P9: Survey Data and Session Summary

Survey Responses

Background

P9 is an independent UX researcher and AI specialist with more than 25 years of professional experience. She holds a master's degree and runs her own consultancy, two podcasts (one focused on AI for good), and a nonprofit that provides free AI education. She is also a fellow at the Conversation Design Institute. Her entry into AI predates the generative AI wave: she began with NLU and NLP through conversation design four years before ChatGPT launched, and was among the earliest ChatGPT adopters when it arrived.

P9 is one of the most deeply integrated AI users in this study. Her daily workflow begins with an 8 a.m. planning brief with Claude that connects to her calendar and a system of roughly 40 Claude projects. She has replaced multiple SaaS subscriptions by building her own tools with AI-assisted coding, using a stack centered on Whisper Flow, Granola, Claude Code, and VS Code. She draws a sharp distinction between "vibe coding" (which she associates with tools like Lovable and with users who lack development understanding) and "AI assisted coding," which she learned through guidance from developers in the Agentics Foundation.

What makes P9 distinctive in this study is the tension between her deep integration and her active discomfort with it. She describes herself as "monopolized" by Claude and Anthropic products and is developing a "Plan B" strategy for vendor lock-in. She also identifies as neurodivergent (ADHD) and frames AI not as a productivity tool but as a cognitive prosthetic that closes cognitive loops and manages the multiple simultaneous "streams of consciousness" she experiences daily.

Key Findings

The Cognitive Prosthetic: AI as ADHD Management

P9 provides the most vivid articulation of the AI-as-cognitive-prosthetic pattern in this study. Where P6 described AI compensating for a general cognitive tendency, P9 names a specific clinical condition and describes a precise mechanism: she experiences "six streams of consciousness" simultaneously, and Claude projects give each stream a persistent container. Externalizing an idea into a project "closes the loop" in her brain, allowing her to move on without the cognitive burden of holding it in working memory.

She places Claude in the same category as exercise, not alongside productivity software: both are non-pharmaceutical coping strategies for her neurodivergence. The 8 a.m. planning brief replaced "giant stacks of Post-its with reminders and lists" and integrates with her calendar so Claude can block time and manage her schedule.

"That gift of being able to get [my thoughts] out, put it in a suitcase, know it's safe, and visit with it whenever I want to work on that."

The Organization Feedback Loop

P9's AI-powered organization system has produced an unintended consequence: a false sense of capacity. Because Claude keeps her organized across calendar, projects, and filing, she feels capable of taking on more commitments than she can sustain. She runs a consultancy, two podcasts, free classes, free mentoring, social media across multiple platforms, and Agentics Foundation meetups. She recognizes that she's saying yes to too much.

Her response to this overcapacity is itself AI-mediated: she has Claude block time on her calendar. This creates what she calls "a little bit of a loop," where AI simultaneously enables overcommitment and serves as the corrective for it. She has also built an assessment tool called "AI for Your Type" that evaluates a user's neurology, psychology, and AI usage to generate personalized recommendations, including warnings about overuse.

"I think it gives me a false sense to say yes too much because I feel so organized and I'm like saying yes to way too much."

The Plan B: Vendor Lock-in as Personal Risk

P9 is the first participant in this study to articulate vendor-specific dependency as a personal strategic risk and to be actively building a mitigation strategy. She describes herself as "monopolized" by Claude and Anthropic products and is writing about "what's your plan B" as a public-facing piece.

Her concern operates on two levels. Operationally, she notes that Claude Code goes down periodically and asks whether that should stop her workday. Economically, she projects a scenario where pricing escalates to $500 or $800 per month and asks what happens to people who have built their workflows around it. This second concern connects to her broader worry about an AI-driven class divide.

"I'm so integrated with Claude and Claude products and Anthropic that I don't like it. I feel monopolized."

The Interview-Then-Arrange Workflow

P9 has developed a specific, replicable technique for creating content that preserves her authentic voice while leveraging AI for structure. She asks Claude to interview her on a subject, provides her unedited answers, and instructs it to arrange them into a LinkedIn post without changing the words. She then asks Claude to identify knowledge gaps in what she said, turning each content creation session into a learning opportunity.

