational structure that pursues them.
12Women and AI adoption
- Women score higher on verbal and communication benchmarks in every country studied; the gender adoption gap is an access and visibility problem, not a capability one.
14Analog life and AI balance
- Restoring analog time is not a productivity hack; it is what makes the AI-heavy parts of work sustainable.
15AI video and what the next 5 years look like
- Software AI progress is consistently faster than predictions; hardware and robotics progress is consistently slower — calibrate your timeline fears and bets accordingly.
Terms worth knowing.
- Autonomy scale
- A benchmark from METR measuring the length of a human task an AI can complete uninterrupted at a given reliability level. Currently doubling every four to five months.
- GDP-val
- A benchmark measuring AI capability on tasks that have direct economic impact across real-world occupations. AI systems currently score above 80% on this measure.
- SWE-bench
- A software engineering benchmark testing AI ability to resolve real GitHub issues. Used as a proxy for overall coding capability; Claude scored 94 in 2026, up from 17 two years prior.
- Multi-agent orchestration
- A system where one primary AI agent delegates tasks to a network of specialized sub-agents, each defined by a file containing its role, tools, memory, and personality.
- Level 4 / Level 5 ownership
- From a five-level proactivity pyramid: level 4 is identifying a problem, proposing solutions with trade-off analysis, and committing to one; level 5 adds having already executed the solution and set up success tracking before being asked.
- LM Studio
- A free desktop application that runs open-source AI models fully locally, keeping all data on the user machine rather than sending it to cloud servers.
- Digital workforce
- A structured fleet of AI agents — each a file with a defined role, memory, and tool access — that operate as a team with delegation chains, analogous to a small company org chart.
Things they pointed at.
29:04linkGDP-val benchmark
34:00bookProactivity pyramid (Alex Lieberman)
Lines you could clip.
24:00
“If you only focus on productivity, you're screwed in five years. There is so much more to be had in the AI age. And transformation — that is the word to actually be going for.”
38:20
“I mean, I could make it sound way cooler than it is. Let's be honest about it. They are fucking files.”
13:00
“Ask yourself the next three things you're going to do with that knowledge anyways, and then go to these systems and say: what would it take for you to complete those three things for me on a schedule every single morning?”
1:00:20
“The more sci-fi my job becomes, the more analog my free time becomes.”
49:10
“The two main ways I talk to AI: yapping and uploading images.”
Where the conversation goes.
01:30 – 24:00denseAI as transformation vs. productivity
05:46 – 10:22denseCapability benchmarks and technical state of AI
10:22 – 38:10denseAgent mode and multi-agent orchestration
25:44 – 38:10denseJob market and career strategy
15:04 – 18:06steadyPrivacy and data security
43:52 – 49:38steadyGender and AI adoption
55:16 – 58:36steadyAnalog life and AI balance
58:36 – 1:02:36steadyAI video and future predictions
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
00:00AI, if used in an awesome way, can help you become the best version of yourself. You can also completely abuse it and become a worse version. The biggest thing in knowing how to use AI to transform and get the life you want.
00:13You first have to know what life you want. Ali k Miller is an AI expert and one of the leading voices helping people understand how artificial intelligence is changing the way we work and live. In this episode, she breaks down how AI can help us become better versions of ourselves and the practical ways we can start using it in everyday life.
00:30Even if AI has a large overlap with a job, it's still not clear whether that is equivalent to displacement. You only focus on productivity, you're screwed in five years. There is so much more to be had in the AI age.
00:43And transformation, that is the word to actually be going for. So many people just jump into this thing, they're like, write my essay for me.
00:50That is not at all a building more essay. The two main ways that I talk to AI, it is
00:55and Do you feel like there is a security issue with some of the stuff that people are putting into these chats? Once you put in that prompt Hi there.
01:09Thank you so much for tuning in to today's episode with Ally Miller. Your support and you showing up here today truly helps us continue bringing you conversations like these.
01:19So please do us a quick favor before we get into it and hit that subscribe button. Support us. We love you so much for it, and let's get into our chat.
01:30Ali, you are an AI expert, and you've said that AI can help us become the best version of ourselves. I think that everybody listening to this podcast right now wants to become the best version of themself.
01:39So break down that statement for us. First, AI, if used in an awesome way, can help you become the best version of yourself. You can also completely abuse it and become a worse version.
01:52So let's just start there. The biggest superpower that I think people can have in their life is knowing what they want. Like, knowing what you want, knowing what you wanna get out of life, knowing your goals, everything falls out after that.
02:05Right? Like, the ability to set your North Star and then go after it with whatever resources you have, including AI. So starting with whatever it is that you want.
02:15And then the second, the advice that I always take for my own life is just like, what would it take? And so anytime that I'm working with an AI system and I want to make a new product or I want to get a new client or I want to launch a new business line or whatever the thing is, I start from the goal or outcome that I want.
02:34And then I talk back and forth with AI. I'm just like, what would it take to be able to get this done for less than $5,000 in under three months with a team of this size, etcetera, and just give all this context on myself.
02:46But the biggest thing in knowing how to use AI to transform and get the life you want, you first have to know what life you want and then being able to use it. But so many people just jump into this thing and they're like, write my essay for me.
02:59That is not at all building the life that you want. And for people that aren't familiar with you and your work, can you give us a little bit of your background and how you ended up in this space? I advise Fortune 500 companies, private financial institutions,
03:11AI Labs on what is going on in AI for enterprises, how to actually use it and transform their business and their culture, uh, for the AI age.
03:23I started in AI almost twenty years ago.
03:26I've worked in it every single day for the last decade. I've posted online almost every single day for the last decade, which sounds positively insane, and it would have been a lot easier if I'd had AI helping me throughout that.
03:38But I am really trying to help transform lives of a billion people into the AI age. And sometimes that's in a personal life transformation or family or hobbies, And sometimes that is a career pivot or someone quitting their job to start their own business.
03:57But it has been a decade of building AI products and then shifting into business advising.
04:04And within that twenty years, at what point did you realize that AI was gonna fundamentally change society? I think so I had started doing ML research in college, like, nineteen years ago. And I think at the time, I was fascinated by it.
04:17But if you had told me that we'd be here in twenty twenty six, I would laugh you out of the room. I think I was fascinated by the combination of science and humanity, and the combo of that just was enough to get me excited by it and this this idea that you could, like, combine qualitative and quantitative.
04:36So twenty years ago, it was an interest, intrigue side. I think starting about a decade ago or, yeah, eleven years ago, I was like, this is the thing that is going to transform our entire world.
04:51It is going to come like a tidal wave. 99% of the world has no idea that this even exists. I'm gonna dedicate my whole life to it.
04:58And so I completely switched and just said, every single thing that I'm gonna put out into the world, every way that I'm gonna think about my own career, it is gonna be with AI in mind.
05:08And this is, like, before the majority of people even knew what AI was. The before, really any single AI pitch even had AI at the core of it, before my parents even knew what I did for a living. And it has just been every single day focused on what is the latest thing, what assumptions do I have that are changing, what is the pace of change, what do I think is coming around the bend, and what do I think 99% or even 20% of business professionals, what do they not know that's coming, and how can I
05:39help them prepare in the best way possible or at least give them options that they can better prepare? I like that you said what are the assumptions that I have that are changing? Because I feel like adaptability is the biggest thing with AI, especially for young professionals.
