In this episode
Kaz Ohta, co-founder and CEO of Treasure AI, says AI safety is an engineering problem: enterprises keep autonomous agents on-brand by starting them at zero permission, auditing everything they do, and running them on governed customer data. Recorded live at Treasure AI’s Agentic World 2026 in Miami, Ohta and Greg Kihlström cover why Treasure Data rebranded as Treasure AI, why Ohta expects most users of the platform to be AI agents within about two years, why Treasure AI built Personalization Studio to close the loop between data and activation, and what changes for marketing teams when people, not software, become the bottleneck.
Key takeaways
- AI is commoditizing faster than electricity or compute did. Ohta says what costs $100 to do with AI today will cost $1 next year, while the intelligence keeps improving.
- Top-of-funnel discovery has moved into AI. Brands Treasure AI works with have seen web traffic decline; Ohta says the first search result once drew 20–30% click-through and now draws under 10% because AI overviews answer most queries.
- Consumer AI is moving from search to action. Assistants now find unused subscriptions, book movie tickets, and draft email, and Ohta argues the only way an enterprise can handle billions of consumer assistants is with AI of its own.
- 90–95% of implementing an AI agent is clean data and context. Building the agent gets the attention, but governed customer data is what makes it work.
- Most platform users will soon be AI, not people. Today 90–95% of Treasure AI’s users are humans building segments and journeys; Ohta expects about 90% of platform access to come from AI within roughly two years.
- Real-time agent decisions favor consolidating data and activation. Customers can keep best-of-breed tools, but Ohta says they pay an extra tax to make separate vendors close the loop.
- Start every agent at zero permission. Treasure AI runs each AI thread in a secure containerized environment, traces every file and network access, and keeps credentials behind a gateway the AI never reaches.
- Faster AI output makes humans the bottleneck. Treasure AI’s engineers now produce three to four times more code than six to nine months ago, so the company moved code and security review to AI and redesigned how it ships.
- You own what your agents produce. Ohta warns against a new kind of laziness, passing AI output along without review, and says people without AI will be replaced by people with AI.
Chapters
- 00:48 — Live from Agentic World 2026: are AI agents still deciding the way your brand would?
- 02:15 — Kaz Ohta’s path from CTO to CEO, and from Treasure Data to Treasure AI
- 03:04 — AI is commoditizing faster than any technology in history
- 05:36 — Why web traffic is falling: discovery moved into AI
- 06:28 — Consumer AI shifts from search to taking action
- 08:04 — Why the value moves from data to decisioning
- 09:38 — Does the CDP category survive?
- 12:36 — Personalization Studio and closing the loop between data and activation
- 16:10 — Governing autonomous agents with humans in the loop
- 19:12 — When AI speeds up the work, people become the bottleneck
- 22:31 — People without AI will be replaced by people with AI
- 23:55 — What Ohta wants to learn at Agentic World, and how he stays agile
Why Treasure Data became Treasure AI
Ohta started the company 15 years ago as CTO and became CEO five years ago. It began as a data platform, became a CDP company, and rebranded as Treasure AI in April. His reasoning: consumer AI is moving from answering questions to taking actions, and enterprises will need AI of their own to respond. After 15 years of collecting customer data for brands, Treasure AI holds clean data with governance on top, which he says is exactly what AI needs. The next source of value is the decisioning that runs on that data.
Why brands are losing web traffic to AI
Ohta says top-of-funnel discovery has moved into AI. People ask ChatGPT for product recommendations, and AI overviews in search answer most queries without a click. He puts the first search result’s click-through rate at 20–30% a few years ago and under 10% now, with lower positions faring worse. Brands Treasure AI works with have seen the resulting traffic declines over the last few years.
Is the CDP category going away?
Ohta expects CDP to become a subcategory of something broader as CDP, CEP and cloud data warehouse blur together. He is not sure the label survives five or ten years, but the function does: enterprises will need governed customer data somewhere, because that is what powers AI agents dealing with billions of consumer assistants.
Why consolidate data and activation for real-time agents
Treasure AI tries to stay vendor-neutral and maintains its partner connections. But when AI agents make decisions almost in real time, Ohta says data and activation have to work together in real time too, and that is hard across two vendors. Customers can choose best-of-breed and pay an extra tax to make it work, or combine Treasure AI’s CDP, Engage, and Personalization Studio so every customer interaction runs as a real-time closed-loop AI decision.
