五个切实产生业务影响的实用AI工作流
人工智能正迅速融入日常运营，但许多人要么高估了其复杂性，要么低估了它带来的实际价值。结合与谷歌云（Google Cloud）领导团队的深入交流，本文提炼出五个精准、高效的AI工作流程和工具——它们无需深厚技术背景或大规模改造，却能带来清晰可衡量的商业回报。
1. Gemini 深度研究：重塑市场与竞品分析
过去，无论是市场调研、竞争对手对标，还是新产品探索，往往需要数小时甚至数天时间手动查阅资料。而Gemini的“深度研究”功能，借助实时谷歌搜索与自动化信息整合能力，彻底简化了这一过程。
只需几分钟，Gemini就能回答原本需数日才能完成的复杂问题。它从150多个相关网络来源中聚合洞察，以报告形式呈现结果，并将用户假设与真实市场数据进行对比。对于希望打破内部偏见、验证新方向，或快速理解行业变化的企业而言，这项功能能在极低资源投入下提供精确且可执行的建议。
核心价值：Gemini不仅回答问题，更能帮助你发现“该问什么问题”。高管可以上传专有文件、整合上下文数据，并获得综合性的市场视角或反方论点，在投入更多资源前做出更明智的决策。
2. NotebookLM：基于语境的学习与知识管理
传统的入职培训和内部学习材料常存在知识断层，难以满足个性化需求。NotebookLM是谷歌推出的AI驱动、上下文感知的学习工具，正在彻底改变企业对新员工入职和持续员工发展的支持方式。
用户可将公司文档、链接、Drive文件甚至YouTube视频导入NotebookLM。该工具随后支持互动问答、个性化解释，并生成多种格式的输出内容（如播客、视频、闪卡），适配不同学习风格。值得注意的是，NotebookLM并非仅限企业客户使用——它是免费且公开可用的。
核心价值：企业主可创建动态、基于真实资料的入职环境，让每位员工都能从公司原始材料中获得实时解答。这显著缩短了上手时间，提高了知识留存率，减少了重复咨询和再培训需求，直接提升生产力并控制成本。
3. Gemini CLI 与代码助手：赋能非技术人员实现流程自动化
定制化自动化不再是大型企业的专属。谷歌的Gemini命令行接口（CLI）和代码助手工具，已不再局限于资深开发者。如今，业务分析师、财务经理等非技术人员也能构建轻量级应用、自动处理繁琐的电子表格任务，或在无需IT支持的情况下快速搭建解决方案原型。
这些工具提供联网的实时代码建议、函数补全，甚至能解析遗留技术文档。对非技术人员而言，关键在于“意图”：你只需描述想要实现的目标，系统便会生成可行方案。虽然正式上线仍需专业审核，但业务团队现在可以在投入资源前，自行构建概念验证的工作流和自动化流程。
核心价值：更快的实验与开发意味着企业无需花费数月撰写计划书或与外部供应商谈判。公司可以快速验证想法，从备忘录直接跳转到演示原型，并在规模化前不断迭代，从而实现更敏捷的创新周期和更低的项目开销。
4. 自主智能体：通过 Google Jewels 实现后台自动执行
持久、自主的后台任务正在成为现实——Google Jewels 就是一种能够根据机器可读规范独立执行任务的AI智能体。企业不再需要耗时的手动迭代，而是将重复性或复杂的任务交给Jewels，由其自动执行并返回可操作的结果。
对技术团队而言，这意味着代码更新、文档编写或测试生成可以“全天候”自动完成。管理者可协调多个智能体并行推进任务。尽管清晰、结构化的指令至关重要，但AI智能体异步工作的能力所带来的效率提升极为显著。
核心价值：企业可在不增加人力复杂性的情况下，成倍提升运营吞吐量。团队保留监督与协调权，确保质量和安全，同时将重复性工作交由系统自动处理，释放人才去专注更具战略性的优先事项。
5. AI作为智能界面：无缝集成至各类业务工具
AI功能正快速渗透至谷歌现有产品中，包括Google Sheets、Docs乃至Search。这一转变并非旨在取代现有工作流，而是让诸如创建数据透视表、查找信息或生成报告等常规操作变得更智能、更直观。
如今，管理者和高管不再需要精通每一种软件或编程语言。AI的集成意味着“意图”和“上下文”驱动结果：你表达想达成的目标，系统便自动执行或提供建议。随着客户服务、入职培训和内部分析变得更具主动性与智能化，人与技术之间的交互将趋于无摩擦。
核心价值：自动化门槛降低、培训成本减少、采纳速度加快是立竿见影的优势。长远来看，这种无缝、具备上下文感知能力的AI生态系统将成为企业的竞争差异化优势，提升响应速度、扩展支持能力，并增强整体敏捷性。
实践启示：通过动手实验塑造未来
AI的采用不是放弃控制，而是善用工具来提升速度、增强上下文理解，并获取可执行的洞察。最具影响力的结果源于积极的实践：尝试免费层级的服务，将AI融入现有工作流，并始终保持对自身流程的好奇与谦逊。
商业价值不仅体现在成本节约上，更在于打造更智能、更快速、更具响应力的运营模式。随着AI不断发展，那些主动测试、学习与迭代的组织，不仅能保持竞争力，更能主动塑造自身业务成长的环境。
—英文原文—
原标题: Ep 683: 5 Practical AI Workflows That Actually Matter
原内容:
Episode Categories:
Resources:
Join the discussion: Got something to say? Let us know on LinkedIn and network with other AI leaders Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineup
Connect with Jordan Wilson : LinkedIn Profile
Five Practical AI Workflows Driving Real Business Impact – A Deep Dive on Google’s Latest Strategies
Artificial intelligence is rapidly integrating into daily operations, but many still overcomplicate its adoption or underestimate the tangible benefits. Pulling from a candid discussion with Google Cloud’s leadership, this article outlines five pinpointed AI workflows and tools that demonstrate clear business value—without requiring deep technical expertise or sweeping transformations.
