The Secret Tool That Automates Anything (Better Than AI Agents)

Isaac shows how UI automation bots can handle repetitive tasks like logging into tax portals and reconciling transactions in accounting software. Using Axiom.ai, he builds a Google search bot, then demonstrates real workflows for downloading tax documents and bulk cash-coding in Xero. Unlike AI agents, these bots are predictable and secure when credentials are vaulted.
Chapters
Hey, this is Isaac, and in this video, I'll show you how to automate anything that happens on your screen. Even if the app you're using doesn't have an API or doesn't have integration support, it's not supported by apps like Zapier or Make. Even if you're doing things in websites like the IRS or some other government agency, even if you're using legacy applications like tax software, or you're doing things within accounting software that which traditionally require humans, like checking if a bank fee gets disconnected. In this video, I want to share the tools, automation tools you can use to even in those cases automate a process. Namely, we're going to be talking about browser and desktop automation. That's the topic of today. So get ready to learn a beginner-friendly tutorial of what these tools look like, what these actually are, the pros and cons of using browser and desktop automation, When exactly do you use them and how have we used them within accounting firms? And frankly, also why they're often better than AI agents, at least today. So let's get started. Now, what are browser automations? Essentially, the concept is that instead of moving data in the background like you would with Zapier, with Make, with APIs in general, you would act as a human instead. And what that means is that you can build an automation that it's opening up a browser, going to a specific URL, clicking here, clicking there, downloading a report, filling out a form, and moving information, but specifically making clicks and keystrokes in a screen or a browser like you would. Now, when is it useful to use this? You might have seen this chart that I'll show on the screen now that really Explains our framework internally at Opsware as a company. We believe that the higher up you go in this pyramid, typically the more maintenance something's gonna need, but the more possibilities it unlocks. I'm not gonna go into the details, but you can see that RPA, it's the last step, and there's a reason for that because from our perspective, RPA is now relying on the website or the desktop software, and if those things change, Then that bot or that automation needs adjusted. Doesn't mean it's not worth it, but it's worth keeping in mind that from our perspective, we see that as a last resort. In terms of typical tools in the space, you'll see Power Automate as one that is within the Microsoft environment that has some browser automation and desktop automation functionality. The tool we'll be talking about today specifically, it's Axiom.ai. So the tool we're talking about today is actually a Chrome extension. But it can act as a Chrome extension, although you can have a desktop version as well. We'll be using the cloud version in this case. And when you— once you download that extension, you'll be able to click on it, and this screen will come up where you'll be able to see a little dashboard of sorts where you'll be able to see everything that you currently have set up for yourself. Let's see how easy it is to create an automation. And just for as an example, let's just say I want to create an automation that goes to Google and and looks for a brand name and that's it. Just as a simple example, just to showcase what the tool can do. So in this particular case, I can navigate, which I can go to a particular website. In this case, it's Google by default. It has some other options that allow you to potentially have a store cookie. So if you are logged in, you can kind of continue to use the tool. But in this case, simple example, I click on Google. I'm going to click an element. Guess what? If I am on Google, check this out, I can untoggle or toggle this back on. I know I want to click on this particular search bar. Check how easy that is. I select an element. This is the element I want. I'll click on complete. And that's it. From there, I can also enter text. What text do I want to put in? Let's just say it's observe. And you see here that it's kind of telling me like Instagram DMs, right? There's, if you've ever seen people say comment box in the comments and I'll send you this lead magnet or PDF, whatever, sometimes the way they then get back to you could be with a tool like this just for context. But then let's just say I also want to now press a key, right? After I type my name, I want to click enter. Here I have just press keys and I have the ability to record the thing I do in my computer. So I click on record and I hit enter. I'll see that enter come up. I can stop the recording at this point and I'm able to run this to see it go live. I could schedule this to go in and do whatever it has to do on a regular basis. And you can see some other things, other options that it has to run later on. If I click on run in the cloud, just for us to see this, Run live. We should be able to see another screen come up, and we just you should just simply see the Google page being visited, the search bar being clicked, option name being tapped in, and then enter. Interesting that NetSuite is competing with our terms, but that's as simple as it is. It's a very low-code visual development way of building bots. Most accounting firms think they're moving towards automation, but. But when you look under the hood, what we find is messy spreadsheets, siloed tools, and half-baked workflows that are still slowing them down. So that's why we built our AI and Automation Readiness Quiz. With just 10 quick questions, you'll be able to see where your firm is strong and where you're falling behind. Here's why it matters. Understanding where you stand is the first step our clients take towards a smoother, more scalable practice. That's exactly how Donna Bordeaux cut a full-time bookkeeping role and went from 90 hours a month down to 1 day a month of her time while scaling her firm from 45 to 60 clients without adding new staff. And that's how Gareth, who partnered in a Canadian tax firm, went from dreading every January to looking forward to tax season and actually earning 47% more per hour on their tax return work without price increase. So feel free to check the link in the description, take that quiz now, and see exactly where you stand. All right, now back to the video. Now I do want to talk about