This workflow is notable because it inverts the typical generative AI content pattern (AI writes, human edits) into a pattern where the human provides the substance and AI provides the arrangement. The result is content she can stand behind as authentically hers while still benefiting from AI assistance.

"I will say, 'Hey, Claude, I want to write a LinkedIn post on X subject. Interview me on the subject.' And then take my answers, do not change them, and turn it into a LinkedIn post."

The Chromebook Precedent

When discussing AI in education, P9 draws a historical parallel that no other participant has raised: the decline in student competency that followed the introduction of Chromebooks in schools. She uses this as a concrete precedent for her concern that one-size-fits-all AI implementation in public education will produce similar or worse outcomes, particularly for special education students who need differentiated approaches.

She wants to move her consultancy into public schools, arguing that teachers currently have no governance around AI use, and that governance needs to be specific to teachers, administrators, and student populations respectively.

"The studies, as soon as kids' competency, everything, everything tanked as soon as Chromebooks entered the picture. And now we want to talk about kids using AI."

Emerging Themes

Interview Transcript

00:00:00

Paul: I'd like you to tell me the story of your first "oh wow" moment with AI. What was going on that made you try AI? And what happened that made the light bulb turn on for you?

P9: It's an interesting one. I have a story that I tell about this all the time. So I was working UX/UI. I mean I started in 2001 with websites and graphic design. Used that throughout my career. Had a business, worked in restaurants, had my husband, blah blah blah blah blah. And then one day I was working, I came back and I'm working in UX and I'm learning and I'm doing, and somebody who ended up becoming a mentor of mine, [REDACTED: mentor name], had posted about, this was four years ago, pre-ChatGPT. I forget if the post was about AI or conversation design, but the post, I swear, I had one of those moving moments where the lights came up and the angel sang and I said, "She's talking about the future of everything." And the next day,

00:01:07

P9: so it was, I felt like it was like a divine instinct. The next day, I was learning everything I could about conversation design, NLU, NLP. So when people talk about AI now, it's always generative. I didn't start with generative. I started with NLU NLP.

Paul: You might know my wife Susan Hura. Her name might be familiar because she's in that space.

P9: I have to see her little LinkedIn picture.

Paul: Yeah, I'll send that to you along with the focaccia recipe we spoke about.

P9: Yeah. So literally the next day I was in the Conversation Design Institute learning everything possible, and now I'm actually a fellow at Conversation Design Institute. So that's a testament to my grind, my ability to work 80 hours a week and be happy. And so I started down that path. I started learning everything, intent and utterance and all the NLU NLP stuff.

00:02:07

P9: Started doing freelance, like trying to figure out how to build those old chatbots that were pretty awful. And then ChatGPT hit and I was on it the first day, because I was in that circle of people who were like, "Oh man there's this thing, it's starting, you know, tomorrow." I was in that, you know, they send you that ranking with OpenAI and I was in that like 0.01% first users because I was ready and I took to it like a fish to water, which, this is interesting research. I think neurodivergent people are much more successful in this space than a lot of others.

Paul: What were you trying to do with it for that first light bulb moment?

P9:

00:03:15

P9:

Paul: What about, fast forward to today, what tools have you tried and left by the wayside? And what tools are you using now the most?

00:04:12

P9: So I am lucky enough to be part of the Agentics Foundation and these are like the greatest agentic developers on the planet in my opinion. So I have been very lucky to be guided by them tool-wise, but I'm also part of the design community and a lot of the design tools are absolute garbage. I'm not a fan of Lovable. The only one I would say I could get behind is probably Cursor, but I don't use Cursor, because Cursor is the closest thing I think to being a real coding tool versus Lovable which is a vibe coding tool. And there's a big difference between vibe coding and AI assisted coding. And I have been lucky enough to have been guided by these developers from early on to learn how to do AI assisted coding versus vibe coding. You know, vibe coding is for people who don't understand development.