Deep Research with Google Gemini: Transforming Market and Competitor Analysis
Many research projects—whether market analysis, competitor benchmarking, or new product exploration—previously required hours, sometimes days, of manual effort combing through sources. The deep research capability in Google Gemini, harnessing live Google Search and automated synthesis, streamlines this process.
Using Gemini’s deep research, complex questions can be answered in minutes, not days. The tool aggregates insights from over 150 topically relevant web sources, presents findings in report format, and juxtaposes a user’s assumptions with real market data. Businesses looking to challenge internal bias, validate new directions, or rapidly understand industry shifts stand to gain precise, actionable recommendations at a fraction of current resource expenditure.
Key value: Gemini doesn’t just answer questions—it helps surface what the right questions should be. Executives can upload proprietary files, integrate contextual data, and receive synthesized market perspectives or counter-arguments, supporting better-informed decisions before allocating further resources.
Contextual Learning with NotebookLM: Onboarding and Knowledge Management
Traditional onboarding and internal training materials often leave knowledge gaps and fail to address individual needs. NotebookLM—Google’s AI-powered, context-grounded learning tool—offers a step change in the way companies enable both new hire onboarding and ongoing employee development.
Users can feed NotebookLM curated company documents, links, Drive files, or even YouTube videos. The tool then enables interactive Q&A, tailored explanations, and multi-format outputs (podcast, video, flashcards) that match individual learning styles. NotebookLM isn’t limited to enterprise customers—it’s accessible and free.
Key value: Business owners can create dynamic, contextually grounded onboarding environments where every employee gets real-time answers from company source material. This drives faster ramp-up times, higher knowledge retention, and reduced recurring support tickets or re-training—directly impacting productivity and cost containment.
Gemini CLI & Code Assist: Empowering Nontechnical Builders to Automate Processes
Large enterprises aren’t the only ones who benefit from custom automation. Google’s Gemini CLI (Command Line Interface) and Code Assist tools are no longer reserved for seasoned developers. Nontechnical users—including business analysts and financial managers—can now build lightweight apps, automate time-consuming spreadsheet work, or rapidly prototype solutions without waiting for IT support.
These tools offer internet-connected, real-time code suggestions, function completions, and can even untangle legacy technical documentation. For nontechnical users, the key is intent: Users describe what they want, and the tool generates functional solutions. Production deployment still requires oversight, but business teams can now build proof-of-concept workflows and process automations internally before resource investment.
Key value: Faster experimentation and solution-building mean that businesses don’t spend months in planning documents or external vendor negotiations. Companies can validate ideas quickly, jump from memo to demo, and iterate before scaling—resulting in sharper innovation cycles and less project overhead.
Autonomous Agents: Background Execution with Google Jewels
Persistent, autonomous background work is emerging with Google Jewels—an AI agent that operates on machine-readable specifications and performs tasks independently. Instead of time-consuming manual iteration, businesses offload repetitive or complex tasks to Jewels, which executes changes and returns actionable results.
For technical teams, this means code updates, documentation, or test creation can happen autonomously “around the sun,” with managers coordinating multiple agents for parallel progress. Clear, well-structured specifications are critical, but the lift from having AI agents work asynchronously is significant.
Key value: Companies can multiply operational throughput beyond traditional headcount without adding complexity. Teams retain oversight and orchestration, ensuring quality and security, while repetitive work is handled automatically—freeing talent to focus on strategic priorities.
AI as a Smart Interface: Seamless Integration Across Business Tools
AI features are rapidly rolling out across existing Google products, including Google Sheets, Docs, and even Search. This shift isn’t about replacing workflows—it’s about making routine actions like building pivot tables, searching for data, or generating reports smarter and more intuitive.
Executives and managers no longer need proficiency in every software or programming language. AI’s integration means intent and context drive outcomes: users express what they want to achieve, and the system executes or advises. As customer support, onboarding, and internal analysis become agentic and smarter, the interface between staff and technology becomes frictionless.
Key value: Reduced barriers to entry for automation, lower training costs, and faster adoption are the immediate wins. Over time, this seamless, context-aware AI ecosystem becomes a competitive differentiator—improving response times, scaling support, and boosting overall agility.
Practical Takeaway: Shaping the Future Through Hands-On Experimentation
AI adoption isn’t about surrendering control; it’s about leveraging tools that drive speed, context, and actionable intelligence. The most impactful results stem from active experimentation: trying new free-tier services, integrating AI into existing workflows, and remaining endlessly curious and humble about one’s own processes.
Business value is realized not just in cost savings, but in shaping smarter, faster, and more responsive operations. As AI matures, those leaning in—testing, learning, and iterating—remain not only competitive, but can actively shape the environment in which their business grows.
For more actionable strategies and updates on effective AI integration, consider subscribing to specialized daily newsletters or partnering with experts focused on context-driven, real-world business transformations.
