From Copilot Adoption to Copilot Fluency

Welcome and catching up with Cadie

Mark Smith 0:20

Welcome back to the show. I'm here with Katie today. She's from Microsoft. And for anybody that's seen my book, she's the one that forwarded the book on co-pilot adoption. She's been on the podcast in the past, but today we're gonna catch up and see where she is at now, what's changed. So, Katie, tell me about the last six to nine months. 2026 has been crazy already. What's top of mind for you? What's your focus area? What have you seen in the market?

From Copilot Adoption to Copilot Fluency

Cadie Kneip 0:51

Yeah, well, that's such a specific, interesting, specific set of time because I'd say up until you know two or three months ago, we were very focused on landing and adoption readiness efforts for agents. We were trying to land agentic AI at Microsoft. Um, and we had a lot of success with that. And we did a lot of uh creative programs with champions and with uh some of the gamification efforts that we've had. And then in the last two to three months, we've moved on from AI adoption um and into AI fluency. So we've moved, nobody gets left behind, right? We we don't build programs in a way that's like we're moving on strategically, so too bad if you missed it. Um but strategically, we're really focusing on that make it real, make that value uh personal uh for you with AI.

Mark Smith 1:49

So tell me a bit more, asin'. So uh for adoption, we talked a lot about you know productivity, how do you bring people on the journey um as an organization? Why is why is this move to fluency and what do you mean by fluency? Sure.

Cadie Kneip 2:07

Sure. So in my head, fluency is you know similar to if you think if you're fluent in English, right? It's can I use it natively, naturally, um without thinking. So we we did, I think we did talk about that previously, kind of the concept of making a habit um automatic. Yes. Um we use like I use the analogy all the time of um Lux go book a trip to Disneyland and you pull out your phone right away to check on flights. Uh and it's like you pulled out your phone right away. That was the first thing you did. You didn't go to a travel agent, you didn't do um something, uh, you didn't, you know, go find that uh that information elsewhere. So that's kind of automatic habit, automatic um action. That that's what fluency means uh to me. So when we look at AI fluency in an enterprise, in a business context, it's how do you make AI a fluent part of your workflow?

AI as the new default over Google

Mark Smith 3:07

So you know, one thing I've noticed is that I don't Google anything anymore.

Cadie Kneip 3:13

Yes.

Mark Smith 3:14

It's very rare. Like and and this can be something around that's very local to me. My first default is um I'll probably go to voice mode and I will ask one of three different AI that I have available to me. Um my question, because I think the biggest thing is I've trained it on my context. It knows where I live, it knows all my interests, etc., because we've built this kind of data set over time. And so therefore, it gets me a much better outcome than uh going to Google. And I uh a key a key point here is that because I'm in New Zealand at the bottom of the world, often Google in the past, if I asked a question, would give me something about the Northern Hemisphere or or you know, I I would ask for a product and it would give me an international location, which is I I would buy from it if it wasn't available locally, right? But I now I have all this lot local context and so it gives it gives it to me there. Are you seeing that move as well that when you if you're talking about what do you grab first, is that there's much more a move to, you know, if I'm gonna ask me about my health, I'm gonna open up Copilot to do that. Why? Because it's gated. I know Microsoft's not using it to train anything else. So all my personal, personal, personal stuff I actually use copilot for rather than Chat GPT or Claude.

Cadie Kneip 4:43

Yeah, that's so interesting. I love that. Um I I think so too. I think that there's something comfortable about knowing or trusting a product the way that people trust Microsoft products uh with Copilot. Um, but to your other point with um people kind of using AI in in different ways, um, I I've used that analogy with um web search for a while. And I I kind of think that um that that's a good tell because uh there are it it's just so interesting because there's such like Google is so synonymous with that web search. Yes. And so, you know, in how many years are we gonna move on from that meaning what it means? Um there's gonna be before we know it, probably sooner than we think of it, um, a a real change in how everything gets done. Um yeah, and I kind of do the same thing. I um I I definitely am using copilot and whatever AI uh you know tools uh that I have. That's what I reach for when I'm trying to get an answer for anything. Um and I love that you mentioned using your voice features. I think that's something that um uh is probably a little underused. Um, I think that power users are are really comfortable with that, but I'd love to see more people um give that a shot. I think that that is a good segue into um something that we see in research quite a bit with AI, which is um when people worry about AI or people worry about your human, you know, your day-to-day role as a human or your value as a human, um, you get a lot of sense of like point of view and identity and context when you talk something out. Um and I I think, you know, for for all that it's worth that using your um voice feature uh is a great way to is a great way to do that or include that.

