In their latest video and podcast, Randy Johnston and Brian Tankersley, CPA, review ZOHO in their roundup of AI systems for accounting firms. Watch the video, or listen to the audio podcast below (transcript below):
Or use the below podcast player to listen:
Transcript (Note: There may be typos due to automated transcription errors.)
Brian F. Tankersley, CPA.CITP, CGMA 00:00
Welcome to the accounting Technology Lab sponsored by CPA practice advisor, with your host, Randy Johnston and Brian Tankersley,
Randy Johnston 00:10
Welcome to the accounting Technology Lab. I’m Randy Johnston with my co host Brian Tankersley. Brian has recently attended the Zoho day 2025 with and got to hang with a lot of the analysts. Typically, I go as well, but was unable to this year, so I asked him to be very thoughtful about watching what was being said. We know that the founder and CEO had stepped down earlier in the prior week to the event. So we knew there’d be some items related to that, and as we expected, they also had a fair bit of AI coverage. So what we’d like to talk about in this episode is the AI in practice, as Zoho sees it. And again, we believe that the way Zoho is applying this is the way you’ll see it in other vendors. And I believe that their methodologies, their privacy, their strategies, the way they’re using agents. And of course, they’ve been around with, you know, ask Zia Since 2016 I think nine years ago, if I’ve got my dates right on that, and they’ve made other AI announcements in 2018, and later. So you know, these breakthrough with the agents were the big announcements during the event. So Brian, I’ve probably said too much. I’ll let the more informed authority speak up.
Brian F. Tankersley, CPA.CITP, CGMA 01:39
Well, first off, I have to thank, I have to thank Sandy and and everybody at Zoho for making it possible for me to go to their to Zoho day 2025 it was down at Horseshoe Bay Resort, about an hour outside of Austin. Beautiful experience. Got a lot of really good brisket, including some good brisket chili down there, and their hospitality was just outstanding. Now, the presentation that we’re going to listen to, that we’re going to we’re going to do here, was originally created by Raju vaisna, I believe, is his name. And Raju is the chief evangelist at Zoho, and he is like Mike at Acumatica in the previous episode on Acumatica, Raju is one of the brightest minds in technology platforms out there today, in my mind, that we get to interact with, okay, and he’s interacting with, you know, with Forrester and with Gartner, and with IDC and all the other folks like that. And so I was very privileged to get to to get to hang out with them and just kind of see the reactions to these things. But this was actually raju’s keynote, a significant part of it, and he said that Zoho was really focusing on three areas. There’s the Zia tool that’s built in now. There’s the the Zia agents tool that is kind of an equivalent of the of the power power virtual agents that is part of the power platform for Microsoft. And then there’s the Ask Zia tool that where you can do interactive querying of things. But the goal here, the first one we want to talk about here again, is Zia, which is contextual intelligence. And the idea here of Zia, right now, in in Zoho, is that it’s doing all these things, but you don’t even know it’s doing anything, and you’re certainly not getting an incremental bill for any money, any additional money inside of Zoho, one for the use of their AI. So this, in my mind, was the money shot for the for the AI within Zoho, because it really shows you some of the different kinds of AI that are incorporated in and some of the different kinds of things that it’s doing inside your applications that you may not have realized were happening in the background.
Randy Johnston 03:47
So when Raju and I first began talking about this, 567, years ago, we were talking about very specific use cases, grammar correction or OCR improvement, or trying to do handwriting recognition, or being able to deal with bots and so forth. And I think what’s happened here with Zoho is they’ve continued to to find more and more business cases where they can apply Zia and get a benefit. So in their security division, the malware detection, the phishing detection, those are all great uses of this technology. Phishing is another one. And of course, we know in the Zoho workflow that they’ve been trying to take advantage of this for a while. So again, many, different business use cases. So I agree with you, Brian. You know, when you think about the money shot and all the use cases, this is a good one.