Topics Covered in This Episode:
Five Practical AI Workflows with Google
Gemini Deep Research for Rapid Analysis
NotebookLM AI-Powered Knowledge Exploration
Gemini CLI and Code Assist for Developers
Google Jewels Autonomous Coding Agents
AI Change Management and Workflow Automation
Gemini’s Contextual Integration with Email and Calendar
Gemini and Agentic AI Across Google Products
Episode Transcript
Midroll [00:00:00]: This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Jordan Wilson [00:00:16]: I think sometimes when we think about implementing AI into our day to day workflows, we overcomplicate things. Right? We think sometimes we have to be very technical or it’s a big endeavor to get started, and that’s definitely not the case. And today on everyday AI, I’m excited for today’s show because we’re gonna be going over five simple AI strategies to supercharge your workflow with Google. So I’m excited. I hope you are too. Let’s get into it. Welcome to Everyday AI. If you’re new here, Everyday AI, it’s an unedited, unscripted livestream podcast helping everyday business leaders like you and me not just keep up with everything that’s happening in the world of AI because it is hard to do, but this show helps us make sense of it and grab the information that we actually need to grow our companies and our careers. Jordan Wilson [00:01:08]: If that’s what you’re trying to do, awesome. Starts here. But if you wanna take it to the next level, make sure you go to our website, youreverydayai.com. Sign up for the free daily newsletter . We’re gonna be be recapping the five simple strategies we’re gonna be going over today as well as keeping you up to date with everything else happening in the world of AI, so make sure you do that. But without further chitchat from me, let’s bring on, an expert from Google to help walk us through this. So I’m excited. And livestream audience, if you could, please help me welcome, to the show. Jordan Wilson [00:01:40]: We have Richard, Saroder, who is the, senior director and chief evangelist at Google Cloud. Richard, thank you so much for joining the Everyday AI Show. Yeah. Really happy to be here, Jordan. Alright. So tell us, like, what the heck do you do at Google Cloud? Because it sounds like you do a lot, but walk us through your day to day. Richard Seroter [00:01:59]: Yeah. Definitely can’t even explain it to my parents. Right now it’s a problem. But, look. I lead, teams like developer relations, our technical docs team, our open source program office. Just anybody who’s about how do we inspire and activate builders on Google Cloud. How do you give people the confidence? Right? There’s a lot of information out there, but how do you give them the confidence they can do it too? So they can use cool open source stuff, cloud services, AI stuff. I spend most of my day talking to customers, working with my team, Jordan Wilson [00:02:25]: going hands on. I still code a decent amount. I’m not good at it, but enough to use the products that I don’t think we should be talking about, products we don’t know how to use. And, you know, I’m curious. So throughout your years at Google, right, and, obviously, on the AI side, Google’s been there for a very long time before the large language model, surged from a couple of years ago. But I’m curious. The people that you’re talking to specifically about AI, is it more, yes, the technical dev, type people? Are you talking to the everyday business leader? And is it changing as AI, especially generative AI becomes more and more accessible? Richard Seroter [00:03:02]: Yeah. I mean, I’ve been in this field now too long. I don’t guess my age. But this is the I mean, I think this we’re at Internet level in terms of people who care about this outside of IT. No one outside of IT cared about Kubernetes, serverless, arguably cloud computing, maybe mobile, but we’re back to, like, Internet level conversation of, like, the people I talk to are sometimes in marketing , sometimes in HR, sometimes in CIO roles. It’s not just builders. Of course, for tech folks, it’s awesome. But the difference is this isn’t just about how do I improve my day to day work with tech stuff. Jordan Wilson [00:03:32]: Some of it’s like, how do I change my business mix and products they offer? How do I change how my team works? I don’t think this is just about what you can do. I think it’s about how you do it. And we haven’t had a change like that industry wide in a long time. Mhmm. Yeah. That’s a that’s a good point. And, you know, I think people, when making comparisons about generative AI and and what it can accomplish, they’re not always going back to, you know, cloud or mobile. They’re saying, like, electricity. Jordan Wilson [00:03:58]: Right? Like, so much bigger than that. You you know, I’m curious even for you before we get into our five strategies. Like, how how big or how much of an impact, has Gemini and just Google AI in general had on how you work personally? Richard Seroter [00:04:14]: Yeah. I mean, look, on one hand, I don’t wanna be one of these wacky AI influencer types who says everything’s unbelievable. Everything changes with AI. Like, look. It’s still, hopefully, good people using good tools. You still need human thought. You still need human creativity. Some of these things don’t work as they advertise. Richard Seroter [00:04:31]: Some things are better. Some things are worse. It’s all great. These are tools. These are ways we do better work. Now they’re transformative tools for some teams. And so for myself, and we’ll talk through some of these strategy things, how I research, how I learn, how I build, how I do some of my day to day things. Absolutely. Richard Seroter [00:04:46]: And, look, there’s other areas where I am purposely staying low tech. I write a daily newsletter and I write every word. I don’t want AI to do it for me. And, like, I I learned by writing. I learned by doing that work. And so I think all of us wanna make sure we hold closely to those things we actually love doing and make sure that we’re building depth, not just sort of shallow knowledge because we’ve outsourced all our thinking to the AI. So use this as a tool to augment yourself, not replace yourself. Jordan Wilson [00:05:11]: No. Like what Richard just said there, like, take take that away. That’s so important. Don’t just, you know, kick everything over to AI. You still have to practice those skills, the human side, if you really wanna augment to the level that you can. Right? So real quick, I’m gonna give everyone the five different strategies, and then we’ll unpack them, have Richard unpack them one by one. So number one, Gemini deep research for analysis. Number two, notebook LM for exploration. Jordan Wilson [00:05:38]: Number