how we've used it. within other accounting firms and show you what that looks like. I have 2 examples to share with you. One, it's for a firm we helped them actually go into their government website, the IRD in this case, and download client mail letters on a daily basis and then push that into their systems of choice so they don't have to log in every day to pull that data. Other one I share with you, it's a bookkeeping firm that wanted to— for a bot to go into Xero 3 times a week And clean up everything that had a rule by kind of matching those transactions and cash coding those transactions. Let me show you what that looks like. So let's see this bot running and see it reconcile transactions in Xero. So I'm going to run this in the cloud. We have some of the data that this would use live. Of course, we're probably going to be blurring some things, but I want to kind of show you and walk you through kind of what it's doing in the background. It's clicking the login button, putting in the credentials. We'll talk about security in a second. It kind of like goes in. It kind of has to wait here for a few seconds, a few minutes to follow through this process. But once it is logged in, it's going to loop through all the data that it has. It has all the organizations that the client has and all the accounts. So it's going to go to each organization, to each account. And if there's data, then it's going to click the Save and Reconcile All. If there's not, it's just going to continue. That's essentially what it's doing. I'll pause for a second here so we can see it actually run and do its thing. And again, this is going to be quite blurred, but the automation is running through and seeing, hey, I've saved and reconciled these things. I'm going to move on to the next client or the next account for that particular client. And it just keeps going and going until it's done that for all of them. Okay. So in this particular example, it is a bot that's logging into a government website. Similar process of logging in, followed by getting all the client mail, downloading it for a specific timeframe, exporting it all, and then uploading that file somewhere else for this to run on its own. So we'll see it run as well. I'll pause when it is boring to wait, like it is going to be now. But the idea is that once it's logged in, we'll be able to see it do its thing. Okay. So now that we saw it logged in, You'll see that it kind of goes into the right links. It finds the specific link it needs to click on to see the client mail, and then it's going to pick the specific date ranges, as you'll see, to pick the right time and the right date ranges that it needs. And then it's going to download those letters to process on a separate automation. You can see, select everything, export to zip, and then it keeps going. So those are just a couple of examples, but effectively we've kind of seen other things. For example, with the IRS and downloading transcripts and other notices, or with QuickBooks where you have to like see if a bank feed is broken and notify your clients. Those are things that traditionally you would not be able to do, but with a tool like this, this becomes unlocked. Now, I do want to talk about security. As you see, there's often the need to store and share credentials within those tools. Those tools have oftentimes articles on the subject of how to do that safely, because you can still secure those credentials in a vault and then get them from there. We've worked with highly security-minded IT teams in bigger accounting firms, and what we've noticed is that as long as the value is clear of why we are doing this and the data flow, it's contained, and you understand how the data is flowing, typically it's doable. You know, in a secure fashion. In concept, I like to always say like, it's not that different for another human having access. And in practice, you could argue that it's maybe even safer because this is something that's kind of contained and not a human that might be able to, you know, share it with someone else. So there are ways to do this safely. And like everything in life, however, it's not always— it has some pros and some cons, right? So the cost is, of course, the initial time investment to build the workflow in the first place. Test it, make sure it was working correctly, and often the need to update the automations when the website changes their layout, which is typically not something we have to do with other types of automations. But the benefits are clear, right? You make a lot of progress on your tasks without human involvement, tasks that might have taken hours manually, and that frankly, sometimes we hear firms are just not doing because it's not worth putting a human to do it. Now those tasks become unlocked as a result of being able to do this. If of course it's scalability, it's another factor. We ran this particular cash coding example for, I believe, 60 clients are selling zero. If it was 600, it just takes more runtime and you pay a little bit more, but it would still work, right? And some— and also the reason why I like to say that this can be better than AI agents, it's because at least right now, AI agents can be unpredictable. They can make mistakes. Mistakes. From a security standpoint, they can— you can have a prompt being injected and have it act in a way you don't want, especially when you're dealing with sensitive data. That can be a problem. So browser automations and desktop automations will follow the exact steps we're defining, which is kind of like both— it's kind of like pro and con, right? So you have like consistent and reliable results every time, as long as the environment where it is doesn't change. I'd like to say too, you can still add AI To a, to a bot like this, but only where it's needed. Because if you want AI to do everything, that's, I feel like, where it kind of goes rogue or might not do what you want. It is a little bit more unsecure to attempt to do that. With a tool like this, you still maintain control over the process, which I know is very critical for compliance and security in the accounting industry. Now, as I said, this is literally a last resort, and there's many other things you can do in more reliable and secure ways. Which is why I want you to watch this next video, which is all around our favorite automation tool, Make.com, and 10 use cases we've done for accounting firms. So check that out next. And until then, stay efficient.
More from Isaac

This Boring AI Automation Could Save Your Accounting Firm 40 hours (But most ignore it)

How to Build AI Systems With Claude (So Your Accounting Firm Runs Without You)