Brian F. Tankersley, CPA.CITP, CGMA 04:47
Well, if you think about you know, you and I have been writing with word processors since the since the beginning of them, and if you think about the things that word perfect would do way back in. Uh, you know, way back in the late 80s, early 90s, versus what we have today in Microsoft Word and Zoho, right? And and all of the other, all of the other tools out there, it will almost write your documents for you. It, you know, if you add on the the generative AI tools now, so, so I want you to see that, that even though you may not have noticed these features getting incorporated into your Word processors and your other tools you’re using over the years, I want you to see that these things are in their platform. And they’re, again, they’re, they’re drilling all the way through. And again, they emphasized here, just like our friends at Thompson, emphasized that, that they didn’t charge for API calls. Zoe says, Hey, we don’t charge anything for this contextual AI, it’s just part of our platform, and we have it in here. Now, they also talked about the Zia agents tool. And so what this does, this is kind of their tool for for creating, creating tools that will solve different business problems for you, the idea is we have four areas, data, tools, llms, large language models and skills, and those are put into a trusted platform and used within Zia agents. So the large language models have kind of become the gateway to agents. Now you have to decide which large language models you’re going to use, and with certain kinds of data. You might decide that you’re not going to use some of these things, or you might decide that you want to use, you might want to use some of the public ones, or you might want to use some of the self hosted ones in here. And so again, they’ve got a bridge that will connect to multiple external engines. You decide which ones you want to use with your data, depending on what your regulatory compliance requirements are. So again, with the agents, the idea is that with the data now, you can connect to Zoho apps. You can connect to external business apps, or you can use the Xia search tool to look at all the data stored in Zoho platform. You can also work with external platforms like, again, like Zoho analytics tool, or like snowflake, or again, Google Cloud, or other things like that. Or you can bridge over to your on premises, data lakes and and other tool sets in here. Now, as we again, a large, but a large amount of the data is in business, is for your business is in emails and documents. What they’re doing here is they’re trying to, you know, we’ve heard about data warehouses for years, and we’ve heard about data lakes for years. And of course, a data lake has all different kinds of data, structured, unstructured, semi structured in it. This concept of a data lake house is that you have structured, semi structured and unstructured data in rapid retrieval formats. So if you have structured data, for example, it’s going to be in a SQL database inside the data lake. And if you have unstructured data, it’s going to be stored bitwise in there with continuous data protection activation in here, that, again, is syncing with your with your with your sources, from where they are. The next thing we have are skills, things that the agents can do or or transformations, or other actions they can take. And so here’s another here’s another set of skills that they’ve developed. Again, this is another one of the brilliant slides Raju has created in here. I don’t mean to be too effusive in my in my praise of him, but it’s he. Every year, he blows me away with the way he He teaches me new concepts. And this is one of my favorites, because, again, you can see we’ve got handwriting recognition and trend detection and grammar correction and package detection and email assistance and tone correction and similarity recommendations and transcript intelligence and all kinds of things like this. In here, we then have a set of tools that gets us information and interacts with things. So this is the IO function, and so we have APIs and app actions and prompts in llms, we have code execution and browsers. We have flow tools like Zoho flow, that are integration platforms as a service, or as I call them, data data plumbing tools, digital plumbing tools. You have workflows. It can use computers to do things, and then all kinds of other agents. So here’s an example of an agent in here, where, where we were taking a Zoho form. We’re adding an action that will make it possible for us to insert an image to an app from a browser. And the so the what we’re trying to do here is we’re actually trying to set up a visitor self check in, and then we want to add, we want to add, right down here in the bottom, a photo capture so that, so that we can, when somebody checks into our office, they can sign the non disclosure agreement or the terms and conditions, they can take the picture. And we don’t have to pay somebody, you know, somebody, $100 a month or something for this service. We just do it within our system, and it also works. With our email and with our instant messaging tool called click in the Zoho suite. So here is the video of that, and this takes about a minute and a half, but you can see that that again, they’ve created the form. They’re now going out here into Zoho Creator, and they’re creating a new component to pull this information in. They have the AI assistant turned on, and so the AI assistant is going to create the code in Zoho, proprietary low code language called deluge. And so it’s going to now go out here and create this tool that will take the take the browser’s ability to talk to the camera, capture the information, and then create the component that’s needed to capture the picture of the person that’s showing up for the appointment. And so now we go through we we drag the component that it has created over to this, and now we’re going to test it. So we’re now generating the preview for it. You can see that the person puts in their phone number, their name, the email, the person they’re visiting, the purpose of the visit, and then they go in and they activate the webcam. And by activating the webcam, it’s going to turn that on. She takes a picture of herself. She agrees to the terms. She captures the photo, she agrees to the terms and conditions, and then submits it. And then, when it submits, it actually notifies that person through the click Text management, it inserts them into a report and a list. And you can see here that it’s actually flown through on the instant messaging for the person to let that person know that their visitors there and in the in the entry of the place. Okay, so any comments on that Randy?