three, Gemini CLI and code assist to build. Number four, jewels for background work. And five, AI just kind of rolling out everywhere. So let’s get into them. Let’s start at the top. A tool that I love and use all the time, Gemini Deep Research. Richard, can you explain it for us, and how can people what’s a good strategy to put this into to your daily workflow? Richard Seroter [00:06:01]: Yeah. Yeah. I mean, with all of these and the the ones you called out there, these aren’t just about using new tools. To me, these are about forming new habits. Mhmm. And that’s the hard part. That’s the change management piece. Right? Like, you could one off use any of these and then never use it again. Richard Seroter [00:06:14]: So what we’re all trying to do is almost reprogram ourselves and be like, when I get a hard question, what do I do first? To me, that’s what AI first means. AI first does not mean I use AI for everything. It means that when a situation comes up, I ask myself, there any help AI can do here? No? Fine. Do your thing. Yes? Do it. So Gemini deep research is part of the Gemini app, and we’ve added a ton of stuff to that over this year, whether that’s helping you vibe code an app or, you know, build a storybook for your kids, which is crazy. All sorts of cool things. But deep research is awesome. Richard Seroter [00:06:43]: And there’s other deep research y things out there, but I’ll focus on this one. The idea and I just used it, last week. I had a complex problem. I was trying to actually bias it and say, like, look. We’re at Google. We’re only focused on this part of the application delivery right now. Is the rest of it kind of boring? Should I ignore it? And so what deep research did, it was it went to because it’s connected to Google search, which is awesome. It went and looked at I think I count it was over 150 sites, synthesized it all, gave me a report that repudiated me constantly saying I was missing the boat, which was amazing, fables, charts, all this stuff in about six minutes. Richard Seroter [00:07:19]: So this was work legitimately that would have taken me two to three days. I would have gone to a Google search. I would have typed in words. I would have clicked blue links. Keep doing that. We need that ad money. Like, it is gotta keep the lights on. But if I’m doing a research project, why in the world would I do that today? Instead, I’m going to deep research. Richard Seroter [00:07:35]: I’m getting a really good synthesis, and now you can upload your own files. You can redirect that research. The results of that research, I can turn into different forms and export to a doc. And so I think we change how we do research. No one should say I need weeks and weeks from normal projects that we might be doing personally, planning a vacation, understanding competitors, doing analysis of a market, figuring out a business scheme, figuring out a technical architecture. Instead of just kinda going in and doing all that yourself, how about you kick start it? To me, AI is the best thing for blank pages. Like, it’s the easiest way to now go from I don’t know where to start to here’s something I can start with. And throw it away, keep 5% of it, but none of us just wanna stare at a blank page going, what do I do next? So deep research is an amazing way to go. Richard Seroter [00:08:22]: I have a question. Might be involve a lot of different angles. Can you give me a a look at that and get it back in minutes and go, that’s not it at all? Or shoot. I’m even asking the wrong question. I don’t wanna find that out three days later, a month later after my giant research project. I wanna know now. And so the ability to learn faster might be the only remaining professional competitive advantage out there. And so how do we all just learn faster? That that’s how you stand out. Jordan Wilson [00:08:49]: My so so much to unpack there, Richard. We’re like we might have to just cancel the other four because I wanna talk to you just about that for for multiple hours, but I won’t. But, you know, a couple things that I heard there that I I really wanna zoom in on. Is this really talking about, like, change management? Right? So even for me personally, this is how I use deep like, deep research. If I wake up and I’m like, oh, I’m gonna go grab a coffee and sit down, I sit down first, have Google Deep Research start on something, get my coffee, and come back. So this is like making those little changes, but something else you said, intentionally have it, like, not challenge your thinking, but sometimes go against a preconceived notion. Right? But doing it with Google deep research. What’s maybe a a strategy that you can leave people with on how they can use deep research for maybe either challenging their thinking or when you are just, you know, having to take on a big project instead of staring at a blank page? What’s what’s maybe a a piece of advice you have to people? Richard Seroter [00:09:48]: Yeah. I would say stop thinking of AI as a great way to get answers. Think of it as a way to get great questions. And we don’t use it that way. But if you go to deep research and even or, frankly, Google AI mode, go to google.com/ai. Go to any of our AI tools and say, I have this issue. What questions should I be asking? What should I be thinking about? I’ve done this with, I use some tools sometimes if I get a really technical doc from a team, and I know they’re just trying to show me up. I can pass that doc and go, what are really smart questions to ask about this doc in my review? Now do I take them all? I don’t know. Richard Seroter [00:10:20]: But you might have sparked something with me. Go, oh, that’s a good angle. I should think about that. So generative AI is pretty good at that. And so sometimes you just don’t know what to ask. Mhmm. And so it’s great to sometimes give adversarial questions to a deep research thing going, this is what I think, but tell me why this is wrong, or tell me what I’m missing, or look for counter views. Because, honestly, I think we do our best strategic work when we take these three sixty views of an issue and don’t just get myopic about the preconceived solution we thought. Richard Seroter [00:10:48]: And, Jordan Wilson [00:10:48]: you know, I don’t think we’ll have time to go feature by feature and update by update. But one thing that I think is important for our audience to know, a a new, update in Gemini Deep Research that I am loving is now the ability for it to, you know, go through your calendar, to go through, you know, if you choose to connect it, right, if you enable that, to go through your email. Because that’s one thing I struggle with so much, and then to combine it with normal, kind of deep research across the web. You