How to keep autonomous agents under human control
Ohta treats AI safety as an engineering problem and says you should always start from zero permission, then gradually add capabilities. Every AI thread runs in a secure containerized environment that customers control, every file and network access can be traced, and a gateway keeps the AI from ever touching credentials. His thesis is that AI has to stay under human control, with every switch and knob available to people.
What changes for teams when agents do the work
Treasure AI’s engineering team now produces three to four times more code than it did six to nine months ago, and human review of every line became the bottleneck. The company moved code review and security review to AI and rethought how it builds and ships. Ohta expects the same shift in every department: your job is to orchestrate your agents, and you are responsible for every output they produce. He also wants more room for younger, AI-native talent, citing an 18-year-old intern who built an agent in four hours that cut a four-hour internal email task to 15 minutes. Treasure AI then productized it.
FAQ
Why did Treasure Data rebrand as Treasure AI? Kaz Ohta says the value is moving from collecting customer data to the AI decisioning that runs on top of it. Treasure AI holds clean, governed customer data, which he argues is what AI agents need to act.
How much of an AI agent implementation is data work? Ohta estimates 90 to 95% of the work in implementing an AI agent is getting clean data and context in place. Building the agent itself is the smaller part.
How should enterprises govern autonomous AI agents? Ohta’s rule is to start every agent at zero permission and add capabilities gradually. Treasure AI runs each AI thread in a secure containerized environment, traces every file and network access, and keeps credentials behind a gateway the AI never touches.
Will the customer data platform category survive agentic AI? Ohta says CDP may become a subcategory as CDP, CEP and cloud data warehouse blur together. He is not sure it will still be called CDP, but enterprises will still need governed customer data somewhere to power their AI agents.
Why combine customer data and activation in one platform? When AI agents make decisions in near real time, data and activation have to work together in real time. Ohta says customers can still use best-of-breed tools, but pay an extra tax to make them work together.
How do marketing teams need to change as AI agents take on execution? Ohta says keeping existing processes turns the human team into the bottleneck, so leaders have to redesign how people and AI work together. People orchestrate the agents and remain responsible for every output they generate.
About Kaz Ohta
Kazuki Ohta is the co-founder and CEO of Treasure AI, an AI-native platform built to enable customer data and agentic experiences. He co-founded Treasure Data, now Treasure AI, in 2011 and served as its CTO during the company’s early development, helping scale the business to more than $100 million in annual recurring revenue.
Kaz Ohta on LinkedIn: https://www.linkedin.com/in/kazukiohta/
———- Resources ———-
Treasure AI: https://www.treasure.ai
The Agile Brand podcast is brought to you by TEKsystems. Learn more here: https://aglbrnd.co/r/2868abd8085a9703
The hardest part of martech right now is telling which tools are actually earning their keep. That’s the question running through the MarTech Summit Boston agenda on October 27. We’re attending MarTech Summit Boston on October 27 as a media partner. If you want to join, the code AGILE15 saves you 15%. https://aglbrnd.co/r/c70c250b8d52dff3
We’re proud to be a media partner for #MAICON26 – Oct. 13-15! Learn how AI can power your marketing and business and help you grow smarter. Use code AGILE150 to save! https://aglbrnd.co/r/7fe458ced0f04658
Need to evaluate martech, AI, or CX platforms and don’t know where to start? The Agile Brand Guide List is now in beta. We’re building the most useful, free-to-use, and up-to-date way to assess what truly matters when evaluating platforms. See more here: https://aglbrnd.co/r/f76be3a33cc491bb
Reach your customers with Reddit. Spend $500 in ad spend, get $500 back in ad credit! Learn more: https://aglbrnd.co/r/491818c79fb1873f
Chaser is the only Slack-native project management platform that helps teams turn messages into tracked tasks, automate follow-ups, and maintain team-wide visibility, without adopting another tool. Now integrated with Claude and other GenAI tools. Learn more at trychaser.com and use code AGILEBRAND for a 3-month free trial (normal trial is 14 days).