00:05:17

P9: So yeah, a lot of those design tools I have left behind. Figma Make was a joke. Lovable is not a professional tool. There are a whole bunch of others in the mix there. My stack now is Whisper Flow, Granola, Claude Code, VS Code. I run Claude Code in VS Code. And that's, I don't want to say that's it, but that's the main pieces of it.

Paul: I'm a little curious what Granola does because I'm not familiar with that. I know Whisper Flow is speech to text, right?

P9: Yeah. Granola is not a notetaker, but no.

Paul: Oh, I actually need one for these meetings because Google Meet does not do a good job.

P9: And Granola does not join. It only works on Mac, I believe, but it doesn't join the meeting. It runs in the background.

00:06:16

P9: So, you have to be mindful to say, "Hey, I'm recording." Because there's no reminder in the window. It's a UX issue that I actually brought up to them and said, "Hey, you need to remind people to say, hey, you know, I'm ethical and I'm recording." But it is awesome. And then it also allows you to send the queryable link to the person you met with so I can send it to you when we're done. And you could get a little practice run in with it.

P9:

Paul: That's interesting that you're planning for the base LLM part of your stack to be replaceable, no one's mentioned that yet explicitly. I appreciate you mentioning that. What's your big concern around that? Other than it going down, is it gouging and lock-in, vendor lock-in?

00:07:21

P9:

Paul: Great. I would love to, and I'm learning that 30 minutes is probably not enough, but as we talked about, I'm going to try to stick to 30, 35 minutes and I'm going to move to the next question.

00:08:23

Paul: Think about one thing that you do regularly, it could be at work or in your personal life or a blend, that AI has changed the most. And then walk me through what you used to do versus what you do now.

P9:

Paul: Does it, do you ask for it or is it ready kind of in your UI waiting for you?

P9:

00:09:21

P9: It's just I hate going into the main conversation with Claude and it not knowing what's in the project. So like every month I'll go into the main projects, have them create a summary, put it in a document and upload it to Claude and say this is everything I'm working on. So it has context. So that's a little bit of time and energy, but really instead of like giant stacks of Post-its with reminders and lists, I just have my morning brief with Claude. But also I think the most important thing though is now my ability to cope.

Paul: So, this sounds like it's a lot better for you. Is there any other effects that are unexpected, maybe not worse, but is this increased ability to cope and organize with your 8 a.m. briefing causing any unintended consequences?

P9: I think it gives me a false sense to say yes too much because I feel so organized and I'm like saying yes to way too much. I mean I have a consultancy.

00:10:27

P9: I have two podcasts. I do tons of free classes. I'm on TikTok, YouTube, LinkedIn. The Agentics Foundation meetups. I do free mentoring. Like so I have to stop saying yes so much. But what I've been doing is having Claude block time on my calendar.

Paul: Interesting. So, not only is it bringing about some of your tendency to say yes, you're also using AI to help ward off that tendency with this calendar block.

P9: A little bit of a loop. Yeah. But when people are like, "oh you know AI tamps down creativity," I actually created a tool called "AI for Your Type," an assessment tool. It assesses your neurology, your psychology, your AI usage and then it makes recommendations. Like, you're lonely, you should not be using AI as a partner, as a girlfriend or a boyfriend. Or for me it's like, you're using it to do too much.

00:11:46

P9:

Paul: I didn't mean to laugh. But that was funny.

P9:

00:12:47

P9:

Paul: What's been your biggest win with AI in your work life so far? And on the flip side, what's been the biggest disappointment or surprise failure?

P9: The biggest, I think, I'm trying not to think financially. I'm trying to think, the I don't know if it's the biggest win, but one of the things I love the most is if I need a tool, I can build it. I'm no longer dependent on Mailchimp or SaaS for this, SaaS for that, $20 here, $50/month, that's gone. I build whatever I need now and I can customize it to my weirdness. That is a gift, right?