know, help us understand what that can unlock for people because even for me personally, that unlocks so much. Richard Seroter [00:11:24]: I mean, that’s the contextualization of loss. LLMs are amazing, and we’ll all keep shipping amazing things, and that’s awesome. But this world of more agent stuff and agent stuff really just how do you give kinda overlay these models with things like tools which access other real time data or your personal data? How do you have long running conversations, not just stateful one offs with an LLM? And so things like deep research are agents and be able to pass in data that might be your inbox or your whole set of style guides that you wanna feed in and ask it for ideas on a look and feel. The model doesn’t know that by itself. And so being able to give this thing context, that’s why you hear this term context engineering. I just don’t wanna write a clever prompt. That’s cool, and that’s a skill, but it’s insufficient when I wanna give it a bunch of things like, hey. You know, here’s a bunch of SOP documents. Richard Seroter [00:12:12]: Help me figure out new standards for my team. I can’t figure it out by itself. But once you give it that context, you get something pretty awesome back. So thinking about your context, what do I need to tell this thing so it can properly give me what I need and not just assume it’s a magic robot who knows all this stuff? Give it a little help. This is again where you stay in control. I think that’s what we’ve learned even the last twelve months. There was a lot of fear of this thing’s just gonna do all our work. These things don’t know as much as we do, regardless of what some influencer types say. Richard Seroter [00:12:40]: Like, they need context. They need certain things. You have that. You’re the orchestrator. You’re the engineer. Every individual is now becoming a manager because you are managing the work of these things. You gotta change your mindset to think about that. Jordan Wilson [00:12:53]: Yeah. It’s it’s a great call out. And even just the concept of providing more and more context is going to make kind of the agentic work from Gemini deep research, much more fruitful in the long run the more context you share. Right? Speaking of sharing context, I mean, notebook l m for exploration. Like, I I can’t and anyone that’s listened to the show, I’ve talked for literally countless hours about how much now I just rely on notebook LM and how it’s completely changed, not just how I work, but how I think and how I organize myself. Right? Richard, maybe, for those who haven’t heard me talk for hours about NotebookLM, explain a little bit what it is, what it does, and then let’s maybe, dive in a little bit deeper after that. Richard Seroter [00:13:41]: Yeah. I didn’t understand what it was when we first announced the Google IO whenever it was. Like, that’s neat. What the heck is the use case for that thing? And then I finally kinda I had some light bulb moments. But how do you have a bunch of data that you collect on your terms? Could be your own data, could be links, could be YouTube videos, could be now it connects to Drive and pull in your own stuff. And then how do you then turn this into a form that you can learn however you want? That’s the big takeaway is at this point, 2025, you can learn what you want how you want to. And I don’t think I don’t think you and I are gen z. Let’s pretend we’re not. Richard Seroter [00:14:13]: You and I learned from teachers teaching us in class one way to 30 kids. We read books, maybe had some CDs, watched maybe an online training. That wasn’t personalized. That was whatever the heck would be delivered to the masses. And if you fell behind, you fell behind. If you had a dumb question, you’d either ask or keep it to yourself. This is the first time where you and I can learn how we want to. I can use notebook l m to turn piles of information into a fifteen minute podcast. Richard Seroter [00:14:38]: Listen to it on the way to work while I’m walking the dog. I can turn that into a video podcast. Maybe I’m a visual learner. Flashcards because I’m about to get tested on it. Sounds good. Have a chat with the data going, I don’t understand this or make sense of this. There’s no company that should have the same onboarding process they have today in two years because they’re all terrible. Like, all the onboarding is just here’s a pile of information, study it, and get to work. Richard Seroter [00:15:01]: Why are we doing that? You should be giving every new hire a link to your notebook l m instance going, here’s our business. What do you wanna know about it? Oh, you wanna understand vacation? We’ll give you all the details. Just chat with it. You wanna figure out, you know, how the org is set up? It’ll figure out the org chart for you, tell you how. Like, all of a sudden, learn it on your terms. And so it’s free. We’ve made it available to students . It’s amazing student tool, personal tool. Richard Seroter [00:15:23]: But, again, this is we talked first about changing how you research. This is changing how you learn. And it’s it’s, again, it’s a new habit. It’s a new muscle. But all of Jordan Wilson [00:15:32]: a sudden, we’re all learning the hard way until we use things like this. Once maybe, you know, even going back to your initial reaction when you heard about it at IO and you’re like, okay. What’s the use case? Right? Now fast forward that it’s been out for, you know, a year and a half, two years. You know, maybe what’s an important takeaway, that you’ve maybe experienced or your team has experienced using NotebookLM that you think, would be helpful for our audience. Are you still running in circles trying to figure out how to actually grow your business with AI? Maybe your company has been tinkering with large language models for a year or more, but can’t really get traction to find ROI on Gen AI. Hey. This is Jordan Wilson, host of this very podcast. Companies like Adobe, Microsoft, and NVIDIA have partnered with us because they trust our expertise in educating the masses around generative AI to get ahead. Jordan Wilson [00:16:30]: And some of the most innovative companies in the country hire us to help with their AI strategy and to train hundreds of their employees on how to use GenAI. So whether you’re looking for chat g p t training for thousands or just need help building your front end AI strategy, you can partner with us too, just like some of the biggest companies in the world do. Go to your everydayai.com/partner to get in contact with our team, or you can just click on the partner section of our website. We’ll help you stop running in those AI circles and help get your team ahead and build a straight path to ROI on GenAI. Richard Seroter [00:17:08]: Yeah. Look. I saw I think it’s a couple of the events Google ran over the summer where the results of it was a notebook l m because no one pays attention to the 200 announcements we just made or the all the videos or the 15 blog posts or whatever. And so even as you’re doing big complex things, guess what? No one’s paying attention. It doesn’t matter if it’s Google’s event, Amazon’s event, your crazy awesome launch. No one cares about it as much as you do. Awesome. Give them a way to synthesize it then. Richard Seroter [00:17:33]: So first off, for every big complex thing even you or your business does, give an easy way for people to digest it all on their terms and learn about it. And then the other way is I think all of these tools are amazing at taking really complex things and helping us finally understand what they mean. And whether that’s AI in Chrome or whether you use notebook LM or Gemini app, whatever it is, give it the terms and conditions to your credit card, which no one in every the history of time has ever read, or your employment agreement where you’re like, this maybe this is fine, but this is 15 pages. Give that to the AI and go, what’s a weird thing in here that I should be aware of? Awesome. Where else can you do that? So look for those applications to either make the complex very simple or to take a very big distributed set of announcements and turn that into something that anybody can figure Yeah. Jordan Wilson [00:18:19]: And it probably would have been helpful for me to set the stage a little bit and just, you know, for people that aren’t familiar with NotebookLM, the concept of of grounding. Right? Because I think sometimes people are like, well, okay. NotebookLM is powered by Gemini. Why wouldn’t I just use Gemini for these things? So, Richard, could you just kind of explain kind of the concept of how notebook LLM just grounds answers in the information that you give it? Richard Seroter [00:18:39]: Yeah. It’s like we said. Look. A lot of these things are all gonna be based on the same, let’s even say, the same Gemini model or whatever your favorite model is. It’s about the experiences on top. That’s where the magic’s happening now. And so can I solve similar problems with different tools in different ways? Maybe. But notebook l m is purpose built to say, let me take a bunch of your information, your preferences, your links, ground it on your truth data, and turn that into a form that you can consume. Richard Seroter [00:19:02]: And that’s just what it’s good at. Do I would I use that to look up the latest baseball scores? I don’t think so. And maybe you can even do it, but that would be a weird use of it. That would jump to the Gemini app or whatever. So just knowing what these things are good for and even Google Cloud customers have it all baked into Gemini enterprise. It can be private just for you. Google doesn’t train on it, all that sort of stuff. So you have enterprise versions of notebook l m, the Gemini deep research experience, all in Gemini enterprise for corporate customers. Richard Seroter [00:19:30]: So a lot of this is just about what’s the interface you need to solve a given problem. And NotebookLM is amazing if you just say, I wanna learn from a lot of material that I’ve curated, and that could be my curriculum for this class this semester. That could be about my business, but it saves it. And then it’s something where I can just keep growing it or removing it or pairing it, learning about it different ways. Such a unique experience. It’s free to use. Everybody can use it on their phone, web app, party on. Jordan Wilson [00:19:58]: Yeah. And it’s, the the fact, number one, that it’s free. But number two, the fact that this technology exists and is this easy is still bonkers to me. Right? Like, thinking back, like, two years ago and, like, before notebook l m and then seeing what we can do with it now, I’m just sometimes like, how is this so easy, and how is it available for everyone? Right? Absolutely. No. Richard Seroter [00:20:19]: I mean, you’re gonna look back even in a year ago. We’ve just all been doing it the hard way, and that’s okay. Sometimes you had to learn it the hard way back to, like, skills and things. And so it’s good to know the hard way to do research. Like, we shouldn’t just have the easy way. Maybe people don’t go to libraries anymore, and no kid under 30 knows what the Dewey Decimal System is. But, like, that was a big part of how you and I probably had to do real research. And it’s good Jordan Wilson [00:20:40]: to know that because you gotta hunt and you gotta figure stuff out. But let’s do it easier now. Alright. So we just, unveiled two of our first five. Maybe for more nontechnical people, our next two maybe if you wanna get a little technical, we’re gonna get there. But before we do, just a real quick break for Richard Seroter [00:20:57]: a word from our sponsors. Jordan Wilson [00:21:02]: Alright. So let’s get into it, Richard. Number three on our list, talking about Gemini, CLI, and Codasys. So explain what is what is it, how does it work, and who should be paying attention? Because I’m even experimenting with this a little bit even though I’m not a developer or code. Richard Seroter [00:21:21]: We’re all developers now, Jordan. I mean, I think that’s the most exciting thing is we’re all builders. You don’t have to wait for an engineer to build the thing for you to at least see it in action. Now would you put build that in production? No. But you everybody’s a builder now, and that’s I can use the Gemini app to vibe code a web app and see what that looks like and all that stuff. But for these things, you know, we talked about learning differently. We talked about researching differently. It’s about building different and say, how do I use tools that help me as a developer, write software, learn my system? So the CLI or the command line interface is something that sits in the terminal. Richard Seroter [00:21:54]: You know, think a a lost prompt or whatever. Being able to come in there and have access to the Gemini model, being able to reference a bunch of extensions so I can reach into third party systems. And maybe I would use that to do something like, hey. I’m trying to, take this really old app and make it new. Okay. And then I think pass that into Gemini, update it, iterate back and forth. It’s really powerful or could be as simple as because it’s Internet connected. Hey. Richard Seroter [00:22:17]: I’m trying to build a chart of, today’s stock rankings, put them into a table though, and factor this in, and it’ll can reach out to the Internet, synthesize it all, put it back into language. So Internet connected, I can use it to build. I can use it to connect to third