Randy Johnston 11:45
You know, I was just thinking about the how many times we visited places like this, and how many firms are pretty loosey goosey with their logs and but just as important being able to see the visual log file, particularly in a Multi Floor operation, what a big deal and probably a security consideration over time, I’ve always recommended having pictures of clients, and that probably comes from the Know Your customer background of banking, but you know, the ability for us to identify visually is kind of a big deal, as I see it.
Brian F. Tankersley, CPA.CITP, CGMA 12:29
Okay? Thank you, sir. All right now, so you can see they have a wide range of tools, a lot of use cases. And again, they’re using the AI to deploy the tools, as opposed in a low code, no code fashion. So notice that you as a business user, if you decide what you want, you get the vision for what you want, maybe like, like Randy and I did by going and visiting tech clients and seeing the tools that they have, you can see that you describe what you want, and then the AI helps you make that happen. Now, when we look at the agents, one of the things we’ve got to also be very careful of, because it’s touching a lot of very sensitive data, and we want it to be very well controlled. And so the big problem here is that we’ve got to get the permissions right, and we’ve got to get the trust right. And so this, a lot of times, this permission, if you underestimate the permission layer, the privacy overlooked. This is a place where Zoho is platform really has, really shines, because they have built into Zoho one and into into this, the Zoho directory tool, which is actually used by manage engine and a lot of large enterprises. And so it has, it’s it’s got, they’ve got their own directory structure laid baked into this thing at a genetic level, so that you can, you can, you can use all of this data to do different things, but again, to get the AI right, remember, I don’t, I don’t want somebody that doesn’t have access to the payroll system to be able to ask how much each of the executives make. Okay? And so this, getting this permission right is is very difficult if you don’t control the directory or it’s not integrated. And so in that, in that way, Microsoft and Oracle, and again, these, these huge tech giants are going to have huge advantage if they’ve got their own directory system. Now, as we kind of drill down into this, the agent layer, we’ve got orchestration, planning and memory that’s happening in here. And again, the the idea is that is that now we want to take it, in addition to just solving the problems in here, we now want to be able to create our own agents to solve problems using these components that we’ve now described. And so they’ve got this platform that works across functions and verticals, and that will broaden existing skills and tools with a strong permission layer that’s built as a platform. Now this agent studio is similar to the Microsoft Power agents tool that’s part of the power platform. And so here is the Xia agent tool, and so it’s got got descriptions, and you can see this. One happens to use chat GPT, you have instructions to the agent, and then it connects up to it. They’ve actually created in the beta. It’s a private beta right now of for this, but you can see that they’ve created at this point at least 16 different agents to solve different kinds of problems, from customer retention to sales development to health care appointments to logistics coordinators to sales coaches and inventory managers and email assistants and next best action planner. You know, we always say that we got to keep doing the next right thing, and that’s what we’re really trying to do here. So what we can do is we can take one or more of these agents and put them together, and then we can create digital employees. Now we’ve seen the concept of digital employees in the past from companies like automation anywhere and others, but the idea here is that these these are truly taking a number of agents things together and a number of workflows together to digitally handle some of these kinds of tasks, much like the the cut the visitor log was handled by that simple workflow that was created in the first example, where these digital employees would take multiple workflows and multiple agent models together to comprehensively solve a particular workflow. And so when we look at this, in this digital employees, you’re gonna have provisioning and access control and policies and audit trails. You’re going to do some tagging of things, and again, you may have different business models. So these are some of the digital employees that they’ve considered. So a meeting assistant, RFP assistant, an Expense Assistant, you know, imagine how valuable it would be. Imagine how much your your life would be better with the sales people if you had an Expense Assistant that would just do their darn expense reports for them and make decent suggestions on them. Okay? So, so this is a pretty big deal, and it’s rich. It’s evolving very, very quickly. Okay, so con Any, any, any thoughts to layer on top of that Randy. You
Randy Johnston 17:02
know, I was just mulling over all of the Zia approaches. And of course, this is evolving. And just before you brought up this concept, everything’s moving fast. For Zoho, certainly moving fast. And when you consider their building block approach, and being able to click these building blocks together that that will make a difference, and the whole idea of an agent marketplace, if their partners will continue to build and release agents, which I think is highly likely, that Agent marketplace is likely to expand very, very rapidly.
Brian F. Tankersley, CPA.CITP, CGMA 17:42
I completely agree with you, and they’re doing, again, some very, very interesting things as we look though at this, at the search platform, again, this, this turns into their database. And again, we’ve got privacy first underpinnings that that, again, are built into the same permissions that you have in the Related Zoho apps and other tools in there. Right now, they’re working with self hosted llama and Mistral models. Okay, so notice that they’re self hosting because they don’t want the data going outside anywhere else. Zoho is very fastidious about data not being abused and being used appropriately, and that’s a in my mind, that’s that’s one of the things that really makes them more of a partner, more of a viable Partner In this information age, because we don’t have to worry about them pimping out the data in bad ways.
Randy Johnston 18:35
And they have always, always been focused around privacy and also around performance. You know that, I know it’s been about 10 years ago. I think it was 2015 that they first built their custom ASICs, and they, I know, spent some time talking to you, Brian, about, you know, using Nvidia chips, but using their own ASICs and some other things as well. So, of course,
Brian F. Tankersley, CPA.CITP, CGMA 19:02
an ASIC is an application specific integrated circuit that is a chip that is designed for the specific thing, the specific task that they want, they want to have done, and so, so that’s what an ASIC is, Randy. And I get Randy and I Randy’s Randy. So used to talking tech that sometimes I have to, I have to take these acronyms and explain them here, just because, you know us accountants, we don’t, we don’t live, breathe and smoke things. You know Randy. Randy designed that little red thing that’s in the middle of the ThinkPad laptop. So, you know, he’s been around the block a couple of times,
Randy Johnston 19:35
a couple of times, but Brian, I appreciate the explanation here on the AI initiatives inside Zoho. So any particular parting thoughts on what you learned about the new AI insights for Zoho in practice?
Brian F. Tankersley, CPA.CITP, CGMA 19:53
I you know we talked about how Acumatica is addressing this with Mike’s presentation in a previous. Podcast in and if you haven’t seen that one, I encourage you to go see it or heard it. We also wanted to talk about Zoho here, because I think Zoho entire platform, you know, the the thing about Zoho is they’ve traditionally been an enterprise platform, but they they also have a very viable solution in Zoho, one for small and mid sized business, and this automation that this enterprise architecture and automation that they’ve baked into all their applications to make them all talk to each other, this level of automation is how enterprise businesses like, you know, like Walmart and and target and, you know, the the huge fortune 500 companies have been this is how they have really, they’ve digitally optimized their businesses. Okay? And the cool thing about this now is, with AI, we don’t have to have an army of programmers the way they had to have to implement SAP and Oracle and all these other things. The beauty of this now is it really makes it possible for us to go in and implement these things in small, mid sized businesses, and you to implement even in your practice. And so that’s the game changer here is that this is now possible for you without an army of programmers.
Randy Johnston 21:12
So you can see why Zoho continues to be a product we like to recommend at all levels, but it is especially applicable in smaller businesses, and there’s even more to learn as we talk to you about Zoho features in other accounting Technology Labs. I know you’ll join us again. We look forward to having you with us. Good day.
Brian F. Tankersley, CPA.CITP, CGMA 21:38 Thank you for sharing your time with us. We’ll be back next Saturday with a new episode of the technology lab from CPA practice advisor. Have a great week.
= END =
Thanks for reading CPA Practice Advisor!
Subscribe Already registered? Log In
Need more information? Read the FAQs