Mark Smith 6:51

Yeah, I just saw that in um Copilot that brought out CarPlay integration so that it um uh works. And but by the way, when I say that about my healthcare data, it's because I own my own tenant. I'm a small business and I do that. I couldn't do that in a corporate environment. I'm not gonna say, hey, I've got this ailment, and uh I don't want to necessarily want the IT admin to see, you know, what that is. And so that's why I lean towards co-pilot in that situation where I control the ecosystem as such. But

Talking ideas out with voice AI

Mark Smith 7:21

what I find with the voice, and this is why I find it interesting with the car play coming out, is that I recently had to do a road trip to see a family member two hours away. My family wasn't coming with me. I had to go. That entire two hours was a conversation with Co-Pilot. Yeah, like I just I had it on and we just audioed backwards, and I just riffed all these ideas that I just hadn't had time to just talk out. And I find it to talk stuff out with AI, um it can reveal a lot about how you're thinking, but also give you uh in uh you know, you can get it to play the devil's advocate. Yeah. You can say, I'm thinking about doing this. Can you ask me 20 questions on it? Yeah. And I find that's and it's not that I'm asking it for an answer. I'm wanting to probe in a structured way my thinking or my thought processes.

Cadie Kneip 8:14

Yeah. Yeah. And I'm sure I do the same thing. And one of my favorite things to do with that is, you know, day-to-day role-playing a little bit and giving it the challenge of tell me why I'm wrong. Like I think that this might be the case, or this is what I'm feeling. Why doesn't this work? Um, because yeah, there there's a a the a thing where our AIs tend to say what we want to hear, I think. You're right. Yeah. And uh and so yeah, giving it the challenge to tell me why, you know, that's not true or or challenge it um or day-to-day role play as a certain, you know, port or a certain type of person, a certain leader or whatever, um, can help with that quite a bit. Um, but I I love it. I did I recently did the same thing, uh thinking about or brainstorming um what an AI fluency plan would look like, what people are challenging, uh uh, what challenges people are having with with AI. Um and yeah, it can get to be kind of a heated discussion when you're like just doing flow consciousness. Yeah, yeah. Yeah, it's it's great. I love it.

Mark Smith 9:19

Great.

Why people resist AI

Mark Smith 9:20

I've seen a number recently. 82% of Americans uh dislike AI or um or have a um a resistance to artificial intelligence. And um that can be anything from data centers, from a misperception or understanding about water consumption, um, to the impact on the electrical grid, that it would happen, you know, uh uh impact on local communities, blah, blah, blah. The list goes on. But 82% is a high number in America for anti-AI kind of sentiment.

Cadie Kneip 9:54

Yeah.

Mark Smith 9:55

Why why do you think this has come about? Um and and then when you're looking at fluency, how have you observed um jumping the barrier or or um dealing with that resistance? And last time I talked to you, one of the things we discussed is that somebody that has tried AI and said, ah, it's not for me, and they've walked away. And then what happened? It's often an appear, a colleague, that is doing something with it, and they go, Oh, how did you do that? Oh my gosh, that's amazing. I'm gonna try that. And that's how they're bought brought back in. How now that you're dealing in influency, how are you dealing with the resistance?

Cadie Kneip 10:37

Yeah, yeah. I I'll go back to the first part of the question um just because I thought that was so interesting. But I in my opinion, there's a lot of bad PR that's very loud. And it it's very popular, it's very clickbaity-y, it's very sexy to say like the the worst thing in the world about AI, and people are so polarized on everything. You can only be one thing or the other. Um, and I I find when I do too much research on that, like people saying, Oh, it's the it's like the industrial revolution and living in the industrial revolution sucked. Like, so make the place fuck. Not for everybody out of bed every morning. Um and and I get I find myself getting into a dark place with that. Um, but you know, just for example, yesterday I was hanging out with my brothers uh and they work for a sm a small startup that's that's AI based, um, and kind of doom saying things with them too, like, oh, I can't believe this. Like, and they were like, we don't have that perspective at all. Why don't you instead like focus on the fact that the advances in AI and the speed that we're finally seeing technology go means that we'll finally get to have, you know, amazing capabilities and technology that we haven't been able to have yet. Um and I I thought that changed my entire perspective. And it it's kind of like a something I think as people who are uh, I don't want to say pro AI, but like pro-advancement, pro things getting better, doing things better more efficiently, that it it's a little bit of responsibility on us to uh do a little bit of a gut check. What what's really going on? Where do the pros really outweigh the cons? Um and so that when we're talking about AI fluency and resistance to it, um it's kind of the same story where somebody can be like, um, I am not comfortable talking about agents, I'm not talking comfortable talking about uh reimagining my workflows with AI because this is my identity. I've been in the work service for X, Y, Z number of years, and I have success because I do things a certain way. Um and so I think that's where a lot of the resistance comes from is uh job market's scary, everything's so scary right now. I'd rather just not have things change. Um, and so