party systems. But for some people, it’s a really good hardcore programming way and and system administrative way to manage things without a point and click GUI. Right? Plenty of people are just faster with the keyboard, easier to build, full power of Gemini, giant free tier that anybody can mess with. Just give us an email address, and that’s it. And party on. And use corporate versions if you want to too. Richard Seroter [00:22:53]: But then sometimes if you’re a coder, use things like an integrated development environment or IDE. I wanna see my code. I wanna write it. I wanna do things. And things like Gemini code assist can help you complete a line of code or say, hey. I just need a function that reverses, you know, or adds two numbers together and it’ll write the function for you. That sounds good. Or I got this giant code from somebody. Richard Seroter [00:23:13]: They just retired. I don’t understand any of this. What does this application even do? And get back an answer in a second versus four days. And so how do I understand code, write code, change code? Again, you’re seeing at this point, I think 90% of devs are using some of these tools at this point, but so can everybody to some extent. I could use the Gemini CLIs and a business analyst or a financial analyst to maybe, hey. Can you look at this spreadsheet and make some, you know, updates to this? Am I jumping to Gemini code assist and go, you know, I have this kinda cool idea for an app. I’m not a programmer, but here’s what I want. And that’s for builders nowadays, I think the biggest takeaway is we’ve moved away from you having to know everything to do anything to you have to know your intent. Richard Seroter [00:23:54]: We all know our intent. What am I trying to do? Now, again, don’t take those things that you don’t understand and then push it all the way to production. You’ll get hacked or something will be screwy. But to prove your ideas faster, there’s a motto in my product area right now. I’m in a product area in engineering that focuses on all these dev tools. And one of our leaders, Ryan and Scott, they both kinda coined demos over memos. Mhmm. Build stuff. Richard Seroter [00:24:17]: Stop writing so many freaking docs. Build your idea out. Prove if it makes sense. And then when it does, write the document. But stop wasting months pixel pushing a doc and tables and perfect prose when your idea is not right. Build it. Build demos. Prove your ideas. Richard Seroter [00:24:32]: Everybody can do that. And then once you have a solid idea, you jump into real building and real scaling. But get that first experiment and learn stage done faster before the next one. That’s a culture change. Like, that transforms a business from being paralyzed by these giant release stages to, like, let’s all be builders. Let’s all prove ideas and and move the blockers. Mhmm. Jordan Wilson [00:24:54]: Richard, I think what you said there is really important. Just responding with, like, Jordan, no. We’re all builders. Right? It reminds me when a a story when I, had Paige Bailey from Google on the show, and I do suggest people go listen to that episode six nineteen. She talked about now at hackathons, it’s nontechnical people that are winning AI hackathons. Right? And and I love what you said there, demos over memos and just encouraging people to build it. Do you think it’s gonna become whether it’s using, you know, Gemini CLI and and Codesys or, you know, other kind of vibe coding tools? Is it gonna become common, or is it already common? And maybe I don’t know because I don’t, live in California where everyone’s just building their own, you know, solutions. Right? Like, oh, this this piece of software stinks or was spending hours, you know, a week just for this one figure. Jordan Wilson [00:25:44]: I should just build something. Is that gonna be couples the Richard Seroter [00:25:47]: de facto way to work? We’ll probably swing the pendulum that far, and we’re all just gonna be building everything. And you might build personal software. You might be like, I’m just trying to track my recipes better because they keep forgetting, what I make, and let me just build an app for myself. We’re just recipes better because I keep forgetting, what I make, and let me just build an app for myself. We’re gonna all do I think we’re gonna see an explosion of software. Now this would be an explosion of software that replaces production grade software. I’m not sure. Like, there’s gonna be times where could you write your own customer relationship management system? My goodness. Richard Seroter [00:26:13]: You could. It’s not feeling like that’s gonna be a competitive differentiator for you or something you wanna maintain. So I think we’re gonna swing the pendulum too far for a while where we all just build everything because we can. It’s super easy. And then we’ll find, as usual, that middle ground of still buy, you know, buy commodity and build differentiators. Like, be careful. Don’t accidentally build everything and then realize you took your eye off the ball of your business because you were geeking out on something that actually doesn’t matter to your success. Alright. Jordan Wilson [00:26:40]: So let’s let’s move from the terminal Yep. To the autonomous AI coding agent, Google Jewels. So explain, you know, explain the real, big use case here, Richard, and specifically the concept of, you know, having a background teammate? Richard Seroter [00:26:58]: Yeah. I mean, this is the world. Look. If you’re a builder now, there are things that you do I mean, think of it as a when I talk about the CLI, that’s really almost like working with a junior engineer. Oh, you’re you’re collaborating. You’re hanging out at the same time. You’re both working the same shift. Amazing. Richard Seroter [00:27:12]: When you work with things like JUULS, you’re actually working around the sun or you’re working with an outsource agent or a partner and saying, let me write a quality spec. You’ll hear the term spec driven development. Let’s write a specification that’s machine readable. Still natural language, but maybe organized really effectively. Let me iterate on a spec with this background agent and then hand it off. And I might go to lunch. I might go home. Might have your cup of coffee. Richard Seroter [00:27:34]: And when it’s done, it it gives me a pull request or gives me the changes going, here’s what I did. Check it out. You can go, do you like that? No. Not doing that. Let’s iterate that again. But it’s almost like having work that you can truly offload. Again, you’re still in control. You’re still the engineer. Richard Seroter [00:27:49]: You’re still the orchestrator. You’re still the coordinator. But the idea of having a bunch of