05:51It's having that skill of be of being adaptable because things are changing so quickly. And I also liked that you said that you never would have predicted that in 2026, this is where we would be with AI as somebody that's been in it for twenty years.
06:03So I'm curious what assumptions have changed for you most recently. So I am a
06:10crazy person when it comes to spreadsheets. We're gonna start there. And for the last eight years, I've shared my AI predictions.
06:18And I keep basically a list of predictions. And every I try for every month, but maybe it's, like, every quarter. I check-in on those predictions, and in every single row, I update what my assumptions are, what my prediction is for when that thing will happen, or how I feel about that topic.
06:35And I can look across the row, and I can get a sense of how my opinions have changed over time. So I can see whether eight years ago, I thought that we would have recursive learning in 2040 and whether I've sped that up or slowed that down.
06:51And so I can put that entire spreadsheet into Claude code, and I can say, analyze all of my predictions over time. What have I sped up on?
07:01What have I slowed down on? What have I changed my opinion on multiple times? What do I seem least confident on and why?
07:07And so I have a data repository of, like, every prediction that I've had. And and publicly, anyone can go into my newsletters and track these over time as well.
07:18I think big assumptions that I've changed, particularly with the last set of models, has been on self learning. So the three big areas that I would have everyone be looking into for AI, the big things that I'd be tracking are, number one, how good is AI at coding?
07:35We have now surpassed the benchmark that most people would be looking out for. So Claude Mythos Preview, the model that Anthropic is keeping to itself, um, and dozens of companies for cybersecurity reasons, that scored a 94 on this big software benchmark.
07:51A year ago, we were at 72, and the year before that, were at 17. Like, complete exponential change on coding benchmarks, coding capabilities. And code is not just, hey.
08:02How much can you help a software engineer? Code is how AI talks to other systems. So it is AI's ability to talk to other systems, connect over here, grab a flight off of Google Flights, search this thing, grab this thing.
08:17That is how solutions are built. It's how AI communicates. So coding is number one.
08:22The second is how autonomous it can be, and that is a factor of how long it can work for uninterrupted and how much humans trust it, which I think if you were to look at both of those stats.
08:37Autonomy scale, we are now depending on which benchmark you're looking at, it can work on its own at about a 50% reliability level for, like, one to a few hours, but it's still less than a day. But we are doubling that length of a human task that it can take on.
08:54Uh, we are doubling that length now about every five months. What is an autonomy scale? Really quick.
08:58Yeah. So, um, there's a benchmark from METR, m e t r, and they are measuring the length of a human task that an AI can take on. So if something takes you thirty minutes, it might take AI five minutes.
09:10It might take it five hours. It doesn't really matter. But how long can it, uh, how long of a task can it handle without you coming in and being like, next step, do this.
09:20So right now, we're tracking this, you know, foundation is tracking all of the state of the art models to see how long of human task it can take on.
09:30And the graph looks like that. We were doubling every six to seven months. We are now doubling every four to five.
09:36There are some people that predict that we're gonna be doubling every three to four. It is a factor of how good the models are, how good the, you know, like, what how do I like, post training, um, the reinforcement learning that they're working on.
09:51It could come down to how good a harness is, like Claude coat. Maybe this is too technical, but it is coming down to a couple things of how fast that doubling is happening. And right now, we have models that can handle tasks at a 50 reliability level for over twelve hours and at an 80% reliability level for, like, one to three hours.
10:17That length of human task is doubling every five months. So if you imagine, probably by the end of this year or by beginning of next year, we will have AI systems that can reliably handle an entire
10:31day's work that a human could do. That is so impressive to me because I feel like and I think this is a really important conversation for for people to hear because it can help so much with just getting your basic job, like jobs done, different thing tasks in your life done by knowing that. I feel like I use AI in a very simple way.
10:48Like, I use it for help prepping me with interviews. I, you know, send it my questions that I've kind of, like, written written out through research. It helps me research people.
10:56You know, x, y, and z. But all I'm doing with AI right now is, like,
11:00kind of talking to it. Like, it's a chatbot. You know?
11:03But you're talking to it. You're already treating it like a weird freaking thing. Like, that mindset shift is already big.
11:08So I wanna give you credit where credit's stupid. But how do you like, what's an example of a task you can give it where it like, how do I get it to work for me for an entire day without talking to it? So so let's say that it is still helping you on podcast related things.
11:25Right? Yeah. So you're not, like, completely changing the way that you're coming to it.
11:30But let's say that you decide to set up and and we'll start easy. Right? Like, if you're going to it for research, every single morning, you should just have it automatically run research for you.
11:40Or if you're coming to it to stalk your potential guests, you should just have it go through your email and every single day look at brand new contacts that you're reaching out to. And it can say, hey.
11:53You recently just started conversations with Ali Miller. Here's a little bit about her, and here's why she might be a really interesting podcast guest. Or it could look at your social profiles, and it could say, hey.
12:04There are these five new people who seem to be kind of in line with guests that you've had before. You know, you've had Broadway guests before. Here are a couple new Broadway guests that seem to be getting more popular based on social media, based on blog posts, based on media coverage.
12:18You might wanna reach out to them. And by the way, here are prewritten emails. And by the way, I've already written and drafted the emails.
12:23They're drafted for you. And by the way, I've already reviewed them from five different lenses and made sure they didn't sound like AI and made sure that they sound like the last 10 other emails that you've written. And by the way, I've already run it by three other people on your team and it's already been blessed.
12:35And I've already laid out what that schedule of that podcast could be. Like, the idea that you could chain multiple tasks together, the idea that you could review it from multiple perspectives, the idea that it could connect into tools and actually get stuff done, that's the big change.
12:50Chatbots back and forth, it's give me a little more information. Tell me a little bit about Allie. Tell me what, you know, awards she's won.
12:57Tell me where she's worked before. Tell me why I should hate her and yell at her. What, like, whatever the thing is, you're trying to elicit knowledge.
13:04In the new paradigm that we're in, in in addition to eliciting knowledge, so in addition to all the things that you've built out, you are moving into an action oriented proactive view. So ask yourself and any person that's watching this, ask yourself right when you ask Trashiputti or Claude or Gemini, whatever question you're asking it, ask yourself the next three things that you're gonna do with that knowledge anyways, and then go to these systems.
13:30Go to Claude Code, Claude Cowork, Codex, antigravity, whatever. Go to these systems and say, hey. I was about to ask you x y z.
13:38I was about to ask you to research Ally Miller about this podcast. After I do that with you, I was gonna go off and do these three things. What would it take for you to help me complete those three things or for you to complete those three things for me on a schedule every single morning or every single hour?
13:55And it is gonna talk you through how to connect it into Gmail and Google Calendar, which, by the way, is literally a toggle that takes two seconds and you log in. I didn't even know you could do that. Go into Claude Cowork.
14:06In the next five seconds after we end this, download the Claude desktop app. Go into Claude Cowork. It's like a tab on the left hand side.
14:13You're gonna go to customize connectors. You're gonna search for Gmail and Google Calendar. Toggle it on, sign in.
14:18That's it. And now, you can go into Claude Cowork and you can say, draft me 100 personalized emails based on the last 50 emails, you know.
14:29Understand my tone structure, point of view, method, attitude in my previous writing. Analyze it.
14:37Get deep into it. Understand the nuance. Understand my secondary intent that I might be putting to these emails.
