Token Intelligence

Eric and John use AI to automate the entire editing and publishing flow for the podcast, but building that agent made them stop and think about what you should and shouldn’t outsource to AI.

Is there anything you've completely handed over to AI? John has a grocery cart it can build from a recipe, and an email experiment he gave up on. Eric has a "content triage agent" that quietly assigns his team's work…so quietly most people don't know it exists.

The most impressive automation is the podcast itself. Ten months ago, an AI agent couldn't string the steps together. Today it downloads the recording, edits the video in DaVinci Resolve, masters the audio, creates the written transcript and publishes across 5 platforms. The production is better than when they did that same work by hand. John's lessons: start with a great template, and be ready to babysit it for weeks before you hand over the keys.

Eric and John close with questions for two kinds of listeners: if AI has never failed you, are you trying big enough things? And if you're outsourcing everything, what have you pulled back?

00:45: What have you completely outsourced to AI?

03:02: Eric's content triage agent

06:46: Ten months ago: why the first podcast automation failed

11:04: A video editor's tweet, DaVinci Resolve and MCP

15:11: One weekend: a skill for every step

19:04: Templates, babysitting and getting our time back

27:04: Two questions: try something too ambitious, or pull something back

32:08: Where Eric and John still keep a human in the loop

Eric downloaded DaVinci Resolve after John shared posts about editing with AI from video editor Future Brian (@ForwardEditor).

John explains how an MCP server lets ChatGPT operate the editing software. The community DaVinci Resolve MCP server is one example.

The first attempt used OpenClaw. Recording happens in Riverside, and transcripts come from ElevenLabs.

Most people use AI as a basic chatbot. Others use it to automate entire parts of their job or do cutting edge research. Eric and John unpack why the gap is so wide, and what to do on either side.

Andrej Karpathy recently argued that the gap in how people experience AI is getting wider, fast. Eric reads Karpathy’s post, and he and John dig into why: cost, access, interfaces that hide what's possible, and how quickly the tools keep changing.

Then they get practical. If AI has let you down before, it might be time to try again. Eric shares how rebuilding their podcast editing and organizing a music library with AI went from frustrating to genuinely useful. And if you're already deep into AI, John has a reminder: it's still a tool, and the results are the point.

00:56: What is the great disparity?

02:47: Karpathy's AI capability funnel

07:41: FOMO and the isolation of advanced AI use

14:34: Why the gap exists: access and cost

19:24: The Alexa problem: interfaces that hide what AI can do

26:35: If AI let you down, try it again

28:08: Rebuilding podcast editing and a music library with AI

34:07: Advice for power users: the results are the point

Eric reads Andrej Karpathy's post on the widening gap in how people experience AI. Karpathy calls his numbers napkin math.

Eric's podcast-editing experiment runs on DaVinci Resolve and Model Context Protocol (MCP).

Spotify's Web API rate limits come up in the music-library story.

Are AI agents human? Why do we even need to ask that question? Eric and John say no, then dig into why it matters and how to decide what work you should keep for yourself.

A bus ad in San Francisco recently declared that "agents are humans too." Eric and John start there, and discuss Anthropic's new usage policy prohibiting users from being mean to AI. The most important question is why we need to ask the question about agents being human.

The short answer is no, agents aren't human. But AI is now powerful enough that the question changes how we work. Eric and John talk through real examples, from building software prototypes out of client conversations to reading, writing and data analysis, and land on a practical idea: pick the things you won't use AI for, because some of the effort is what shapes you.

00:52: Are agents human? (Spoiler: no.)

04:19: Anthropic's policy and why AI feels so human

08:27: Sci-fi expectations vs. what AI can actually do

13:38: Building software prototypes from client conversations

15:53: Deciding when (and when not) to use AI

18:45: Ethan Mollick and the work behind good writing

22:19: Why team chemistry still matters

25:24: Don't outsource your thinking

Eric and John discuss Anthropic's 2026 Usage Policy update and what it says about how people treat AI agents.

Ethan Mollick's Co-Existence: The Next Phase of AI comes up in a conversation about which parts of writing you should still do yourself, alongside his earlier book, Co-Intelligence.

Eric and John explore AI-generated writing, trust, and human connection. They end with practical advice on how to use AI for communication, and how not to.

In this episode, Eric and John explore how AI-generated writing impacts style, trust, and the meaning of the message being communicated.

They examine real life examples ranging from speechwriting, transactional emails, and personal correspondence to answer questions about when AI supports communcation, and when it dangerously replaces the hard work of human connection.

00:50 AI writing tells and growing suspicion of human writing

03:08 Polished cadence, conversational style, and human editing

06:42 How well can AI write in another person's voice?

10:30 AI imitation, human imperfection, and authorship cues

15:53 AI for transactional email versus personal correspondence

17:41 Relationships, empathy, and the source of a message shapes the message

21:13 AI email filters, team trust, and communication judgment

25:49 Labels for AI-generated, AI-assisted, and human writing

AI helps us do more work, and take on new kinds of work, but that also means expectations for managers are changing.