Start building your own apps with Replit and get $20 off. Learn more: https://aglbrnd.co/r/93531742a7625a20
Enjoyed the show? Tell us more and give us a rating so others can find the show: https://aglbrnd.co/r/faaed112fc9887f3
Connect with Greg on LinkedIn: https://www.linkedin.com/in/gregkihlstrom
Don’t miss a thing: get the latest episodes, sign up for our newsletter, and check out Martechipedia, the wiki for marketing technology: https://aglbrnd.co/r/35ded3ccfb6716ba
The Agile Brand is produced by Missing Link—a Latina-owned strategy-driven, creatively fueled production co-op. From ideation to creation, they craft human connections through intelligent, engaging and informative content: https://aglbrnd.co/r/f1782ab12211f765
Transcript
[00:00:48] Greg Kihlström: Hi, I’m Greg Kihlström, host of The Agile Brand, and here’s a question for you. If AI agents are making thousands of customer decisions a day on your behalf, how do you know they’re still making them the way your brand would? Today we’re recording live at Treasure AI’s Agentic World 2026 in Miami, and we’re gonna talk about the shift from managing customer data to acting on it with AI agents and what that means for the enterprise. To help me discuss this topic, I’d like to welcome Kaz Ohta, CEO and co-founder at Treasure AI back to the show. Kaz, welcome back.
[00:01:58] Kaz Ohta: Thanks for the perfect intro, Greg.
[00:02:00] Greg Kihlström: Yeah. [laughs] Oh, yeah, absolutely, yeah.
[00:02:02] Kaz Ohta: Glad to be back.
[00:02:03] Greg Kihlström: Yeah, and I think this is, uh, episode number three you’ve been on, so-
[00:02:07] Greg Kihlström: … yeah, yeah.
[00:02:08] Greg Kihlström: … re-returning champion.
[00:02:08] Kaz Ohta: I’m glad I’m, uh, just keep continuing doing the same thing, yeah, together with you.
[00:02:13] Greg Kihlström: Definitely looking forward to, to diving in.
[00:02:15] Greg Kihlström: Uh, for those that didn’t catch one of your previous episodes, uh, do you mind giving a little background on yourself and your role at Treasure AI?
[00:02:21] Kaz Ohta: Sure. Uh, my name is Kaz. I started this company 15 years ago as a CTO. I became CEO five years ago. It’s been a long journey. Uh, Treasure Data becoming data platform company, then pivoted into CDP company, now rebranded as Treasure AI in April, and going into Agentic journey right now. So there’s a lot of things we can discuss and talk today.
[00:02:45] Greg Kihlström: Yeah, absolutely. Well, let-let’s start there then. You know, la-last time we talked it was at CDP World. Now we’re at Agentic World here. Same, same conference, but Treasure Data became Treasure AI. Talk a little bit about, you know, what, what, what was the reason for that, that pivot, and, um, what does it, what does it kind of enable and unlock for you?
[00:03:04] Kaz Ohta: Yeah. So first of all, I wanna talk a little bit about what’s happening in terms of AI.
[00:03:10] Kaz Ohta: Right? So, um, I’d say, you know, it’s actually one of my background, which is I majored in computer science, and 20 years ago my professor built the world’s fastest supercomputer, right, that combines about half a million computers. You can run a lot of simulations and, you know, it just costs a lot of millions, right?
[00:03:36] Kaz Ohta: Now, recently I bought this little more like bento box. It’s called DGX Spark NVIDIA. It costs about $5,000. Very small. It’s faster than that supercomputer.
[00:03:48] Greg Kihlström: Wow. Wow. [laughs]
[00:03:49] Kaz Ohta: So after 20 years, the fastest supercomputer at the time can now run on your, uh, you know, bookshelf, on your … at home, right? That’s how fast your compute gets commoditized.
[00:04:04] Kaz Ohta: Now, what is very interesting is, uh, AI is also getting commoditized faster than any technology in the human history. You thought, okay, electricity got commoditized, compute got commoditized, but AI is getting commoditized faster than those two, right? So what it will cost $100 today to do with AI will be done in $1 next year. Right? So if you think, okay, it will be very expensive to do with AI, will be just done by 100th of the cost next year, right? And not only the, um, the cost is getting drastically lower, the intelligence is also going up and up, right?