14:42Like, don't just say, memorize my emails. Really dig in and, like, get to the meat of who I am in these emails. Grab all that, and then every single morning, I want you to draft me five scary emails that I wouldn't have even reached out for.
14:58Right? Like, if you've done Broadway people on your on your guest list, right, But you haven't yet reached out to Adena Menzel or Megan Hilty.
15:09Use AI to help you get over that fear. We talked about using I AI to become the person you wanna be. It's using AI to become the person you wanna be.
15:17It's using AI to run the business you want to run or have the impact you wanna have. If you have felt held back from reaching out to the big stars or or reaching out to the big CEOs or whatever it is, like, use AI to help you get over that fear.
15:33I am constantly asking like, I literally built out this entire fear busting tool where I go to AI.
15:40It's this, like, whole interface that I've built out, and I give it my fear. I might say like, I just started strength training. I might say, like, I'm worried that every single person at this insane gym is going to tell me that I'm a loser.
15:54Right? That might be a fear. It is going to come back to me.
15:57It is gonna break down that fear into, like, four levels. It is gonna unearth this, like, subconscious scary feelings that I might have.
16:04It is going to address it with action and with potential, you know, meditations that I can do or or positive reinforcements that I can give myself or things that I can say in front of the mirror.
16:16Right? I am using it every single day or at least every single week.
16:21Like, now I'm less fearful of things, but every single week to, like, bust through these walls. Like, heard someone talk about using AI as a means of seeing any of these blockers or obstacles. Instead of seeing it as brick walls that are 30 feet tall, you're gonna see it as paper walls that are flimsy and and that you can soak in water and that they don't exist anymore.
16:44Go to AI I mean, don't, like, crazy overshare all these personal details, but go to AI and say, I've always wanted to blank. What would it take to blank? Here is context on my life.
16:54Wait. What do you mean don't overshare crazy personal details?
16:58I wouldn't give it crazy detailed financial details or health details. For those specific use cases, there are ways to have AI models run fully offline so you're not sending it into cloud providers.
17:11Everyone's gonna have their own risk posture. Some people are gonna feel completely fine sharing photos of rashes, giving personal bank details.
17:19For those people, fine. Go for it. For the average person that is new to using these tools, I'm gonna say to be a bit more protective.
17:28So if you download something like LM Studio, it is a way to run local models. All these things, zero coding required. Download LM Studio.
17:36Download one open source model like the GPT OSS models. And then you can just use these systems fully offline. You can upload attachments.
17:45You can talk to your data. It can't browse the Internet, so it'll be a little bit limited. But you can just run things that only sit on your laptop.
17:53And the worst thing that could happen is that your laptop gets stolen, but it's not going and putting your data onto the cloud. It's not using third party servers like an AWS
18:02or GCP or Azure or Anthropic or OpenAI. It's just living on your laptop. Do you feel like there is a security issue with some of the stuff that people are putting into
18:12these chats? It is not to me to determine whether there's a security issue or not. It is my, really, role to say, here's what the security looks like.
18:22And it is up to you to determine whether you think that's trustworthy or not. Like, we have people posting, like, naked photos online.
18:30Like, for some people, that is what they want to do, and for some people, that's not what they wanna do. I don't wanna tell people what to do. For the security protocols of AI, if you're sending it out into third party providers, it's important to know that once you put in that prompt and once it gets sent out, your input is getting sent out into the cloud.
18:50The actual output, like the AI's answer is getting brought back. It's getting given back to your computer. That input that you gave it, including the attachments that you gave with it, live on servers for thirty days.
19:02It lives on these servers for longer. If there's a potential, uh, you know, abuse in your input, if you're, like, asking it to make a bomb for you, whatever, all these companies have the right to hold this for longer.
19:14They have the, uh, right to hold it for, you know, a certain amount of time. And then there are also lawsuits happening right now that might change how long that data is retained. If you're on an enterprise account, the data of, retention policy is different.
19:27There are, um, everyone is defaulted to opt in to data sharing. So whether someone has taken the time to go into these settings and toggle one button is gonna change my answer for how protected they are.
19:42For every single person listening to this, and I'm sorry to the AI labs that I work with, including sweatshirts that I'm wearing of them, like, I would tell every single person to opt out of data sharing. Half my time I spend in San Francisco, half the time I spend in New York. And the way that Silicon Valley versus the rest of the country, the rest of the world thinks about AI is very, very different.
20:02Like, I used to live in San Francisco. It's very interesting to see how different cities, different regions, different types of companies are thinking about AI. 100%.
20:10I actually so we were in San Francisco for the Super Bowl this year, uh, with the NFL. Oh my god. Very cool.
20:15Yes. And it was, uh, and it was so funny, like, when we we drove from LA, we were living in LA for the winter, and it was wild as we were driving in to see all the billboards be about AI. Yeah.
20:27I mean, it's so different here in New York. We were talking about Broadway before we started filming. It's like when you're in New York City, it's just it's all about the entertainment industry and different brands and products you can buy.
20:37But the billboards in San Francisco, literally 75% of them are AI billboards. There there are memes of them. Uh-huh.
20:44Uh, it is my favorite thing. Like, whenever I land at SFO,
20:49I take photos of the billboards coming in. And by the way, not only can you see that there are AI billboards, the way that we talk about AI in different cities is different.
20:59So you have billboards in certain areas that are, uh, bragging or goading or even empowering companies to replace humans with AI.
21:09And in certain cities like San Francisco, that is seen as an exciting potential.
21:15Take that out of that city, move it into, I don't know, Dallas or something, you're gonna get a very different reaction. So it's also it's like an interesting temp check. You said that you were in LA for the summer for the winter.
21:27Excuse me. You said that you were in LA over the winter. I went to Sydney, Australia for a month this winter to get out of New York snow, which made the right call there.
21:39And, again, the way that people are even thinking about tech, I would go to the beach not barely anyone was on their phone. I would be waiting in line for some restaurant or something for twenty minutes. No one's on their phone.
21:50They're talking to people around them. Like, just the way that we approach cell phones, the way that we approach apps, and then the way that we approach AI, we are going to see going back to the trust levels of of how we see people interacting, The US has one of the lowest trust levels of AI globally, and that is very different when you break it down by city.
22:12Interesting. And
22:13how is that gonna set The US behind? It's
22:17to the extent that trust matters for innovation, that is what I would be looking at as an assumption. Right?
22:23So if you assume that widespread usage gets you more data, gets you more voices at the table, gets you more feedback, and that that feedback loop helps AI labs iterate and innovate more, then trust would have a massive implication. If you instead believe that it's a couple thousand researchers locked in a room and that they can just think and innovate and create brand new algorithms, brand new models, then widespread adoption doesn't matter as much.
22:49So it's it is gonna be that little bit of a difference. I think right now, literally, even if AI got no better, even if there was no innovation from this point forward, we still have probably a three year lag on enterprise adoption, enterprise capabilities.
23:06Small medium business is even longer. Government is probably even longer than that. Right?
23:11And so trust aside, there's just also people friction that takes a while.
23:18And so if I'm thinking about innovation widespread usage, I'm also looking at how many people are using the most state of the art stuff. 6% of business professional no.
23:306% of all all, uh, employees, not just even business professionals. 6% of of people who work in businesses are using AI agents. 6%.
23:40These things have been out for over a year. 6%. That means that if any single person watching this has even used an agent, they're ahead of 94% of folks.
