How to Stay Irreplaceable When Agents Write Your Tests

My last podcast was with Mike Lawson, right? And he had the exact same thing. He said, "I got fired from my job because I gave my agents so much of my understanding of accessibility and how to test these complex systems. They no longer needed me." But is that true?

Was he really not needed? He was able to be replaced by an AI agent. You've actually seen that in real life.

Yep. And and I'm the same. A few moments later. Is what is your new position in this new world? Because you've got a really big career in front of you, right? We're not having to worry about being replaced. We just need to learn how we add value into it.

Same question, same conversation, 20 minutes apart. What does it all mean? Well, you wanna listen to this whole conversation with the one and only Jonathan Wright, the Automation Cyborg. Check it out. Hey Jonathan, welcome back to the Guild.

It's amazing to be here. Love you, Joe. Big fan of the show. Longtime viewer, longtime fan, right? We must have known each other for what, 15 years now, if not a little bit more?

Yeah, totally, totally. Yeah, I guess like you're always on the forefront of things. I call you the automation cyborg. So I guess what's new that's bubbling up on your radar that you think people need to pay attention to? that maybe they're not thinking of? You seem to look like a few steps ahead of us.

I, I, I started my career in automation, and I've just done automation for 30 years. And only recently, and I would say only recently in the 6 months, is where everything's changed.

All right, hold up, hold up one second. You said 30 years of test automation, and then that the last 6 months changed everything. What specifically changed 6 months ago? Give me the one thing. Let's hear it.

The difference is the questions aren't the same questions as they were last year or the year before or or 10 years before that. The less focus now on traditional automation, should we say, you know, it's more now about how— and this is where we kind of got back in touch— it's around this, what is agentic, you know? And I kind of put that out because suddenly, you know, this, this kind of stat, it's a bit like when we were all doing DevOps, right? That suddenly everyone changed their job title from Unix administrator to DevOps engineer, right? And nothing really had changed apart from they were working in continuous delivery, continuous deployment. And now everyone's using AI. But I would kind of partially say that no one's using AI, right? You know, it's kind of that, you know, that 2%, which you'd normally say, okay, they're using AI. I think most people, which is where my, my message came, when they think of AI, they've, they've got the Hollywood side of it, which is they're thinking about large language models, right? And I know the last time we spoke, which I think was— it might have been STAR West a year ago, right?

Yep.

You know, we had that same thing, you know, and I'd just come from speaking at EuroSTAR and doing the keynote there saying, you know, this is artificial intelligence, this is machine learning, this is deep learning. This tiny little piece here is natural language processing, natural language generation, and that is large language models. And actually, that's not that exciting, and that technology, I don't think Everyone's looking to hit it with a— everything's a nail, we'll hit it with a hammer. When actually that technology has a use, it's good at translation of language, surprisingly by the title, but actually doesn't address the bigger problem. You know, the bigger problem of like, how do we test? How do we test better? And how do we test non-deterministic systems, right? How do we test AI-based systems?

Right.

Because that's kind of where I guess the difference is. And it's where the book came out back in 2020 with the Artificial Intelligence for Software Testing, Testing of AI-Infused Systems. It was really about, okay, how do I know that the system that I'm using, and it might be powered by a large language model, it might be powered by some artificial intelligence and, and, you know, traditional stuff, but it works differently to how the systems we tested before, and therefore everything's starting to change, right?

And so when they were saying AI though. Were they just referencing LLM, the LLM piece, and therefore the conversations weren't making sense to you?

They weren't making any sense at all. So like literally last year, there was Michael Bolton and everyone else were all at Eurostar. And it was the theme was AI is on trial. And about halfway through, we all came to the same agreement that large language models were on trial. it wasn't artificial intelligence, right? OpenAI and these chat interfaces and, you know, suddenly everyone's going, oh, I'll use that for testing, right? And then rightly so, a lot of people are in one camp saying, no, no, no, no, no, you should never use ChatGPT. But you know, there's this, I'm going to call it artificial stupidity side of stuff where it's like, okay, I'm using ChatGPT for everything, right? So that's one landscape. And then suddenly we've seen this word agents appear. And like, you drive through San Francisco and everything is agent, agent, agent, you know. And really, we're talking about artificial intelligence for workforce replacement, right? You know, I have an agent that did a manual task before, hopefully something really boring like sending some finance bills or stuff, and now I have an agent to do it.

Right.

Brilliant. You know, and agents aren't new. Again, they go back to 1990. The difference is We've stuck a large language model onto an agent, and now what was Microsoft Chat Framework from Bot Framework from 10 years ago is now, you know, Microsoft Copilot or something, right?

Are you telling me that the thing about Clippy, it's now available?