Selling the art of the possible

Cadie Kneip 13:20

I'm not a salesperson, but I feel like part of my job is selling people on the possibility of what could come, you know, from what their next day-to-day role looks like. Because whether they like it or not, or they come along with it or not, things are things are changing. Um so you had to put a positive spin on it in terms of the art of the possible, what could, what could be. And to me, that usually ends up looking like what's your favorite thing about what you do? And like, why do you think like at the end of the Um absolutely um so my day, Microsoft hired you for this job and not somebody else who can do the exact same 10 things? Uh, and what that boils down to is something that you can absolutely lean into and do better than you than the AI. And you can use AI to amplify that, to highlight that a little bit more, because you're gonna keep your judgment and you're gonna keep your point of view. That ends up being your biggest value at the end of the Um absolutely um so my day as a human operating an AI job, not the other way around.

Agents beyond chatbots

Mark Smith 14:26

Yeah. I are you seeing people are leaning in the work you're doing, and and I know you focus a lot in internally and inside Microsoft rather than externally, but are you seeing a shift to where people are really using the tools to remove kind of repetitive work that it's not the part they like in their job? And so I think of agents in this context, but agents can mean so many different things to different people, right? As in what is an agent. How are you seeing that that move beyond a chat-based experience into a do things experience with the tools?

Cadie Kneip 15:05

Ah, yeah, I love that. Um, yeah, I think especially at Microsoft in the last four to five months, we've really moved away from uh people generally hearing agents and thinking chatbot. Um, and a lot of that comes from um activities like hackathons or like you know, agent-athons, however weird way you want to say it, but workshops where we walk people through um reimagining their workflows is the title of the workshop I I usually run. Yes. Uh and we do. We we say, like, what's the worst thing uh about your job? Like, what do you do every week, every month, every quarter that makes you really hate it? Um, and then we'll we'll dissect it and we'll it kind of like like you do in a Six Sigma lean stuff a little bit, a gamble walk style, like reverse engineering of how things get done. Uh and then in my workshops, we'll take those, put it in a document, and give it to co-pilot and say, how can we optimize this with AI? Is there an agent? Is there a skill? Um, and then we'll build it. Because, you know, using co-ork and scout and things like that, I don't care if you're a builder. I'm not our I'm not a technical person uh by any stretch of the imagination. And I can build a lot of things um with the the AI tools that I have. Um so that's that's the way that the workshops have been running. Um, and that's a really satisfying, quick, painless way for people to uh see AI, the the value in it in their workflows. My example is always my favorite one. Um we talked about gamification a little bit last time, and we award people with curly badges. Uh and uh so I probably have five different um badges that I award to people. And that means I have five different forms that people fill out with however many different, you know, and it's a manual process and it takes hours and hours every week. And if people don't get their badges on time, I get nasty grams and I just it's the worst. Wow, that's amazing.

Automating Credly badges with Scout

Cadie Kneip 17:17

I can't stand it. And so um now I have it as an automatic uh system with with Scout. So it you could call it an agent. Um, but it's a a series of um, you know, clicks and activities and and downloads and uh data hygiene and all that that Scout runs for me automatically every week. Then all I need to do is download the CSV and upload it to Cradley and I'm done. Uh so it was, you know, hours and hours every week, plus just like a gross feeling in my stomach, like I hate that I have to do this, that is now like an afterthought, but has still has great impact, still something that people see value in and you know, you know, that I take pride in my workflow. It's just now I don't hate it whenever somebody brings up Credly badges.

Mark Smith 18:05

Okay, so it's interesting because on Cridly Front, I just built a tool because I I've got a Cridly account that we produce Credly badges, and they have an API, and I don't have anything to do with it anymore. On autocompletion, automatically issues a badge or seamlessly, just no drama at all. So

Scout inside Microsoft

Mark Smith 18:23

I don't use Scout, I use OpenClaw, I've been using it heavily for nine months, and I've got 26 agents, and but and that was one of the things that I needed done, and I built it out. You've mentioned Scout a few times, and you're the first person that I've talked to that have been using it. Is is it commonly being used a lot inside of Microsoft?

Cadie Kneip 18:44

Good question. Great question.

Mark Smith 18:46

Um and it just in your purview, just in the folks that you're working with, do you see it being used more and more?

Cadie Kneip 18:52

So in Copilot Champs land, yes. Yeah. We've got 10,000 copilot champs, and I ask them for examples every Um absolutely um so my day. Not every Um absolutely um so my day, but you know, I ask them for examples all the time. Uh, and a lot of them are using Scout in ways I would have never thought to. Um, you see different use cases pop up depending on somebody's day-to-day role, job, their the amount of uh experience that they maybe have with Visual Studio and GitHub versus somebody like me who's not tech not technical. So I use it for vibe coding quite a bit. Um now I can I can basically build anything I can think up, right? Yes, yes. Um and I'll sometimes have a co-pilot window open right next to it and have copilot go back and forth with Scout and say, Scout said this, now what do I do? You know? Yeah, yeah. That's all my context. Um, so I I think it's fantastic. I and I I think it's a heavy, heavy hitter. Um and like you, I had tried uh open cloud probably for four months privately um before we tried Scout. Um that's impressive. And I you know, I I think it's I think it's great. I think that using Scout in um the Microsoft Suite all up. I mean, we're talking about work IQ, context, all these things that just it really makes you feel like you can accomplish any task that you can think of uh at work.

Scout vs. Cowork

Mark Smith 20:17

How do you divide between using Scout and co-work?

Cadie Kneip 20:21

That's a great question. Um I I so I'll just say personally, um, because I I'll be honest, I'm trying to build out slides that answer this question for people from a guidance perspective. Um and I struggle memorizing what those correct answers are. Um because I think every Um absolutely um so my day both products feel different. Yes. But I'll I'll just give it a real example of when I use uh co-work versus when I use Scout. Um Scout is much more, in my opinion, of a repeatable execution tool. Um I feel more comfortable vibe coding in there. I love that it's connected to GitHub. Um and and there's there's a lot to get done uh with Scout. Um a lot of times, like if I post something in the Copilot Champs community and I get 30 comments in it in my team's channel, and I want to engage with everybody, but I literally can't, I'll open up Scout, I'll have Scout draft all those answers, approve each one, and it gets done. It's it's it's if you really want that true chief of staff, executive assistant, full of consciousness, you know, experience with co-pilot um or or with AI uh in the Microsoft Suite, Scout's your tool for that, in my opinion. Um and then cowork to me is when I want to build something where I really need my work IQ. I really need all my context windows to be strong. Um, I just I I'm not sure exactly why, but that uh everything that's like context, back history. uh related, um finding better, stronger results in cowork.

What Work IQ is

Mark Smith 22:04

Yeah. Can you just um just uh for the listeners talk about what work IQ is?

Cadie Kneip 22:10

Oh yeah. So uh work IQ is uh co-pilot or AI tapping into the graph. Um so we're talking if um it when you're using cowork or co-pilot um using that forward slash um before something to call on something in your graph call on something within the context of your work environment um so forward slash um Mark Smith on my team is um asking me about forward slash this project yes tell me how to respond to that um so that's how you call on things for specific context um within Microsoft and to me it's um it's so powerful it it brings in elements of um of of work in different ways um that truly feels like a relief of cognitive load um you know i if we've talked about like different learning styles quite a bit I I obviously have a different learning style than a lot of people um and we do we just have too many things going on so it is so helpful and so powerful to have a tool that can help with that recall. Yeah. Yeah.

Mark Smith 23:28

Okay that's that's interesting. For those people that don't know what the graph is it's basically all your digital uh exhaust around whether you answer an email, when you respond to an email, who's in the chain of that email, what the timestamp was, et cetera, all can be fed in and give context to your AI um uh from that. That's really interesting. Last a question on cowork that just popped into my mind is that you know I used cowork when it first um was in beta

How Cowork has evolved

Mark Smith 23:54

um and it was very anthropic right it was very much a copy of anthropic and I'm pretty sure it was just a skin of anthropic. In recent times I feel that Microsoft built it re-engineered it and it's not just Anthropic as the only um model behind it. It seems to be much more all the models that Microsoft provide through the copilot um interface. Is am I correct on that observation? Because it is only an observation that I've made that it seems to be like built now it's not just a skinning of Anthropic's product and like if all the major providers have a cowork tool now.

Cadie Kneip 24:33

Yeah yeah and I think it's really the best of a few different worlds and yeah the experience feels totally different. A lot of people didn't love the version of cowork that was originally released about a year ago. And I if you tried it then and you hated it and never went back, go back today. Like it's it's going to do what you always hoped that you know AAI would do uh in in the context of co-pilot. Uh and it's it's great. It is the best of all of them. And um I don't know if everybody has it yet but now you do literally have the option to select what um model you're using um within co-pilot or cowork. And actually in Scout uh also and then the the level of participation you want in it. So there's a lot of I'd say you know I don't want to speak in generalities or too but in my experience I would say I've observed that there's been a little bit of um we we iterate something, we release something new in our product and it feels like it's another way of giving more power in the hands of the user um which which is really really great with AI because um you're you're giving some kind of power away but you're relinquishing some of that in terms of uh sure help to help me figure out how to get something done but I have a little bit more control over how it gets done, um, what the AI is doing, you know, um some of that power gets kind of like balanced in in that sense. So I wouldn't be surprised if as um Microsoft continues to make updates that we see more options, more things come out that feel like um user control uh customized.

Auto vs. choosing your model

Mark Smith 26:25

Yeah so so let's give the classic one auto for model selection um it's funny because you know my wife Meg she is an or uh auto is what she goes for. Me, I never use auto. I am always selecting I want to know what my model is because I know what's going to give me you know the results are it needed. She recently had to do a governance paper and um she had been doing questions backwards with auto and I said oh why don't you just do a a review of what you've written but use Claude um to do it and she was just like you know as in the the the um it played the devil's advocate with her much harder than using auto. Are you an auto or do you like to select your models or do you really use a blend?

Cadie Kneip 27:14

I do use a blend and I'll start with auto because that's what it defaults to for sure. And then it just if I notice that I don't that something's not what I like then I'll I'll switch it. Most of the time I'm finding myself switching um if it's running low medium or high and I just high is my sweet spot forgetting what I want with code 100%. Yeah so but I'm you know uh it so yeah um and then I you can always look it up like you can always ask Copilot or ask ChatGPT like I'm running XYZ type of workload. I'm running something in Excel uh which model is going to be the best for what I want to run I'm running something where I need to build a 30 page deck what's the best model for that um and the answers might change like pretty frequently to the same question. But sometimes like if I'm really stuck and not getting the results that I'm really hoping for, um I'll ask which one I should use.

Mark Smith 28:13

But yeah do you think

Microsoft's own models and image generation

Mark Smith 28:17

do you think that um in time we might see Microsoft's own models appearing in there and I'm not asking you for insights or direction but may I type models because I'll give you a reason why I'm asking is that uh I think Mustafa Saliman recently posted that I think the image model from Microsoft has outranked everything um on the image scale of ranking and I'm like um I wondered therefore is Microsoft gonna start allowing that to be a model that we could pick as well.

Cadie Kneip 28:52

Yeah yeah uh yeah I I haven't heard anything um but I also wouldn't be surprised and and speaking of the the image one I think last time we talked I said that I'm a big user of Mid Journey uh for images um and you know paid the the highest subscription there for a long long time and then in the last week or so I've started using Copilot image generator more and more and it's it's great.

Mark Smith 29:19

I I I canceled my subscription to Midjourney some time ago other things are just doing everything that I needed right so good.

Cadie Kneip 29:26

Yeah that's I mean for somebody like creates tons of images um and uh yeah kind of like picky about the things that you know that can be I I definitely was surprised when I started being like oh co-pilot's doing it better. Um so yeah I I don't know I I think uh anything to keep things moving and keep uh whoever is building any model competitive to make uh better models again and again you know the more the merrier from that perspective.

Mark Smith 29:55

Yeah yeah just pulling back to to fluency AI fluency

What makes Copilot fluency stick

Mark Smith 30:02

are there any lessons learned now after you know you've been two two three plus years down the space of what really lands for people that that are using it as information workers in the context of work yeah like if they get the skill you're like you know they're off to the races. If they get beyond this hurdle you know they're gonna be fine they're gonna be self-selecting and so as they'll they'll um yeah their default will be grab the phone is you know the Disney example you gave where where do you see that tipping point happening?

Cadie Kneip 30:36

I think up until like a year ago we would say um the answer is when somebody's like really good at writing a prompt or feels like fluent in in writing their prompts but now that doesn't matter so much. And if anything learn how to use prompt coach and just you know figure figure it out but uh absolutely prompting is not that measurement anymore. It gets to be more of a thing with are you comfortable with agents and are you comfortable with skills? One or the other really um and I'm trying to think of like where we really see people kind of get through that mindset shift that's kind of it. And it does kind of start with like can you build an agent with agent builder and copilot um you know like a sp a very simple chatbot style agent like can you do that and feel proud of what you've done. That's usually like the spark uh for a lot of people is like that was really easy like building something just using one prompt in agent builder is very very simple and very satisfying. And I also think that like in that back end view if if folks haven't been in the back end of agent builder in M365 Copilot, it's I think very eye-opening to see oh these are these plugins and this goes here and this is what it's doing with my prompt to make it happen. I think that demystifies the entire agentic experience a little bit um and we see people graduate from I'm gonna build something simple with agent builder to I'm gonna build um something more advanced and now I want to I want something powerful in Copilot Studio and then very quickly I need something more powerful in in GitHub or something that that's at a pro code level. Those are kind of the stages that we see people go through. And um it's a lot of the time we talked about those workshops a little bit a lot of time it is talking or walking somebody through that experience of just coming up with a use case that makes sense and then showing them how simple it can be. And yeah like for me in particular I know um I uh was not comfortable with like the concept of uh you know using or building an agent definitely not building um until I saw like that that back end view in agent builder and was like oh this is this is what it's doing with my prompt. I get it now.

Mark Smith 33:10

Interesting.

Cadie Kneip 33:11

So yeah there's a little bit of demystification that needs to go on. Sure.

Predictions and excitement for the rest of 2026

Mark Smith 33:16

We're we're three months for four months before the end of the year 2026. I think it's I feel it's been an absolute hectic here in that I think we're going to uh you know new models and stuff dropping every two weeks Dow, if not more frequently we've seen you know the open weights model um dominate the the news cycles we've heard a lot about data centers. What are you most excited about for the for the rest of this year, just for the next four months and and and are there any predictions that you see changing or happening?

Cadie Kneip 33:49

I will I'm very excited to get my team in as AI fluent and I want to kind of get my team through it and kind of check them all as AI fluent meaning that they've got at least one workflow uh that they've uh I don't want to say agenfied but improved with AI. And then I want to see that scale across Microsoft and then I want to see other people use that model and become AI fluent in some capacity. Then I want to see or I'm excited to see what people do with that time, with that efficiency, with what's been given back to them because the heavy cognitive load things, you know, the the tedious stuff has been offloaded. And for people who have been doing the same job for many years in a the same way that's one big mystery is what are you going to do now? And there's a huge opportunity uh to like you know we say so much, but to to lean into the the human value, to have a point of view, um to uh to do things in a way that your judgment uh you know shines through and so that's what I'm the most excited for is to see like what what people what comes out of it right I mean obviously we all want to see the Um absolutely um so my day that AI can uh fold up my laundry and do my dishes for me. Like if we could have that by the end of the year I would be so stoked. But in the meantime what are people gonna do? Uh is just incredible. It's so exciting.

Mark Smith 35:26

How soon do you eat your first robot?

Cadie Kneip 35:30

I mean we all have like the the the mops right the the roombas and things uh so I I'm the kind in my husband too my husband's such a tech guy in a gearhead uh the second that one's available we're getting it um but you know in terms of like when does that hit the market I'm I think before you know it I think we're gonna see something yeah Optimus from Elon will be surely available. I think it'll be yeah I don't yeah no but I think you know we're gonna see stuff come out in December like we're gonna see Christmas presents that that are things that we wouldn't have thought of before yeah and that's one of the exciting things with the the pace of change.

Wrap-up

Mark Smith 36:12

Yeah. Katie thank you so much for coming back on the show again.

Cadie Kneip 36:15

Oh my god thanks so much for having me good to see you.