23:51Like, that is how much of a lag we're on even with the current trust of today. And it seems like to what you're saying is or one of the assumptions would be that AI can make people so much more productive if they know how to use it, and productivity at scale is probably good for our country.
24:09Right? It is an interesting debate. But or is that an that's an assumption.
24:13I love the assumptions. Is
24:16is Assumption. More philosophers than AI too. I'm not a philosopher, but I would love it more than we're here.
24:21Um, I think that productivity is an important component to the AI story. Right?
24:28Like, the industrial revolution, all these things have helped us become more productive. What you've also seen over the last several decades is that productivity and wages used to be growing together, and now they are not. And so there is a new trade off that individual employees or team leads are taking where productivity gains are not benefiting them as an individual as much as we saw decades ago.
24:51And so you have this weird tension or I mean, not weird, like, a very obvious tension where you have business leaders who very clearly want productivity.
25:00They want the same output for lower cost or at the same cost, much higher output so that they can grow new business lines, grow their market, but you have lower incentives for the individuals or team leads to actually do that.
25:14And so that is one tension that everyone should be solving for on both sides. And I'm not saying that you should maximize productivity. That's number one.
25:21The second is the a harsh truth that the majority of businesses that I talk to don't even consider until we have these hard conversations.
25:31But I have to talk to like, I work with CEOs of Fortune 500 companies. Obviously, not wearing sweatpants, but I show up to these meetings, and we talk about the future of technology and the future of their business.
25:43And, obviously, I have to tell them, if you only focus on productivity, you're screwed in five years. Do you think that productivity is what wins in the AI age? Are you insane?
25:52Like, there is so much more to be had in the AI age. And transformation, that is the word to actually be going for.
26:00I want benchmarks of people having fun. I want benchmarks of people living rewarding lives, having a rewarding job. I want benchmarks of net new things, new growth, new reinvented workflows that literally couldn't exist before.
26:17The majority of businesses, vast majority of businesses are like, ah, you used to write one blog post. Now you will write five. Are you kidding me?
26:26That is, like, the most boring job I could possibly draw. And you're going to we're not quite there yet. Right?
26:34We're still at this tension point where we have some layoffs happening. We have headcount switches. We are going to get to a point where you are going to have employees, the ones who are left.
26:47They're gonna say, you gave me the worst job ever. I don't wanna do this. This is not rewarding, and I can't show up to this job with the same joy, vigor, excitement, you know, impact driven motivation that I used to have.
27:02We're in the middle of this. Like, we're probably actually at the beginning of it, but I would if you are an I mean, if you're an employee at a company obviously, everyone's an employee at a company.
27:12But if you're an employee at a company and you just work for yourself, take a really hard look at what your company and your leadership has done in 2023, '24, '25, and beginning of twenty twenty six, and ask yourself, have they only prioritized productivity?
27:27And that might give you a leading answer to how they might behave for the next three years. So when you're consulting these companies, how are you consulting them to adapt AI in their workforce?
27:39It's so funny. I think, like, so my the the method of my advising has changed over time. 2023, it was a lot of don't ban it.
27:48Like Yeah. And then a lot of board meetings.
27:512024, it was a lot of conversations around ROI and efficiency.
27:562025, it was, hey. AI agents are a thing. Let's start talking about it.
28:01The way that we're working is changing. And then 2026, it is a complete, uh, new paradigm shift talking about multi agents and digital workforces.
28:10Like, I have a team of 34 AI agents where I am talking to my AI chief of staff, Simon. Simon has his own assistant, Toby.
28:19Simon has six direct reports that are all named after the friends, Monica, Chandler, Ross, etcetera. Wait. When you say AI agents, these are not real people.
28:26These are this is AI? Correct. It is it is, um, I mean, I could make it sound way cooler than it is.
28:33Let's be honest about it. There are files that describe the personality and the, uh, guardrails and the tool usage and the model for what I want each of these AI agents to have.
28:44But, like, let's make it very simple. They are fucking files. And each one of these, Simon, Toby, all the six friends, all of these sub agents underneath the six friends, each one of that, uh, each one of those agents, it's a file.
28:58So as I'm talking to Simon, Simon not only has the Simon file of what his tasks and goals and personality and methods are, but Simon also has Simon's memory.
29:11Simon also has a company state file that he can go in and check-in. Simon also has interaction, uh, with these agents and shared files among all of those.
29:21Simon also has the ability to delegate to Toby, the the, like, assistant, to mark down what is happening, to keep a list of what agents are being used when. It is a complex, orchestrated set of 34 files that have a lot of memory and support systems around it.
29:42But the reason I bring that up is because the way that I advise companies very clearly has changed over time. And it was changing about every year, and now it's changing every three to six months. And the companies that I have worked with now for years I mean, if you had looked at our twenty twenty three conversations, we're in a completely different world.
30:03I think the way that I meet with them in that first meeting is not, hey. Let's use this tool.
30:09Let's open this button. Let's click this thing. Let's start this file.
30:13The way that I meet with them in that first meeting, half the time, we're not talking about AI. Half the time, I'm talking to them about their culture, about how they make shifts today, what it means to change, like, their work process, what the reaction has been from their employee base for the last several years.
30:31Is their leadership even bought in? Like, there are some times where I'm talking to a CEO, and they're like, my CHRO is out.
30:40Like, they they are not in it with me. How do I get them on board?
30:45Or, um, my CHRO is the loudest person in the room. Right? Like, how do I support that person?
30:51And so every single company is gonna be different. But if someone, uh, listening is, like, trying to move into the AI advising space, which, like, by all means, we need more people teaching these skills, especially to small, medium businesses. It is such a big business opportunity, especially if you're, like, a go getter who has the stamina to, like, run a marathon.
31:12Like, you have the stamina to learn these tools. I would go in and have that entire first conversation just be about goals, blockers, resources, and then have AI just be one of those resource capabilities.
31:25Okay. Quick questions. Your AI agents, do you build everything on Claude?
31:30Right now so it's interesting. All these things can be ported over because they're just files. Like, we used to have all of these chat logs sitting inside of Chattypie Tea, where you had, like, three years of conversations.
31:41You're like, oh my god. There's no way I could bring that over to another provider. Because these things are literally files sitting on my actual laptop, that means you can switch between providers.
31:52All of the initial digital workforce build out that I've done is inside of Claude code specifically.
31:58Got it. And how nice are you to your chat agents? How nice are you to your chat agents?
32:03It it changes. Right? Like, I at at default, I'm, like, treating them like an AI system, um, and not like, oh my god.
32:12You make me feel so happy inside. Like,
32:15I Do you say thank you?
32:17Yeah. I say thank you. I say please, um, not because it performs better.
32:23Like, it has been scientifically proven that being polite or being mean and threatening that the AI systems doesn't actually change the performance. But if it meet like, there are times we're just being effusive and super helpful and like, yeah.
32:39Let's, like, go get it. The AI will reflect your language. And so if you feel like you're about to get into this, you know, crazy work mode where you're like, we are working for the next twenty four hours straight no matter what we do, you might want to have that tone of voice to your AI so that it reflects back to you.
32:58Or if you're in a really big cheerleading mode, you want might wanna speak to it of like, hell yeah, Claude. Five exclamation points. You go, girl.
33:06And it is going to respond back. If you want a permanent personality, uh, then you put it in the actual, like, system instructions.
33:14So as one example, I cannot stand, like, emojis in the text responses.