background work, this is the next cultural change. I think the future for a lot of builder teams is that you’re gonna have multiple of these going at the same time. If you’re a software developer, here’s an agent. You go write the docs. I’ll check back with you. You add some tests to this code. Richard Seroter [00:28:06]: Got it. And you add this new feature because I’m wondering about this. And you’re just coordinating responses and things, and maybe it won’t be that extreme for everyone. But this idea of multiple independent agents doing work that you’ve directed, which then you pull back together when it’s done, that’s part of the future, one way or the other. And so how do you get ready for that? Look. The most important programming language for the next few years is English or whatever, your spoken language. How do you communicate intent to these agents? Because if you write terrible specs, you will get packed terrible responses from these agents. So how do I convey my intent effectively? How do I think about communicating guardrails? Hey. Richard Seroter [00:28:43]: Hey. Don’t do this. You should only be scoped here. Or, hey. Don’t don’t, you know, do things that are insecure. Like, how do I know enough? This is where we still need expertise. So if I give incomplete instructions, I’m gonna get incomplete results. So we gotta keep building our expertise so that we can properly narrate this, And then we’re just doing work all over the place at a faster speed at higher quality, but there are prereqs to getting that right. Jordan Wilson [00:29:07]: So, that’s number four. Let’s go straight to number five. So just AI rolling out everywhere, and it’s like I kid you not. I have a a working notebook l m document of just everything Google Gemini rolls out because it’s hard every day. Every day something’s coming out. But tell us, Richard, what does this ultimately mean? Because even, right, like, Google AI mode is, you know, coming out with with with Canvas and and and all these other right. It’s it’s all these features that I’m using from different products are seemingly being rolled into even just Google search, but just Gemini AI everywhere. What does this look like? Richard Seroter [00:29:47]: Part of me hopes that we stop worrying that it’s AI pretty soon, and it’s just we have smarter things. I mean, I can go to google.com/ai, and I just get a a more interesting way to search. Or I can use Google Sheets and have an AI function that can help me build a table real quick. It’s cool. It’s awesome. Or I could use our public cloud and Google Cloud and go to BigQuery and just explain what I wanna do and have it turn my natural language into a proper SQL statement, which I forget how to write now. That’s amazing. I don’t even care that it’s AI personally. Richard Seroter [00:30:16]: I just care that it’s smarter. And, again, I don’t have to know everything to do anything. And so as you look across all these tools, I think you’re just gonna see these interfaces have changed. Like, the there’s a it’s the new interface of technology. Right? It’s not just UIs or APIs or, you know, it’s AI. AI is a new interface. And so how does this just make a smarter way where I can express my intent to some of these systems and build a pivot table, write a resume, understand a document, perform a search. You know? And then when you get into people who build agents, how can we change customer support? How can we change onboarding and knowledge management? And so I think what the future is just gonna look like is is a smarter way to interface with these technology systems without being locked out by our lack of knowledge about the nuance of every programming language and syntax and tool and how do I call this? I don’t care. Richard Seroter [00:31:06]: I just wanna go on vacation. Lock it in that system. I don’t I don’t even care what system it is. And so I hope we just keep returning more autonomy to the human by offloading a lot of this different mucking around between systems to the agents and and the AI that can do it. Jordan Wilson [00:31:22]: Alright. So, Richard, we’ve covered a ridiculous amount of great information on today’s show. So everything from talking about cultural change to shifting our mindset of being we’re all builders, demos over memos. I mean, I think we’re literally just going to put up, like, 50 of your quotes in today’s newsletter. But, you know, as we wrap up, you know, after going over these five simple AI strategies, maybe what’s the one most important takeaway? Because, you know, you and your team are really helping build the future of work. So maybe what is your one strategy or one most important takeaway for people to be able to take advantage of all these things that we’ve talked about today? Richard Seroter [00:31:59]: Yeah. I don’t I mean, I think to some extent, either AI is gonna happen to you or you’re gonna happen to AI. And I think you have to decide if you’re gonna lean in or not because this is all coming in some way, shape, or form. If you wanna be ahead of it and then be in control of it, be smart about it. Understand the best ways to use it. Make it work for you. I don’t wanna work for AI. I want it to work for me. Richard Seroter [00:32:18]: But that requires me to lean into it then and understand how to use it well and stay up to date and listen to podcasts like this and lean in. Because you know what? Plenty of people won’t. And I think that you wanna be on the side that’s shaping how this industry is going to look and how work is going to look, not just be subjected to what’s gonna happen to you. So take some command of the of the scenario by being smart, going hands on. I think everything we talked about today has free tiers of service. Try stuff. You know, to me, the most important two traits every single human should have right now in in professional world is curiosity and humility. Be endlessly curious. Richard Seroter [00:32:52]: Keep learning. Don’t ever calcify your knowledge because it’s changing every week. And, yeah, be super humble because all these opinions you have today are probably wrong next week, and it’s okay. So if you have those two things, you are set up for success. Jordan Wilson [00:33:04]: My gosh. What an inspiring and, invigorating thirty one minutes from Richard. Richard, thank you so much for taking time out of your busy day to join the Everyday AI Show. We really appreciate it. Hey. Thank you so much for having me. Alright, y’all. And if you miss anything, don’t worry. Jordan Wilson [00:33:19]: We’re gonna be recapping it all in our newsletter. So if you haven’t, go to your everydayai.com. Thank you for tuning in. Hope to see you back tomorrow and everyday for more everyday AI. Thanks, y’all. Midroll [00:33: