A (0:00)

Good morning, good afternoon, or good evening, depending on where you are in the world. And welcome back to episode seven of the AI Lean Edge podcast, the show where we cut through the hype and make AI simple, practical, and profitable for small business owners like you. I'm Brian Bricker, coming to you from my office at AI Lean Solutions. And today we're diving into a topic that affects every single business, accounting. Although today we're focused on small accounting and bookkeeping businesses, all small business owners will find some accounting nuggets throughout the next 20 minutes or so. So today we have some great information for you. An update on the current state of AI since things are changing at a fast and furious pace. A book review of another one of my favorite Lean books, and a deep dive into using Lean and AI to enhance your your accounting practice. Because, let's face it, accounting is the language of business. But for many owners, it feels like ancient Greek. Endless receipts, unpaid invoices, and tax season stress. If you've ever wished for a smarter way to handle your books, stick around. Okay, so let's get into it chaos

B (1:11)

out clarity in Cut the noise and dial in the wind Hey, I Lean Edge Stream it dream it lean it tight Edge Everything just works right hey, Eileen Edge. Say goodbye to the sludge Run it clean run it mean.

A (1:36)

Hey, everyone. Welcome back to AI Lean Edge, the podcast where we dive deep into lean thinking and the world of artificial intelligence. To keep you ahead of the curve, I'm your host, Brian Bricker, and today we're starting with the most significant advancements in AI tech over the past 10 days. If you're like me, you know AI is moving at warp speed and it's easy to miss the big stuff. But don't worry, I've got you covered with a rundown of the key breakthroughs that could shape the future. We'll aim to keep this around 10 minutes, so grab your coffee and we'll jump in together. First off, let's start with Microsoft's big splash. On August 28, they launched their Mai models, including Mai Voice 1 and Mai 1 Preview, amid some reported strains in their partnership with OpenAI, makers of ChatGPT. These models are designed for advanced voice interactions and multimodal capabilities, so they're pushing the boundaries of how AI integrates into everyday enterprise tools. So what's exciting here is that Mai Voice 1 excels in natural language processing for voice commands, making it really a game changer for virtual assistants and customer service bots. This comes at a time when Microsoft is doubling down on its own AI infrastructure, potentially signaling a shift away from heavy reliance on external partners like OpenAI. And if you're in business tech, this could mean more customizable in house AI solutions that reduce costs and improve efficiency. Sticking with August 28th researchers unveiled a new procedural memory framework that's probably promising cheaper and more resilient AI agents. This breakthrough focuses on enabling AI to remember and apply procedures over time, much like how humans build muscle memory for tasks. Traditional AI agents often forget context after a session. You know, as you you probably have discovered this if you've been chatting with an AI, you know it doesn't always remember what you've already talked about. But this framework allows for persistent learning, making agents more reliable for long term applications like autonomous driving or robotic assembly lines. The best part? It's designed to work with lower computational resources which could democratize AI development for smaller teams and startups. This isn't just incremental, it's a step toward AI that's truly adaptive and less prone to errors in real world scenarios. Now one of the most mind blowing updates came from Singapore based Sapient on August 28th as well. They introduced a reasoning model with just 27 million parameters that outperforms giants like GPT based models Claude and Deep Seq on the ARC AGI benchmark for context on this ARC AGI is a tough test for measuring progress toward artificial general intelligence and this tiny model uses brain inspired multilayered processing with iterative refinement bursts. Instead of the usual chain of thought prompting, it mimics human like hierarchical thinking to solving problems faster and more efficiently. With only 1,000 training samples needed, this could slash training costs dramatically and make advanced AI accessible on everyday hardware. We're talking about a paradigm shift where size doesn't equal smarts, efficiency does, and in lean thinking we love efficiency. So shifting gears in generative AI, the last week of August saw a flurry of updates that changed the game forever. On August 30, reports highlighted rapid releases like Google's Nano Banana. I highly recommend it if you haven't played with it yet, which is likely a nod to an ultra efficient version of Gemini 2.5 flash for image generation, which is now available to everyone. Then there's cling 2.1 which supports start and end frames of 1080p resolution for video generation, making it easier to create seamless clips. Speak 2.0 brings advanced voice synthesis and concepts like digital Twins are definitely gaining traction for simulating real world objects in AI environments. And these aren't isolated, they're part of a broader wave where generative tools are becoming faster, higher quality and more user friendly. Imagine editing videos or generating images in real time without massive compute. That's the direction we're heading. OpenAI wasn't sitting idle either through any of this. Throughout late August, whispers of GPT5's launch surfaced, complete with the GPT OSS for open source accessibility. This model boasts human level intelligence in domains like competition math, achieving near perfect scores paired with enhanced reasoning and efficiency. GPT5 represents a leap in large language models, but it's not just about the closed version. The open source push could accelerate global innovation, letting developers tweak and deploy it for custom needs. This ties into broader discussions on unhobbling AI, removing constraints so models act as autonomous agents in the economy, handling multi week projects independently. If realized, we could see AI as drop in remote workers, boosting GDP growth by about 30% annually. On the hardware and infrastructure front, September 5 brought news of OpenAI partnering with Broadcom to develop custom AI chips. This $10 billion deal aims to optimize chips for AI workloads, reducing dependency on Nvidia and potentially lowering costs. It's a strategic move as AI compute demands are skyrocketing. Meanwhile, Sierra, an AI startup, hit a $10 billion valuation, underscoring investor confidence in next gen AI plat. And of course, Deep Seq is building advanced agent AI, focusing on autonomous systems that could totally revolutionize industries like logistics and health care. Google made waves this this week too, with VO3 integrating into Google Photos for animating still images into videos. Announced around late August, this uses advanced diffusion models for realistic animations, turning your photo library into dynamic stories. It's part of Google's broader push, including next gen healthcare AI tools unveiled at GOOG IO 2025 but highlighted in late August updates like Genai for imaging interpretation and clinical documentation. These could transform medicine by speeding up diagnoses and reducing paperwork for doctors. Xai's Grok got an update with Grok Code Fast one, emphasizing speed in code generation. This, along with other models like Genie 3 and AlphaFold 3, shows how even smaller compute setups are yielding magical results. But the real excitement? Massive supercomputers with over a million GPUs coming online by late 2025, which could unlock unprecedented capabilities in AI. Other notables include Mistral AI targeting a $10 billion valuation, Databricks acquiring Tecton for better agent offerings, and Salesforce's Agent Force for public sector tasks like tax consultations, plus emerging research on multi robot orchestration and reinforcement learning for physical systems hints at coordinating complex real world operations, which is what I know that I've been waiting for. We need to take AI from the the just the informational resource it is and plug it into real world operations. Wrapping up these past 10 days highlight AI's relentless pace right now. From efficient tiny models to autonomous agents and custom hardware, we're on the cusp of AI reshaping jobs, healthcare and creativity in ways we can barely Predict. Predictions like AGI by 2027 feel more plausible with these straight line improvements in compute algorithms and unhobbling. Keep an eye on scaling laws and agentic AI because they're the ones to watch and we'll keep an eye on them for you and give you updates on future podcasts. Thanks for tuning in. If you enjoy these updates, please hit subscribe, leave a review and share with with another small business owner. Drop your thought comments below and and let us know what advancement in AI is exciting you the most right now. Okay, moving towards our main topic today. Accounting and bookkeeping as we know are essential, but they can be exhausting. I want you to picture this. A small bookkeeping firm I spoke with recently spends more than 15 hours a week just entering receipts into their system. That's nearly two full working days each week on a task that adds no strategic value. This is the type of work AI was born to take off your plate. Not to replace your accountant, not to make things more complicated, but to handle the heavy lifting so you can focus on what really matters. Growing your business, serving your clients, and having a little more life outside of work. So let's start with the pain points. If you're a business owner or accountant, these will sound familiar. Receipts scattered in shoeboxes or globe compartments, late nights reconciling statements, the constant stress of chasing clients who owe money, and of course, the yearly ritual of tax season panic. Now here's the hard truth. These pain points don't just cost you time. They cost you money. They cost you morale and sometimes even client trust. I've seen firms lose clients simply because invoices were delayed or reports were late. But here's the hopeful part. AI is not some vague promise. It is already solving these problems. Let me walk you through exactly how AI thrives on repetition and rules based tasks. It recognizes patterns, applies rules, and learns as it goes. That's why it fits perfectly into accounting. Here are a few examples. Data Entry and Categorization AI can scan a receipt, pull out the date, the