00:13:52

P9: Like this week alone, I built my own version of Mailchimp and they actually reached out to me because I went on their page and I'm like, "What is the, why won't this work?" And they reached out to me via email and by the time they reached out to me I was like, "You know what, this is going to freak you guys out but I'm cancelling because I built my own version of Mailchimp, does everything I need to do and it's free." They didn't respond to that. But that's the SaaS apocalypse. That's coming for everyone.

P9: The biggest loss is I actually started writing a book called The AI Hypocrite. And it's all about me and my internal struggle with the water usage, the consumption, the electricity, the job losses. Trying to make a difference from the inside is easier said than done. And a lot of those anti-AI people troll me and they're like, you know, there's no such thing as being an ethical AI consultant.

00:14:56

P9: There's no, and I'm like, listen, man, you make your money from being an extremist. There's a lot less money to be made by trying to be a voice of reason in the middle ground. But that's where I am. Why don't you go troll like some bro asshole and not me who's trying to reach the masses, teach ethics, teach how to use it, pull women into the space and teach, teaching for free. I have a nonprofit that teaches people. Like I'm the last one you should be trolling and yet here you are, you know. So it definitely has brought out the trolls and that's a negative.

Paul: That's interesting. I might circle back to that, but I want to make sure I cover this one, which is everyone has had an experience where AI outputs something that's just not true. How do you decide when and whether to trust what AI gives you? How do you operate and course correct when you find an obvious hallucination?

00:16:01

P9:

00:17:06

P9:

Paul: Do you see norms being developed around disclosing when people have used AI in the organizations and people you talk to?

P9:

Paul: I like that alliteration.

P9:

00:18:13

P9: When I, I've spoken so much about my process, I don't advertise it on LinkedIn every time I post because I use the same process. It's public. It's in my featured section. And I teach it.

Paul: I like that technique.

P9: Yeah. Go on, my featured section is a treasure trove. But also voiceofai.io is my website and it's a lot of the same stuff just put in a really nice tidy fashion. And then what I'll do too is say, "All right, now where's the knowledge gaps with what I said to you?"

00:19:16

P9:

Paul: Very cool. So let's zoom out for a second. How does this increasing presence of AI everywhere make you feel?

P9:

00:20:33

P9:

Paul: Do you worry about losing certain skills for yourself?

P9: No.

Paul: Do you worry about yourself losing skills because you're using AI?

00:21:40

P9:

00:22:41

P9:

Paul: And this is related to, you know, you're talking about balance. And we, I've heard a lot about fears and concerns very similar if not identical to ones you just expressed about the next generation, gaps in education and wealth and gender creating divides. What do you think's the big promise of AI? What's the big breakthrough? It doesn't have to be in medicine or science, but what's the promise that you're kind of hoping for in your most optimistic moments?

P9: Exactly that, science, medical. I have two podcasts. One of them is about AI for good and we had a bunch of environmental people on.

00:23:32

P9: As much damage as it does, it can be used to fight back against that damage. But really, you know, it comes down to greed, people, ego, corporation, you know, like everything always comes back to how are we going to use it and how do we teach people to use it for good. It's similar to guns really, right? Like it can sit on the table 400 years. It's the person that picks it up. And to me, everything comes down to, like, so much of the bad actors come down to greed in one bucket, but also that other bucket of people burning down $25 million warehouses because that company that makes bazillions of dollars isn't paying them a livable wage. If it wasn't so hard to keep yourself fed, clothed, and housed, I think people would be very different than they are right now.

00:24:41

P9: So, I think a lot comes back down to that.

Paul: What's the biggest gap between what AI can do for you right now and what you actually need it to do? And what would it take to close that gap?

P9: I'm working in the space that I work in, agentics. So I want to be able to say to my phone, "Hey, get me some leads," or "get me some airline tickets for me and the kids," or, you know, I want that concierge service. There is a huge tradeoff for that and that's data. So yeah, I think a lot of people are going to be willing to give that data for that convenience. Mother of three, works a bajillion hours a week, that concierge convenience service is, I think, going to be beautiful for me.

Paul: Great. That's a great place to stop. I'm going to stop the recording now.