33:20So I'm like, scrap all emojis. But also at the end of every single Claude response that I get, I ask Claude to give me three to five additional proactive, responsible, autonomous tasks that it can take off my plate or that it can set up as automations.
33:37So you talked about, like, at the very beginning, how do I just make this thing more action oriented? How do I work with it as a task doer or as an agent and not just a knowledge repo?
33:47Literally make it work for you. Right? Like and I don't mean work for you.
33:52I mean, like, work for you, vibe with you. Like You have to. Because, like, sometimes Jeremy will go on my chat or my Claude.
34:00At actually, Claude, we haven't really talked about with each other yet, but sometimes he goes on my chat and he'll talk to it and ask it specific questions. And he's always like, damn, like, your chat is not smart. Oh my god.
34:11Don't insult my chat like that. I love my chat. I really do.
34:14Do you name your chat? No. Should I?
34:16No. Okay.
34:18Again, unless you think that that, like, helps you interact with it better, but for the most part, probably not. I think, like okay. You know how if you go onto someone else's Netflix?
34:27Like, have you ever, like, stayed at an Airbnb? Yes. I have.
34:30And you open up the Netflix, you're like, what is this shit? They're like, damn. They love anime.
34:33Yeah. Exactly. Which, like, anime is great.
34:35But if you open to it and you're like, why is everything a murder show? Like, should have people coming onto your chat or Claude or whatever your AI experience is and going, what is this?
34:44That is how customized it should be for you. It should be behaving in the way that you want. It should be orchestrating in the way you want, saving things in the way you want.
34:52It should be really hard for people to come in and use your systems because that is a reflection of how customized you have made it and how goal oriented you've made it for you.
35:03Right. Okay. I've heard you say that women are adopting AI 25% less than men.
35:08Stats say that. The stats say that. Okay.
35:10So the stats say that women are adopting AI 25% less than men.
35:15Why do you think this is happening, and what effect does it have? First of all, like, I so I get asked that question a lot, and, like, my answer is getting angrier and angrier each time. My answer now is that there is literal scientific research that women are better at building and using AI agents.
35:35Like, these are new papers that are coming out. And so maybe we haven't had as many, like, public examples of people using AI. Like, lot of the creators and amazing engineers that are putting out content, they're men.
35:48It could be that we're busier. It could be I mean, there's a thousand reasons. But, like, AI is for all genders, all people, all ages, and there should be actually nothing holding holding us back.
36:02Like, these tools have never been more accessible, and we literally have scientific papers coming out saying that women are better at the current paradigm that we're in. Why?
36:13Is do they give reasons? You can you can theorize that, like, maybe there's something around theory of mind.
36:20Maybe there's something around empathy or delegation. Maybe there's just something around really strong communication. Like, women in there's a research paper that, like, in 75 different countries, they looked at women's writing skills and verbal skills as compared to men's.
36:34Women outperformed in all 75 countries. Like, we are I I I think on average, we are very strong communicators.
36:44These systems, the the language of choice is natural language. It's English or whatever language you speak. It is not crazy coding, if this, then that spreadsheet, whatever.
36:57You can speak to it in voice. Like, I am constantly talking to my AI systems, dictating, not just typing, and, like, visuals.
37:06Like, the two main ways that I talk to AI, it is yapping and uploading images. Like, we have so many strengths.
37:15I think if people don't want to adopt AI, like, you don't have to adopt AI, you get to make that choice. You will get to continue to make that choice and see what new assumptions or new data come, and you can change in either direction.
37:31But there is no reason that you can't. Right?
37:35The the the obstacles that existed when I when I was first starting in AI, the obstacles were massive. Even when GBT two came out, like one of the first big models that you could actually deploy in an API, you had to know what an API was.
37:51You had to know how to code to be able to bring it in. And, like, at the time, you could write a sentence and AI could finish that sentence. That was extremely gatekept because it was only accessible to developers.
38:03We have systems that are free that you can just yap to, and it can literally take hours of work off of your plate every day. So why do you think
38:15women or men, if they are, are hesitant to adapt AI? Like, what are some of the biggest misunderstandings you hear about?
38:22I think
38:23so first, I think the productivity story is told way too much. Like, I I mean and I I don't wanna put words in your mouth. Like, I don't I don't make assumptions of anyone.
38:32But if I just heard someone tell me, like, you can do 800 things in the time of like, that's not motivating for me. Maybe it's not motivating for many people out there.
38:42I think a lot of people don't wanna be these productivity drones that are surgically attached to their laptop. And so the more stories that we can tell that are about getting away from your laptop, not just talking about productivity, but maybe, like, having a family that bonds together or friends that bond together.
39:01Like, I use AI with my friend group as a means to bring us closer together. I think if more of those stories were told and it wasn't just a productivity story, we might actually have people who are, um, hesitant or opting out be leaning in.
39:17I think if we also were more honest about data privacy, again, just sharing what the reality is and not just telling scary stories or only telling hype. If we just had the knowledge more cleanly out there, that would be helpful. Like, what I was just sharing about data privacy, like, just that knowledge is pretty rare to know.
39:35All of those rests, just being able to lay it out on the table and say, you are an adult. Like, you are a brilliant person. You've lived for decades.
39:42Here's the information. If you wanna use it, great. Here's some possibilities out there.
39:48I personally think that you should give it a go so that you have a seat at the table. But if you totally decide to opt out, like, that is up to you. Right.
39:55And I also feel like it's kind of flipping the script on productivity and saying more so, like,
40:00well, productivity is one way to think about it, but it's also, like, if you're thinking about using AI in your work,
40:07you could be having your work become a little bit more passive for certain tasks that previously took you a while. Yeah. Which also helps you be more productive in other things you wanna do in life.
40:16Like, productivity is important. If you don't use AI to become more productive, like, you're clearly leaving something on the table.
40:24It's more that Productivity is getting a bad rep. Productivity is a a trap because we're we're being told this story that the only thing that businesses care about, the only thing that AI labs care about, the only thing that individuals should care about is productivity.
40:38But, like, if you're using it to, um, I don't know, write social media posts for you, whatever. Instead of just using it to write the social media post because I'm still writing my own post, use it to review your social post from 10 different angles.
40:53If you're using it to make a workout plan, don't just say, make me a workout plan for every single day of the week. Say, make my workout plan. By the way, I'm not gonna do whatever you say because I'm gonna wake up tired.
41:04I'm gonna tell you that I hate my outfit. I'm gonna tell you that every single time I go running, I always wear things that are way too warm and now I'm too hot. I'm gonna tell you that I stayed up until 5AM the night before.
41:13I'm gonna tell you all these excuses. Now rebuild this five different ways. Let me evaluate it in a brand new interface.
41:20Let me dictate to you back and forth for the next hour about how we're gonna make this system that much more robust. What did I just do in that last hour? I didn't make I wasn't more productive.
41:31I just made the system better. I made it more robust. I increased the quality.
41:35Right? I didn't just make more, more, more, I meaningfully transformed the way I got that task done in the first place.
41:44And again, whatever task you're working on, client outreach, whatever it is, don't just say instead of writing one email, write me a 100.
41:52Say, review it from all these different angles. Make it more in my voice.
41:56Um, you know, review it against these seven risk, uh, these seven risks and help me mitigate against them. Um, I wanna grow into this brand new business line.
42:06Coach me through this. Listen to my call live. Literally live coach me while I'm in this meeting.
42:14Right? Like, that is not more productive.
42:17It's not just shoving more out into the world, which is just like a lot of AI slop. It is looking at your core. There's a lot of, like, self reflection that's required with AI transformation.
42:28It is looking at your core and going, who the f do I wanna be in life? If you wanna be a podcaster, a runner, if you wanna be a CEO, if you wanna get promoted, if you wanna pivot your job, if you want to, uh, you know, move to a new city, whatever the thing is that you are trying to do, yes, look at it from a productivity lens because I do think it's important.
42:48But literally ask AI and be like, Ali told me that I am too focused on productivity.
42:54Give me 400 different ways of thinking about this from a non productivity lens. Yell at me.
43:01Let's argue back and forth. Let's bring in five other personalities so that we can all argue in a room. Make it a a PowerPoint and convince me that I'm wrong.
43:10Turn this into a video so that I can show it to my husband so that we can get more involved in it together. Turn it into a funny meme that I can post as a poster on my computer. Like, thinking about it through the weirdness angle can sometimes, like, elicit cooler AI interaction
43:29to be able to get to the life you want. I really feel like there needs to be courses on how to use AI because I'm teasing some.
43:36You're You could No. You are. This podcast is so interesting because you really are educating us on so many ways to think about using AI to help us.
43:44It's just like ways that I never even would have thought of, you know? I just yeah. Yes.
43:48As somebody that uses AI every single day. And, like, again, just just
43:53validating and empathizing with people, These labs are not giving you a manual. Like, we're supposed to show up, we see a blank white box, and we're like, ah, yes.
44:01Reinvent your whole life with this. It's not obvious how to do that. And so it is helpful to look at people like me who have been using this for a decade, who are hammering it for hours every single day, who have built out digital workforces, who have done all this stuff, to say, I don't wanna spend sixty hours watching videos to learn the thirty minute thing.
44:22Like, I'm gonna go to advisers, influencers, content creators, educators, people at my work, whatever it is, to just speed up my learning curve. Like, that is my my goal in helping a billion people is to speed up the learning curves that people don't make the same mistakes I do, give people the resources to make those empowered decisions, and to teach people like you or whoever's listening so that they can teach the next 100 people.
44:45Like, we all have to help each other and pay it forward, but there is a certain amount of, like, you just gotta lean into the weirdness and just ask Claude or ChatGPT and say, I keep hearing on these podcasts that I'm supposed to use it for action. Ask me questions.
45:00Or I keep hearing that I'm supposed to build AI agents or 34 agents or whatever. Can you interview me about my day? And then can you figure out what this even looks like?
45:09Like, I don't even know what this means. Turn it into your personal tutor, your personal coach, your personal doer, but use it as, like, a real time context gatherer
45:17to get into that weirdness. So the future of work, I feel like we talked about this when it as it relates to San Francisco, there's a lot of companies that are celebrating being able to have less employees working for them and more jobs being taken over by AI. While it's being celebrated in San Francisco, I think it's one of the biggest fears that people have.
45:36You know, I hear about it all the time from my friends that work in different industries from marketing to public relations to people that are artists. Like, it's scary, the concept of AI taking over our jobs. So from your perspective, what are the careers that are at risk when it comes to AI?
45:54So first, again, we're kind of in the muck of it. And and by the way, the muck might go on for years or decades.
46:01Um, it is not clear even if AI has a large overlap with a job, which there are several jobs where AI's overlap with the job is, like, 70 plus percent. It's still not clear whether that is equivalent to displacement.
46:18There are some jobs that I think are more at risk. I think a lot about, like, people who write closed caption for TV.
46:26Um, I think a lot about, like, manual data entry, uh, or or junior research roles.
46:34Like, things where you can kind of already see that the average person is able to use AI even with bad prompts to complete that task. Uh, anything that is, like, very physical, nuanced, or has high liability.
46:47So, like, I always say that, like, phlebotomists are very safe. But think of, like, electricians.
46:53Right? The the liability of what happens when that goes wrong, very high.
46:58The law the field of law, the field of medicine, those air I mean, architect, like building skyscrapers in New York.
47:07Anything that feels like there's higher liability, you're gonna be in a better spot. I think that the the next year, the thing that I would be looking at is there's a benchmark called GDP val.
47:23GDP, you know, economic, um, val, like like, um, evaluation.
47:28So it's measuring AI's capability of completing real world tasks that have, like, meaningful economic impact on the I think it's on The US GDP, but it might be on global GDP.
47:40And so they look at all these different, uh, like, 44 different occupations and all these different tasks, and it's the, uh, highest leverage task for each of those jobs.
47:51And they look to see whether AI could complete that. The systems that we have today are scoring 80% plus on GDP value.
47:59Like, we're already in a situation where AI is performing knowledge tasks at a very, very high performance level. Whether that means we trust it at that level, whether we think, uh, companies can can run teams of one, um, like, our our systems are still set up for a lot of people working together.
48:24Um, you're gonna see teams get smaller. I think that's probably the biggest thing that I would be looking at. The companies that I work with, the best ideas, the best products, the best work is more often than not being done by teams of, like, two to eight people and not teams of 300.
48:45So even if you're I mean, if you're a subject matter expert, if you're an engineer, whatever it is, like, find your pod, and it's a go further together kind of thing.
48:55All of these things it's it's not that being on the small team makes you future proof.
49:02It's that being on the small team helps you learn faster and adapt faster, and it's the power of adaptation
49:08that makes you more future proof, if that makes sense. So what are you saying? Like, what advice would you give to a 22 year old that's fresh out of school and is worried that they're not gonna be able to land a great job because of AI?
49:23My advice is is not immediately AI tied, but I'll give the general advice, and then there's an AI component to it.
49:30For whatever reason, when I interview really, really early in career folks, I feel like I'm missing this, like, proactive ownership level and the ability to look at a swath of options and measure trade offs.
49:49And I don't know why it's missing, but it's really missing from early career folks.
49:56So I, um, stole this idea from Alex Lieberman, but it's like a pyramid of proactivity. And I think he stole it from his house. But Alex Lieberman, friend of the show.
50:06There's levels one through five of productivity. And I would tell every single 22 year old listening, I guarantee you, you are at a level one or two.
50:16And if you are at a level four, you will actually be hired. So level one is that you come to your boss or you come to a potential business that you wanna get hired by and you're like, here's some problems that I've labeled not gonna get hired.
50:28Level two is like, here are some problems that I found and some potential, uh, causes. Level three is here's the problem, potential causes, and potential solutions.
50:39Level four is here's the problem that I found. Here's the potential causes, potential solutions.
50:45Here's the one I think that we should go after. Here are the trade offs for why I didn't pick the other four, and I really think that we should pick option three, and I can execute it in three days. If you talk that way and you start using, like, the STAR method and you're able to talk about ownership and the trade offs and the plan back, you're you're top 1%, at least top 10%.
51:08Once you are inside of a business, I am moving every single person on my team up to a level five. The more AI also comes in, the more ownership every single person's gonna have, the smaller the teams are gonna be. Right?
51:20The more autonomous you're gonna be because you're also gonna have this team of, like, a 100 agents. Level five is there's this problem.
51:27Here are potential solutions. Here is the one that I picked. By the way, I already went ahead and solved it.
51:33Here are the things that I'm gonna look for to know whether the problem, uh, was solved or not, and here's how I'm gonna track it.
51:41And at every single one of those levels, there are ways that you can use AI for it. I can send you this, like, pyramid thing. But I want someone showing up and just showing up as a capable, like, gremlin mode, high ownership, high creativity, long term problem solver.
52:01And that tells me that they have the flexibility, the adaptation, the curiosity, the agency to actually be an unbelievable not just, like, a helpful contributing member.
52:12I mean, a big freaking leader on my team. And you can do that at age 57, 22, it doesn't matter, but I need to see the the strong sense of wonder and the extremely high agency because AI is enabling that at scale.
52:28And so you want the people on your team that have those qualities in spades. Absolutely. And I mean, we're also in an entrepreneurial space where we have smaller teams.
52:37So having somebody like that is really important. I want every single person on my team to be a systems thinker who is extremely goal and action oriented. So if I get someone and they're like, I don't know.
52:47What are we doing? Not that I think you were doing this. But, like, I want someone to be like, ah, yes.
52:50I'm in I am the manager of conversion rates.
52:54I'm the manager of client success. I'm the manager of fun for the team.
52:58Whatever the thing is. I want someone to know that space that they're owning. I want someone to know that North Star and to have full agency, creativity, um, good at setting criteria, good at gathering resources to be able to move toward that.
53:12The the flip and where I actually crazy disagree is that I think we are going to see job titles slowly melt away.
53:22Like, I truly believe that we are going to have kind of categories or buckets of jobs where you're gonna have, like, this bucket is just product related. Whether you come at it from a slightly more engineering angle or design angle or product angle, we're all kind of going after the same thing.
53:39There's gonna be another zone for, like, external stuff. And by that, I mean, like, marketing and sales and customer success, things that are in there.
53:47And then there's gonna be another one for, like, back office y operations y things. And these crazy specific titles of, like, SEO specialist for Mid America, they're gone.
53:56The people that I hire for my team have more generic job titles than I had when I first started, than I had at Amazon, than I had at IBM.
54:06And I think that is more the trend. You want generalists. You want utility players.
54:11You want people who are able to say, yes. This is my role today, but I'm gonna teach myself that other job that I'm 10% away from or 20% away from. You want people that are these, like, crazy generalist go getters.
54:22And what that means is that, like, I just interviewed someone, and they're like, I just feel like it's too general and it's too amorphous, and I don't know.
54:30And I'm like, I mean this with so much love. You're not meant for my team. Like, maybe you are meant for a more detail oriented, very niche vertical focused space, but you are not meant for how malleable my team, and I think many teams, particularly in small meeting business space, are going to be in 2026 and beyond.
54:50So if you are a brand new grad and you are like, every single coffee mug has to be turned in this specific way, and if I don't know exactly what I'm supposed to do every day, I'm gonna throw up in my mouth. You're gonna have a very difficult next five years.
55:02I also think, like, you're a very entrepreneurial person, so that's it's it is a different way of thinking. But it's also who has strength. Like, I don't know.
55:09I was yelled at, like, every single day fifteen years ago of, like, just stay in your lane. If you go back to my employee review, it literally says, Ali is, like, too distracted by learning what everyone does around.
55:22And I'm like, yeah. But I'm also doing my job. Like, I'm just I I'm I'm able to bring this whole SpokaneWille thing together.
55:28Anyways, I quit that job. It doesn't matter. Us Us entrepreneurs did not get great employee reviews.
55:32Let me tell you. I I will say I, like, loved big tech, though. Like, the messiness of it.
55:38Anyways, I think that the people who are able to not just complete their job but are able to have that flexibility, they're in a very, very strong position. But if I get someone that is very messy, that's still not what I want.
55:52I want someone who is focused, who's lethally gonna say, like, this is the goal. Here's how I'm gonna hit it. Here are five options.
55:57Here's the one I'm gonna do. Here are the three ways I'm gonna measure it. I've already set up AI systems that can track this.
56:02I'm gonna get back to you in four days to know whether it worked. And if it didn't, here are two potential options. Okay.
56:06I saw this article today that had a thesis that said, the future belongs to people that have artistic vision, adaptability,
56:11communication, and technological intelligence.
56:14I think that sounds right. I think the thing that is probably missing from that list is the agency side and the systems thinking side.
56:25Like, I think I think creativity for sure, adaptability for sure, um, the ability to just, like, learn how to talk to these systems, not necessarily in a super technical coding way, but, like, maybe talking to these systems means clicking a couple buttons or dragging boxes.
56:40Maybe it means learning these crazy if then statements even though I don't think that's gonna happen. Like, right now, the interface is only getting more accessible.
56:48So I I agree with all those. I think the people who understand these crazy complex moves are in a very strong position for 2026, 2027. Because, again, the AI of 2025 and earlier was an AI of one.
57:04It was you talking to Chatuchupti about setting up your podcast, and it was a one to one interaction that people were naming their ChatGPT, and they were having this one to one thing.
57:15No more. Like, the way that 2026 AI and beyond is moving is multi threadedness. It's managing multiple things at once.
57:24It's being able to bring all these pieces together. It is figuring out which AI systems are shared between us, which AI systems benefit a whole system. So I would really, really emphasize systems thinking and, like, literally reading books and training people on organizations.
57:41Like, the AI has moved from a one to one to a one to many.
57:46Every single person kinda needs to be, like, the greatest organization manager ever. I need to build out a team of AI video editors. That could be one, but, like, that's your starting spot.
57:57What about your like like, my Phoebe on my team is just this, like, creative naysayer that is constantly poking, prodding, adding five more ideas, thinking more creatively. Like, maybe you have some that are video editors.
58:11Maybe you have some that are video dreamers. Maybe you have some that are video cleaners. Like, whatever the the thing is.
58:17I think that would be really big. And maybe you decide to delegate that entire video management to your video team, and then you orchestrate some other team. So systems thinking I think is missing.
58:27And then what was the third one?
58:29It's artistic vision, adaptability, communication,
58:34and technological intelligence. I think, like, you can hire someone who is extremely, uh, skilled with AI, and they know how to talk to it, and they know how to be creative with it.
58:44That still doesn't tell me that they're gonna be a 100 x performer. That tells me that if I give them the perfect task, that they will be able to, like, effectively execute it.
58:55I want someone who if I'm not micromanaging them because I don't ever wanna micromanage someone. I want to give someone like, you are going to own this space. You're gonna have this goal, and I want them to have the level of ownership that I know that they're going to be able to govern this whole system to work on their behalf with, uh, like, artistic vision and creativity and adaptability and the resilience and emotional fortitude to, like, push back on AI when it says the complete wrong stuff.
59:20Like, that is who I would want. That is the character that I want in someone. There's a little bit of, like I call it literally gremlin mode, but there's a little bit of, like, hunger
59:31and a little bit of that entrepreneurial mindset that's missing from that description. So we talked about this a little bit before we started, but I think with the rise of AI and people really kinda, like, living on their phone, we're really living in this digital world. Even I talked to so many people that are content creators, and we're all like, we are chronically online.
59:47You know? I went down an AI wormhole of freaking mountain goats that are climbing cliffs the other day, and I was on it for, an hour. It's very popular.
59:54What? Sorry. It's very popular online right now.
59:56You're watching AI mountain goats just climb mountains? Yes.
59:59Because you're inspired by them? I just, like, got trapped. You know?
1:00:02Like, we're chronically online. Okay. So so That's a good question.
1:00:07have to show you these videos. It's so good. Um,
1:00:10but I feel like we're all chronically online, and I liked what you were saying that a lot of your friends that are in Silicon Valley in the tech space, you guys do completely offline retreats.
1:00:21So how much more important important are real life activities gonna be now with the rise of AI? The more the more sci fi my job becomes, the more analog my free time becomes.
1:00:32And I think, like, the middle of a lot of things are disappearing. Right? And I'm hoping that some of these, like, I'm addicted to my phone.
1:00:40I'm scrolling on, you know, Instagram, TikTok, whatever, for five hours. Like, I want that to go away. Very clearly, that's like a negative behavior.
1:00:47So, like, I've set up friend retreats in Cleveland where we all had just like a big phone detox. And those, by the way, are all friends that work in AI. I just went on a hiking trip in Salt Lake City, and none of us had our phones out.
1:01:01It it's it is so important to have that break. Like, some of my best work comes from when I'm on a one hour walk and I'm dictating to Claude, but I'm not staring at a screen.
1:01:13I'm not falling into mountain goat.
1:01:16I was mountain goat. I was mountain goat deep. Let me tell you a few of my friend by the way, people listening, not on not an unusual Have you guys seen the mountain goats?
1:01:24Oh my god. Like, screaming television. Seen Have the mountain goats?
1:01:27Oh, damn it. Okay. People They're stopping on track.
1:01:30Oh, well, that's different. Are we doing? Okay.
1:01:33I need to send you one because the for whatever reason, I'm getting these, like, fat AI stuffed animals that are, like, doing TikTok dances. I love those. You've okay.
1:01:41Have you seen the fruit ones? No. Oh my god.
1:01:44They're addicting, those little videos. See, but, like, I'm not I'm not, like, scrolling through them for five hours. Mhmm.
1:01:49They're, like Oh, I follow the the accounts.
1:01:53If I see them and it's good, I'm pressing follow. I okay. Going back to, like, the productivity thing, not everything in AI is meant to be this, like, a hyperproductive thing.
1:02:02Like, we're allowed to just, like, laugh in new ways and just see absurdist things, And maybe it's a nice distraction for, like, a day or two or a month in your kids' mind. I think of, like when I think of
1:02:13the dancing and, like, AI, fruit, and animals, to me, it's like watching Disney Channel.
1:02:19Like, sometimes you just need to turn your brain off and rot a little bit Interesting. To get away from the nonsense online. I would pay good money to see Hilary Duff perform.
1:02:30I would not pay good money to see a stuffed animal dance on the screen. No.
1:02:35So in terms of, like, continued commercial but I I I appreciate that these videos are in my feed every it'll be like AI AI AI AI stuffed animal dancing AI AI.
1:02:48Barbershop AI. Sometimes I if AI videos are getting so good, I, like, can't tell the difference. Like, it takes me a second often to figure it out.
1:02:55On the mountain goats or on humans? Mountain goats, I well, I started from a real video, and then I went deep into AI videos. We are,
1:03:02like, just getting past the point on being able to distinguish real video from fake video. Right now, the the main thing is that AI video can really only be, like, one to two minutes long at a decent quality. But there's still I always talk about, like, the viscosity of AI videos.
1:03:16I don't if you noticed this. What does that mean? It's like, you know, like, when you, um, put something into molasses that it just, like, moves more slowly.
1:03:23Right? Like, if you, like, pour out gel versus pouring out water, the viscosity of the thing. So if I'm watching an AI video, I still feel like it looks like they're filmed in, a different gravity space, that they're filmed, like, in space or on Mars.
1:03:35Like, everything's moving, like, a little too slowly. Or, like, if you're bouncing a ball.
1:03:40So my like, all the how many teeth do they have in hands? Like, all of that is completely indistinguishable.
1:03:48Uh, if you are trying to prove that something is or is not AI, have them make wild, crazy gestures that are not just the same on repeated and have them, like, drop several balls.
1:03:59Yeah. Okay. Quick question, though.
1:04:01Is it because I know we have dropped. But quick question. On your assumption list, are you assuming that video is gonna get so good that we can't tell the difference?
1:04:07Absolutely. But the but the time frame is changing.
1:04:11Okay. Yeah. So how soon do you think that'll happen?
1:04:14Again, all the like, I I put my predictions out there. I'm not one of these people who like, it depends. Yeah.
1:04:19I think that video has been doubling in length every, like, year or so.
1:04:27So you can imagine that within five years, we could get, like, a, like, a short film.
1:04:34But I also add on as a layer that, like, most people's predictions are too slow in software space and too long in the hardware space.
1:04:45So if if people are listening and they're like, oh, AI is gonna, I don't know, take over my robe like, my manufacturing job tomorrow. Robotics moves very, very slowly, and software that does not involve three d or video moves a lot faster.
1:05:02It's that weird middle ground of when it's software, but it has to do with, like, spatial or three d that it's a little messier.
1:05:08But, yeah, within five years. Ali, last question. Final piece of advice that you wanna leave our audience with as it relates to AI.
1:05:15Taking the exact like,
1:05:17a lot of life advice that people have been given is the right life advice going into AI. Like, if you are someone that has been goal driven, then apply that into whatever you are hoping to do with AI.
1:05:29I always ask myself, what would it take or what would it look like? As you open up the next, you know, Claude window or Gemini window or whatever, look at this window and say, I heard on this podcast over here, or I saw this tweet over here, I saw this LinkedIn post.
1:05:46Put that into the system and just say, I want to do this. My goals are these five goals. Walk me through it for the next, like, three hours.
1:05:54Like, let's get deep. And I think if people are just willing to have these planning conversations, iterate, add that context, you are going to be in a completely different position for your work and your life in in, like, actually three hours.
1:06:12You just have to sit down and instead of saying, let's build one agent that refills my water bottle. Like, take five steps back and just say, I'm a blank who's trying to blank.
1:06:23Like like, I'm a, uh, running coach, let's just say. I'm a running coach.
1:06:29I have five clients. I'm looking to expand to 15. Every single time I reach out to people, they don't respond.
1:06:35The way that I'm reaching out to them is via Reddit sub threads and cold emails. I don't feel like calling people. I only have three hours a day to call people.
1:06:44Here is what I believe in as a running coach. Here's what I'm afraid of. Here's what my, you know, sentiment is that I get back from clients, whatever.
1:06:51All of this context, give it in and just say, what would it take to three x my business? And then go back and forth with AI and
1:06:59see what you get. Ali, where can everybody follow you and stay up to date with everything you got going on? Ali came around pretty much everything,
1:07:06uh, mostly on Instagram and Twitter and LinkedIn. And I've been posting more and more, like, deep dive demos on YouTube and my newsletters where I share all these forecasts and, like, deeper thoughts on how to prepare for the future.
1:07:21So Ali K. Miller.
1:07:23Or just ask AI and say,
1:07:26how can I find her? Yeah. Yeah.
1:07:28Thank you so much for being on Post Run High today. Thank you for having me. Appreciate it.
The Hook
The bait, then the rug-pull.
An AI consultant who has spent two decades watching the field move from lab curiosity to civilization-reshaping force opens with a split: the same tool can compound your best qualities or hollow out your judgment entirely. The fork is not in the technology — it is in whether you know what you actually want.
How they asked for the click.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
OTHER LINKSAlso linked in the description.