[00:04:53] Kaz Ohta: So I kind of feel like, okay, at certain point of time, intelligence is becoming more like water, where everyone can just consume, and it’s available, it’s always on. Right? So that’s one thing that’s happening. The second wave I’m seeing these days is a lot of people’s use of AI is more like search. Right? So I’m like, okay, I wanted to buy some shaper. I always ask OpenAI or ChatGPT, “Hey, I wanted to buy something. I travel often, right? What are the products available?” So I’ve started asking those type of recommendations and questions to AI.
[00:05:36] Kaz Ohta: Right? So last few years a lot of brands we’re working with observe the decrease of the web traffic. Right? It’s because top of the funnel, TOFU, moved to AI a lot.
[00:05:53] Kaz Ohta: So a lot of people just ask questions for AI or they go to Google and then search, uh, search it. It also has what is called AI overview- A few years ago, the first place in search engine got about 20 to 30% click-through rate. Now it’s less than 10%, which means any search query, about 90% we are satisfied with the AI overview answer. Only 10% will click for the first link. If you’re just down there, it’s even lower.
[00:06:28] Kaz Ohta: Right? So that’s creating a lot of shift around how to e- be exposed to your buyers, right? The final piece that’s actually happening last few weeks is the consumer AI adoption is now moving from search alike to taking actions.
[00:06:49] Kaz Ohta: And with the release of something like Grok Bot or Facebook Meta Muse, that’s now became a Apple Store’s number one app.
[00:06:58] Greg Kihlström: Right. Yeah, and quickly. [laughs]
[00:07:01] Kaz Ohta: Yeah. It, it got about 5 million downloads in last, um, you know, three weeks or so.
[00:07:07] Kaz Ohta: Right? So imagine you now have 5 million AI assistants working for 5 million people, and it’s gonna be around 120 million or billions of people having AI assistant for each, right?
[00:07:22] Kaz Ohta: Now the use of AI shift more from search to taking actions. People are excited to have those AI assistants finding some subscriptions who are not using it, or booking the movie theater, um, based on your preference or times, right? Or drafting the email for you, right? So if that consumer shift happens, company or enterprise has to adapt because imagine we have 5 billion internet connected people having AI assistant for each. The only way to handle those AI assistants, billions of AI assistants, is to have an AI for you.
[00:08:04] Kaz Ohta: Right? So with those combined, we think, “You know what? Okay, it’s not just about data. It’s really the decisioning the AI on top of it that matters next.”
[00:08:15] Kaz Ohta: Right? And when it comes to CDP, we’ve been working with a lot of brands the last 15 years, collecting a lot of data, and it turned out we’re in the best place to run the AI on top because we have the cleanest data. We also have govern- governance on top of it, right? Turned out what AI needs is you actually need governed customer data, and that’s within CDP, right?
[00:08:46] Kaz Ohta: And the other things we’re also believing in it is right now majority of our user, I would say 90 to 95%, is human, right? So you go to CDP, create segment, a journey, and et cetera, et cetera. We believe in next few years 90 to 95% of users over our platform will be AI.
[00:09:07] Kaz Ohta: There will be no manual segment creation. There will be no journey creation manually, right? Literally, you ask AI to do so, and probably in, in next two years from now, 90% of access to our platform will be by AI. But, you know, a lot of our customer will orchestrate a lot of AI agent to finish the job.
[00:09:32] Greg Kihlström: Yeah. Yeah.
[00:09:32] Kaz Ohta: Right?
[00:09:32] Greg Kihlström: So it’s, it’s AI from the consumer. It’s AI from the marketing team or from the brand, let’s say, as well.
[00:09:38] Kaz Ohta: Right.
[00:09:38] Greg Kihlström: So, yeah, I mean, what… where does that leave the… You know, ’cause as you mentioned, you know, Treasure Data started as a, as a data platform, moved to a CDP. Is the CDP segment itself, uh, do, do you see that staying around as it is? Or is… Are, are you just part of this continuum that, that others are gonna be on as well?
[00:09:59] Kaz Ohta: Yeah. So I would say in reality of implementing AI agent, 90 to 95% of the work is actually having clean data and context. Right? Actually im- implementing AI itself is a interesting job. It creates a lot of buzz, but at the same time what you really need is governed customer data, right? So-
[00:10:24] Greg Kihlström: Yeah. Garbage in, garbage out. Yeah
[00:10:25] Kaz Ohta: … we believe CDP might be just a subcategory of broader category. I know, you know, CDP, CEP, cloud data warehouse just becoming blur. But it is true maybe from five years from now or 10 years from now, you have to have governed customer data somewhere. And I don’t know if we call CDP at that point, but conceptually that’s what powers all of the AI agent that, you know, supports the customer interaction with billions of AI assistants.
[00:11:01] Greg Kihlström: A new landing page turns into a pile of tickets and handoffs. Framer helps your team move faster. Teams can collaborate in real time, iterate on the same page, and publish instantly. Agents help you go beyond the vibe coded site and a first draft to bring you a production-ready site faster than ever. Framer is the AI website builder that helps creators, teams, and businesses ship production-ready sites faster than ever while getting every detail right. Prompt, inspect, edit, and publish in one place at a whole new pace. Agents and Framer work alongside teams to streamline collaboration on the same canvas, build custom code components, create and manage CMS content, optimize SEO settings, audit for issues, and ship everything all in one place.
[00:11:46] Greg Kihlström: Learn how you can get more out of your site from a Framer specialist or get started building for free today at framer.com/agile for 30% off a Framer Pro annual plan. That’s framer.com/agile. Rules and restrictions may apply.
[00:12:36] Greg Kihlström: Let’s talk a little bit about the activation part as well. So, um, you’re rolling out Personalization Studio-
[00:12:43] Greg Kihlström: … here at Agentic World which, you know, it, it seems to replace some of the capabilities from some of those standalone testing and, and-
[00:12:50] Greg Kihlström: … personalization tools, some DXPs. You know, what, what was the strategic driver? I mean, I think you already touched on it, but, you know, what was the strategic driver to build this capability and, and, you know, how can it harness the, the capabilities you already have in new ways?
[00:13:06] Kaz Ohta: Yeah. So first of all, um, we want to be as neutral as possible as a vendor right? So we keep maintaining all the connections with other partners and tech ecosystems, and that’s why, you know, a lot of our customer love us, right? But at the same time, if AI agent come in again and making decision almost real time, you almost have to have data and activation working together in real time.
[00:13:35] Kaz Ohta: And that has been a lot of challenge if you’re using two vendors from the user’s perspective, right? So it will be a decision as a user where, okay, Treasure Data has customer data, and Engage, that’s our email channel activations, and then personalization, you can close the loop almost instantaneously or real time, or to avoid any vendor lock-in, you can use multiple, but you have to pay some extra tax to make it work, right? I think that’s a choice, and there’s a lot of customer who’s asking, “You know what? I’m just consolidating a lot of software stack into one, and then that has to work perfectly.”
[00:14:21] Kaz Ohta: That’s one path you can go. The other path is, “You know what? I’m just using best of breed,” and I think we’re trying to provide a choice But again, if you were to choose our Engage, Personalization, and CDP together, not only you’re consolidating your software stack, we’re making sure, you know, every single request and customer interactions, AI will make a real time closed loop decision all the time.
[00:14:50] Greg Kihlström: Well, and that’s, I mean, and, you know, I, as, as a consultant, I, I work with, I talk with a lot of customers that face this challenge of activation in, you know, i- in some ways they’re getting closer and closer to the source, you know, from the data perspective, and yet they have all these disparate systems kind of disconnected and/or the route to get real time personalization, you know, when you have three different… Which a lot of these, these large organizations, they have, they have one of everything basically, right?
[00:15:22] Greg Kihlström: So, so closing that loop, get, you know, getting it as close to the source as possible makes a lot of sense-
[00:15:30] Greg Kihlström: … um, despite, you know, some… I, I know there’s certainly a, a, a cause to be made for composability and, a- and things like that, but, but definitely the, the speed required of being able to make decisions so quickly, it just seems like the case is being made again for kind of, uh, consolidating, right?
[00:15:50] Greg Kihlström: Imagine you have millions of customers or consumers using AI to access to your products and services, right?
[00:15:58] Kaz Ohta: You need a lot of speed and real time to act on it.