vendor and the amount, and drop it into the right expense category. Expense Tracking Once it sees your preferences, it starts to auto categorize with surprising accuracy. Fraud detection AI flags odd transactions that don't match your usual patterns. Invoice reminders, automated nudges keep cash flowing without you making awkward phone calls and trend spotting. AI can show you seasonal spending patterns or alert you to rising costs. Think of AI as an assistant that works quietly in the background. It doesn't replace judgment. It doesn't replace relationships. It gives you space to be the expert and advisor your clients actually need. Before we go deeper, I want to take a quick break and tell you about something exciting. My new book, Smarter, Leaner, Faster How AI and Lean Thinking Can Transform your Small Business, is available on Amazon right now. This book is a roadmap for small business owners who want to cut waste, harness AI, and work smarter instead of harder. It's packed with case studies, checklists, and practical strategies. If you've been enjoying this podcast, the book takes these ideas to a whole new level. There's even a section that outlines how to run your own AI Lean Audit for accounting tasks, including which tools to test first and how to measure the return on investment. I also share industry specific checklists like how to cut invoice delays and streamline reconciliations with AI. So if accounting is your world, this book has direct takeaways you can use right away. Order today on Amazon and be ready to hit the ground running. Just search for Smarter, Leaner, Faster on Amazon. All right, back to our topic. Let's look at the tools that actually deliver results. QuickBooks Online It's a classic, but with AI categorization that learns your habits. This alone can save you hours. Every month, Xero and Hubdoc snap a picture of a receipt, and Hubdoc pulls the details right into Xero. No typing, no mistakes. Dext Formerly called Receipt Bank, Dext is beloved by bookkeepers because clients can upload receipts from anywhere. AI extracts and organizes the details automatically. Then you've got Zoho Books and Zia Zoho's AI Assistant. Zia can forecast cash flow, send reminders, and automate workflows for small firms. It's a surprisingly powerful system. At a lower price point, AI writing assistance tools like Grok, ChatGPT, or Claude Draft emails, explain reports in plain language, and summarize financial data for clients. That means fewer hours sweating over the perfect wording. These tools aren't just bells and whistles. They tackle the very pain points we listed earlier receipts, invoices, reconciliations, and communication. And here's the kicker. Most of these tools already exist inside platforms you might be paying for right now. The challenge isn't finding them, the challenge is turning them on and letting them work for you. So I want to pause for a minute and review a book that has shaped my own thinking. It's called the Lean Toolbox for Service Systems by John Boscheno. Most people know lean thinking for manufacturing, but Basheno focuses on applying lean to service industries. And accounting is very much a service industry. He emphasizes concepts like flow, variation, and customer value. For example, if your accounting process creates delays for your clients, that is waste. If your invoices are unclear and lead to confusion, that is variation. What I appreciate most about this book is that it provides not just theory, but practical tools, flowcharts, frameworks and models that you can adapt to your firm. If you are serious about combining lean thinking with accounting, this is a book worth having on your shelf. Again, it's the Lean Toolbox or for Service Systems by John Baseno. All right, let's bring it home now with how to get started, because I know you might be thinking Brian, this sounds great, but I don't have time to learn another system. Here's my advice. Pick one tool. Just one. If receipts are your headache, start with Dexter Hubdoc. If invoices keep you up at night, set up automated reminders. Train the AI, so correct it when it miscategorizes and treat it like a new staff member. Review weekly. AI is powerful, but you need to keep an eye on it and track the time you save. Even five hours a month adds up to 60 hours a year. It can be significant. Small wins do matter. You don't need to overhaul your entire system overnight. One smart step at a time will make a huge difference. So here's the big takeaway from today. AI in accounting is not hype. It's here. It's practical and it saves time. Whether you're a business owner doing your own books or a CPA running a firm, these tools can give you your evenings back and help you focus on higher value work. Don't forget to Download the free AI accounting starter checklist at aileansolutions.com in the Free Resources section. It's a simple one page guide to help you take the first step. So as always, thank you for listening to AI Lean Edge. Remember, if you like what we're doing here, like follow, share and review the podcast so we can bring even more value to small business owners worldwide. And don't forget to check out my new book, Smarter Leaner How AI and Lean Thinking can Transform your Small Business So until next time. Keep working smarter. Keep eliminating waste. Keep. And most importantly, keep leaning into AI.