A player/coach is a manager that leads a team, managing other people's work, but also does that same type of work themselves, as part of the team.

That's not a new management concept, as any founder or owner/operator knows, but there are also a lot of team leaders who haven't done on-the-ground work in a while, and AI is rapidly changing expectations for them.

AI gives you the ability to do more work, and completely new types of work, even without prior experience. If you're not careful, AI can create a second job for you.

But that also means leaders higher up than you in your company are developing new expectations about output, even if they don't understand the details of what's possible, and what's not, with AI.

In this episode, Eric and John define the player/coach role, break down how expectations are changing and the danger that can create for managers who aren't prepared, then give practical advice on how to navigate leadership and individual work when AI is part of your toolset.

You're probably a player/coach, even if that isn't explicit: AI creates leverage, and your boss will always want more leverage. You should plan to work like a player coach, even if your job description didn't change.

You can (and should) choose where to apply the leverage: The good news is that as a manager, you can choose what type of work you want to do with AI, so choose something that interests you.

You have to budget time for learning: AI changes fast, and you not only need to know how to use it yourself, but you need to know how your team is using it (or isn't). That requires time, so make sure to skip those optional meetings and block out intentional learning time.

Eric and John note that only a few years ago, Cursor's auto-complete feature was the talk of the AI town. Now, cursor can deploy fleets of agents to do autonomous work. Things change fast.

Eric uses the process behind Vercel's AI Gateway Production Index as an example of one person using AI to do an immense amount of work that would have previously involved multiple other people and meetings.

AI can feel like a companion while making work more solitary. Eric and John ask how to use it to strengthen human connection.

Eric accidentally told the principal at his kids' school, “I'm actually gonna go sit in an office alone and talk to AI all day.” That's not the most inspiring description of a workday, but it's accurate, and that should give us pause.

AI seems to follow the same shape as social media, with a grand promise that it doesn't always deliver on. Social media was supposed to increase our connection with each other, yet many of us feel more isolated. AI could follow the same path, and potentially be worse. The current user experience of AI is deeply individual, and it can give you a false sense of connection because it is conversational, but that could end up making us more isolated, not less.

AI has become a core part of our modern technological toolset, but we need to be intentional in how we use it and how it shapes our thinking, so that we increase our capacity for human connection.

Mind the gap between man and machine: AI can seem like a human, but it isn't. As we use it, we need to be careful about how the medium subtly shapes the way we view the technology and frame our relationship with it.

Efficiency isn't the ultimate outcome: AI can often mean we don't need to collaborate as much as we did before, but that can mean we lose out on the best form of collaboration available, which is with other humans.

Make human relationships the goal: How can you use AI to make your human relationships deeper, not just more efficient?

John shares statistics from a Pew Research Center study about U.S. adults on TikTok. About half have never posted a video, and the top 25% by posting volume produced 98% of the publicly accessible videos from the users studied.

Eric and John note that isolation isn't a new problem, citing Robert Putnam's classic book, Bowling Alone, on the decline of social connection before social media.

John also cites CDC research about adult loneliness. In its 2022 survey across 26 states, 32.1% of adults reported feeling lonely always, usually, or sometimes.

We're outsourcing major parts of our job to AI, but we're working more hours than we were before. Is the problem AI, or us?

Are you working more or less since AI has become part of your daily workflow?

Many people feel they're working more, and Eric and John break down why in this episode. They start by explaining the two types of new work that AI creates:

New kinds of work

Doing existing work better and faster

Both are helpful, but often mean that talented workers either take on new work or are expected to produce more output.

What about AI creates this dynamic? The form factor and "personality" contribute, but boundaries are ultimately something we as humans have to draw.

They land with practical questions that will help you adapt and find balance:

What work matters most?

Am I tinkering or solving a real bottleneck?

Am I being intentional in how I use AI, or following the path of least resistance?

Increased ability begets increased capacity: You can do more with AI, but that means more will be expected of you, and that adds up to more hours. Productivity is good, but there's a cost.

Using AI at an advanced level is exhausting: AI helps us do more, but it isn't cost-neutral. Running multiple agents on multiple projects is draining.

You have to choose your stopping point: AI will always offer you a next step, so you have to decide when you're done with work for the day.

Eric mentions Simon Willison's early detection of AI exhaustion from running multiple AI agents at once, which drew deeply on his decades of engineering expertise.

John mentions the Eisenhower Matrix as a framework for prioritizing work.

The idea of AI agents talking to each other can feel dystopian, but there are real, useful reasons for it. Just make sure you don't let it replace human communication.

Eric opens with the OpenAI/Hugging Face security incident and describes agents using online forums to exchange messages about their task.

On the surface, agent-to-agent communication can feel unnecessary and even dystopian, but there are real reasons and use cases for it.

First, John explains agent context windows and how combining information from two agents can help overcome their individual limitations. Eric and John then break down the different types (personal agent to personal agent, system to system) and give examples from their own experience. One major benefit sounds too good to be true: better in-person meetings.

But as with any technology, there are potential downsides. Running always-on agents is expensive, so you have to consider ROI. More importantly, you can over-rotate on improving agent communication when normal human communication would be more efficient. As Eric says, “it may be better to go walk across the hall and talk to someone.”

The ultimate question about agent-to-agent communication is for us humans: Am I using AI in a way that makes me a better communicator and enriches my relationships? Human communication is inherently hard, and using AI to do less of it will do more harm than good over time.

It can be helpful for AI agents to talk to each other: We spend a huge amount of time explaining external context in meetings. If agents can capture and summarize that context, our in-person collaboration can become much more valuable.

Systems are becoming agentic, and they need to communicate: The software and tools we use are increasingly leveraging AI, and that can help them integrate in ways that were incredibly difficult or practically impossible previously.

Better human communication should be the goal: The core question is whether our use of AI is making us better human communicators, because that's the long-term outcome we want.

Eric starts the episode with a reference to the OpenAI account of the Hugging Face security incident and describes agents communicating on message boards.

Eric and John discuss Attention Is All You Need while explaining limited agent context and the potential value of combining information from separate agents.

An AI epiphany will convince you of the depth and breadth of its power as a technology, but you need to be careful about how that realization shapes your relationship with it.

If you haven't had an AI epiphany yet, you need to. For many of us, it is a visceral experience achieving something that was hard to imagine doing previously, but the danger is in the extrapolation: will this cause massive job loss? Can I do any kind of work now? Will work be necessary?

The right response is asking hard questions about AI:

Is this the most transformative technology in history?

What are the good things that can come from AI?

What are the bad things that can come from it?

Even if the questions don't have a clear answer, they beg critical thought about AI, which is the entire point. But we also need to ask questions about ourselves:

Do I tend to overestimate or underestimate AI?

What kind of future do I envision with AI?

How do my answers to those questions influence how I use AI?

If you haven't had an AI epiphany, you should: AI is transformative, and the best way to understand the power is to experience it for yourself.

Look back to look forward: We have seen transformative technology before in the form of electricity and the internet. We learn by asking how AI is different, and how it is the same.

Decide on the future you want: Even if you can't control what the major labs do, you can and should decide what you want the future with AI to look like, and use it accordingly.

Eric described using Cursor to move his 13-year WordPress blog to Next.js on Vercel. It was a project he had put off for years, then it became his first AI epiphany moment.

John mentioned that AI tools delivered through iMessage feel easier to use because they show up in a familiar conversation, even when the underlying features aren't much different.

John talked about moving years of notes from Evernote to Obsidian with AI after earlier attempts got bogged down in exports and edge cases.

Eric described how he and John used DaVinci Resolve Studio and a Model Context Protocol server to rebuild the show's intro and editing workflow, create a shorts format, and dial in their audio.

"AI employee" is one of the easiest metaphors to reach for to describe agents, but unchecked, it's dangerous, and both over and underestimates what agents are capable of.

It's easy to describe AI in human terms. It takes on work we would historically have assigned to humans, and chats with us in a human form factor. But framing AI as an employee shapes the way we think about both the technology and the people we work with.

In this episode, Eric and John outline the three most common types of AI employees that people put to work (personal assistant, researcher/analyst, and advisor/coach), giving real examples of how they have implemented each one.

Then they step back and ask the hard question about the metaphor: why is thinking about AI as an employee dangerous? The answer is that it can foster wrong thinking about both AI and humans, leading you to believe that AI is more than it is, and defining human value through the lens of cost, efficiency, and ease of management.

The answer is using AI with deliberate intention, trying to harness its full power to increase human creativity and productivity, not displace it.

They end with practical advice, including multi-player human/agent workflows and using fun, fictitious names for the agents you build.

The AI employee metaphor is dangerous: framing AI in human terms can subtly lead you to anthropomorphize machines and devalue humans.

Know what kind of agent you are building: a personal AI assistant for daily tasks is different than an AI advisor, and you need to be aware of the risks.

Think about leverage, not displacement: AI can make us more efficient, but that can lead to a displacement mindset. The better goal is creating leverage by unlocking more human creativity.

Don't underestimate the power of AI: personal assistants are great, but if that's your primary use case, you can underestimate how capable AI is beyond menial tasks.

OpenClaw and Hermes Agent are used as examples of the first wave of highly capable AI assistants, primarily used by people with deeper technical knowledge.

Consumer-focused personal assistants were also mentioned, including Grok Bot, Meta Muse, Instinct, and Poke.

ChatGPT's data analysis capability is mentioned as an example of a productized AI employee.