[00:16:02] Greg Kihlström: Yeah. Well, and even moving the data from one system to a… There’s, yeah, there’s just a lot of, yeah, a lot of-
[00:16:08] Kaz Ohta: Exactly
[00:16:08] Greg Kihlström: … lot of considerations there.
[00:16:10] Kaz Ohta: Right.
[00:16:10] Greg Kihlström: I also wanna talk about, um, the, the governance aspect. You, you, you’ve touched on that a little bit there. You know, your messaging mentions autonomous ROI-driven AI agents, um, but also mentions, you know, uh, keeping humans in the loop. Now that agentic, uh, like AI can mean a lot of things to a lot of people, but it also, you know, it, it certainly covers a lot of ground. How do marketers approach this in a, in a way that, you know, any- anything from the re- you know, highly regulated to just any large brand needs governance. How do they, how do they get the reassurance that autonomous also means, you know, brand compliance and you know, there, there is that human in the loop?
[00:16:53] Kaz Ohta: Yeah. I feel like AI safety is an engineering problem.
[00:16:57] Kaz Ohta: Right? There’s a lot of threat and rumors and, you know, about, like, AI taking over the world and all this, but, you know, I would say, in fact, you should always start from zero permission from AI.
[00:17:11] Greg Kihlström: Yeah.
[00:17:12] Greg Kihlström: Yeah.
[00:17:12] Kaz Ohta: And then you can gradually open up, adding s- a couple permissions to do it, you know, some of the capabilities to do it, right? So our approach is every single AI conversations or threat to start, we have a secure containerized environment, and then pretty much all of our customers can control what you can do within this containerized environment.
[00:17:38] Kaz Ohta: And then also, you can also trace what AI has done, every single, let’s say, file access or network access. We can also wrap around the credential management where AI will never, ever access to these credential by having a gateway. Anyway, long story short, we think having those containerized and secure environment for AI is a given and must-have for enterprise to adapt, and auditability and the governance, that’s what customers are asking for and looking for, and we’ve spent a lot of engineering time and resources to achieve that. And then to be honest, a lot of enterprise t- On AI use cases with the security and governance aspect of it.
[00:18:27] Kaz Ohta: And, uh, we’ve been working with o- one of the largest brands, some of the wor- world’s largest brand, and just learning and implementing what they want.
[00:18:37] Greg Kihlström: Yeah.
[00:18:37] Kaz Ohta: Right?
[00:18:37] Greg Kihlström: Yeah. Makes sense.
[00:18:38] Kaz Ohta: And that’s an advantage.
[00:18:39] Greg Kihlström: So I mean, built in versus after the fact, like, well, what about, [laughs] what about this or what about that, right?
[00:18:47] Kaz Ohta: But again, I think it’s important to say, okay, s- always start from zero permission and start adding some, right? And then you audit everything, right? And you can basically control.
[00:19:00] Kaz Ohta: Right? So our thesis, AI has to be under human control, and you have to have every switches and knobs for human, right?
[00:19:10] Kaz Ohta: Yeah, which we provide.
[00:19:12] Greg Kihlström: Love it. Yeah. So last thing I wanna talk about is just some of the, the people a- and process change required. You know, certainly there’s been a ton of change just, um, in, in the world, you know, in the world of work and marketing over the last few years. Um, as agents take on more of the execution, teams, they’re not necessarily shrinking, but their roles are changing-
[00:19:36] Greg Kihlström: … and, and what they do and what they should focus on. For those leaders out there listening, you know, how should they set up their organizations to really keep adapting? Because I don’t think we’re done with the change either. So, you know, how do you, how, how do you, um, you know, recommend that a, the CMO, uh, you know, marketing leader out there just, you know, sets their team up to, to keep pace with change?
[00:19:59] Kaz Ohta: I think one of the hard things we learned is if you just keep using existing processes in team, even if AI drastically shrinks some of the work-
[00:20:09] Greg Kihlström: Yeah
[00:20:09] Kaz Ohta: … um, human team is still the bottleneck.
[00:20:13] Kaz Ohta: Right? And, uh, one of the example I have is, um, you know, uh, the m- major chunk of AI use case right now is coding, right? So our engineering team is now producing three to four times more code compared to six to nine months ago, thanks to the AI. But, you know, first of all, we reviewed every single code, um, by human, and then review becomes much more like, okay, human bottleneck, right? And then we started adapting AI to actually doing reviews, security reviews, and, you know, we rethink how we build and ship, uh, software products, and trying to figure out what
[00:20:58] Kaz Ohta: are the right, uh, I would say, cooperation model between human and AI, right? So I think that’s gonna come into every single department or every single job. Your job basically is to orchestrate your AI agents, and then let it work together, but in fact, you’re responsible for every single outputs they’re generating.
[00:21:25] Greg Kihlström: I’m actually start observing some sort of laziness in terms of AI, which is, okay, you ask some work, and you just delegate to AI, and then you just respond with the output without any review. And that’s the new type of laziness-
[00:21:39] Greg Kihlström: Yeah, yeah
[00:21:39] Kaz Ohta: … I think that’s happening in the world. Right? But, you know, y- y- you have to prevent that in a way, okay, what are the right input and output between human and AI so that you as an organization can move forward, right?
[00:21:56] Greg Kihlström: Yeah.
[00:21:56] Kaz Ohta: I think that’s the type of thinking the new leader has to come in.
[00:22:01] Greg Kihlström: Yeah, I mean, I, I think of it as, you know, you wouldn’t just take… You have an intern, let’s say, you wouldn’t just take what they produce and, you know, give it to your boss to, you know, as finished product. But also, there are times where you actually learn something new from what your intern puts together. You know, so there’s-
[00:22:18] Greg Kihlström: … there’s a, I think there’s a, there’s a through line there of like, it doesn’t mean that you just pass it straight from AI to, to wherever it needs to go, but there’s also some, you know, some opportunity to, to grow even by you know, being part of the process, right?
[00:22:31] Kaz Ohta: Yeah. And then one of the belief I also have is, okay, a lot of people say, okay, AI could, you know, um, replace your job and all these, but one thing I can clearly say is people without AI will be replaced by people with AI. And I also have a interesting story with our intern, which is, okay, you know, we’ve been, uh, writing a lot of internal emails, uh, you know, distributing it. That’s done by manual. One of the intern who is, like 18 years old then came in, and, “You know what? I could just write an agent to do it.”
[00:23:12] Kaz Ohta: And he has built it in four hours, and it used to take four hours to write an email. Became like 15 minutes, and then we productize it. So a lot of people are now saying, okay, younger talent’s job will be gone, et cetera, et cetera. But wait, they also have a fresh mindset with AI, which, you know, we didn’t have this level of AI until… I’m, I’m 40 right now, but a few years ago, right? So they also have different mindset. So I would say I wanna give more opportunity for younger people who’s more AI native in bringing in, you know, different perspective, and that’s also a interesting part of leadership job right now, right?
[00:23:55] Greg Kihlström: I love that. Well, Kaz, it’s always great talking with you. Um, got two last questions before we wrap up here. First, uh, you know, we’re about to kick off Agentic World here in Miami. Uh, what are you looking forward to most?
[00:24:07] Kaz Ohta: You know, we have a lot of brands here, you know? We got five billion people internet connected, and we manage two to three bil- billion people’s data inside our system. So in this room, we’re delighting billions of people’s life every day. At the same time, we also have to be a custodian of that data as a community. Right? So I’m looking for, you know, talking with a lot of leaders who is thinking about how AI’s gonna transform, how consumer adoption of AI will change the marketing. At the same time, how to protect that, you know, uh, customer data with the governance, security, and compliance. You know, that’s w- what I really wanna learn from them, and then just evolve together.
[00:24:55] Greg Kihlström: Yeah. Love it. And last question for you, uh, what do you do to stay agile in your role, and how do you find a way to do it consistently?
[00:25:02] Kaz Ohta: Um, I would say, you know, being a founder is very interesting. You know, every six month or so, you know, your role is very different, right? And, um, you know, I’m always trying to learn. You know, for example, to learn AI, I bought a couple GPU at home, started deploying some of the agents for me. You know, I got, like, maybe 40 to 45 small agents running for every day, right? So I think curiosity is one thing that, you know, that’s something you really need to have. And, uh, I think what I learned in school, that’s called knowledge, is, like, probably gonna be fully commoditized, and the only way you keep just innovating is just, you’re just very curious about what questions to ask, and I’m, I’m trying to keep ahead of it.