Clippy coming back, he's now he, she, they, them, is actually part of this open test platform that I've been working on, is that you can actually, like I said, with the voice capability, you can speak to Clippy, and Clippy will use context reasoning. So, you know, we saw this quite quick emergence where even myself and Tariq and, and Jason all had something to blame in the early 2020s when we were saying, oh, become a prompt engineer. And then, you know, 2022, 2023, people discovered RAG, retrieval augmented generation. They all got really excited that they could stick, like Bill Gates told us in 1980, 640K is all you'll ever need in memory. You put your 640K markdown file and into a RAG, add a large language model, bish bash boom, it now knows everything about testing, right? You know, you drop in the ISTQB book or whatever, we're all good. We don't need testers anymore. And then it'll take the test, the ISTQB theory practice, and it'll pass it like Jason did 5 years ago. But that doesn't mean it's a tester, right? And I think this is where suddenly. Testers aren't like developers, right? Maybe they're a little bit more savvy or more questioning. They've got the critical thinking, whatever it may be, the special sauce that makes testers very unique. Then sat there going, wait a second, if I put this in, it doesn't give me the right answer, or it doesn't— they're questioning everything, which is brilliant. But if you think about how much developers have just dived in. the testing community are on the fence going, oh, actually, you know, well, let's run a pilot, or let's check to see if what's done manually is as good as what's being done automation. You know, this is the same conversation we've been having since the '90s. It's like, oh no, automation does checking and manual tests do testing, and, you know, they're completely different things. But we're not— we're talking the same language, right? A computer was invented It was because somebody, some human, was doing a really annoying task, sorting checks or whatever else, and they realized, oh, a computer could do that, right? And then they realized, you know, we realized in the '70s, '80s, Dorothy did, and then, you know, you know, who's been doing it now for 40 years, that actually computers can also automate repeating tasks, right? Which is why we got Word. Well, we don't have mail merge anymore, but we have Computers to save us time, right? And in a way, everything's automation. And now agents are automating workflows. We've been here before, robotic process automation. It's, you're doing process mining. Joe sends out lots of emails. We just automate it for him, right? But we do okay by having a program or an automation script or some type of agent that nearly does 10, 20, 30% of the workload, right? But doesn't have all of the insight and the intellect behind it. And so agents have come along and then we're all suddenly going, oh yeah, we're using agents. So we're using agentic. Like literally my Slack's pinging me because someone, a lady just sent me a message that, oh, I'm building agents. Can I borrow some of yours? And they're doing stuff that I need doing, right? You know, they're doing computer use automation where they're trying to go through. Actually, one of them is Just reading your website at the moment and look at the 400 test tool vendors that are on there, right? Because it's just trying to work out and it's trying to learn intent because I've just asked it a question when I was waiting for the to get on the Zoom for you. I was like looking at it going, "Oh, okay." I'll ask my agent to go and have a look at this, right? And so I lean onto agents, and I I've spent the last two years building. You know, I launched an agentic platform last year, which went. GA. And as you saw from the Gartner quadrant and the Forrester quadrant, it sat there snugly in between Tricentis and UiPath, right? And at the point when you looked at that, you were looking at it going, okay, the landscape's moving, right? You know, it's a bit like what we saw when Mercury was there, right? Mercury had 50% of the market share.

Yeah.

And then what happened? Well, potentially the worst thing that could have happened to the automation industry in the world Which was selenium, right? Selenium being the cure for mercury poisoning. However, I did speak to Jason Huggs the other day, and he and he also mentioned that none of the team that ThoughtWorks had ever used any of the mercury products, right? So I don't know what they thought they were curing because they weren't actually suffering from the problem, right? And the problem that we were both suffering from with my with my mercury box up here, right? Actually, I think. That's the Mercury box was, you know, we've actually gone backwards 3 decades instead of going forwards because I'm using computer use agents now and I'm using the same technology that DeepSeek Harnish dropped last week, which opened up their KUAs, which actually allows you to take over desktops, takes over VNCs, taking over Citrix, whatever else it is. And I'm testing cyber robotic systems, right? So I've got, you know, the Omniverse, which is the NVIDIA platform for digital simulations of robotics. And on the other side, I've got my DoBot, right, which is about 5 foot 2, is a real robot just stood looking at the moment, not looking at anything because they don't, don't have a face. It has a hole there in the middle and it just has a LiDAR camera that's looking down, right? And on one side, I simulate that I want it to go over and pick something up off the desk because we're trying to do some stuff for a banking customer. On the other side, I'm trying to measure within 2 millimeters of accuracy. Is this robot doing— is the locomotion and all the different sensors working to do what it says it's doing? And it's actually moving in the direction it is. Now, I've not seen a browser for 5, 10 years, and that might make me different to most of the people who are on this call. But unfortunately, our overfocus of browsers has meant that we've not really innovated. If you think about all those smart guys in Israel who were Still the same smart guys. They were coming up with ways to deal with highly complex systems. And then what we've done is we've automated browsers, right? Which literally takes 30 seconds. And I can tell you how to do this. Just don't go into ChatGPT, just go to Codex or, you know, Claude Code and just type in, create me a CDP, you know, tool to replace Playwright, hit enter. Obviously the Playwright engineers have gone from Microsoft now, and then 30 seconds later you'll have something that's faster than ViVM, better than, more reliable than any of the other platforms because it's an easy problem to solve. But we've been over obsessed with it. So one of the things I did, and I did 4 months ago, is I jokingly launched OpenWrite, and OpenWrite was Like Playwright, but open. And what's really fun about it is it converts from anything to anything. So if you've got Tricentis, or you've got, you know, UFT OpenText, or you've got Appium, or you've got Robot Framework, you can convert it to any other language. So if you really want to go from Vibium to back to Appium, or go back to, you know, any of the other good tools that I've noticed your, your website ChatGPT recommends, it's so easy now and it's open, right? Because the real problem is, is scripting, right? Is scripting languages has kind of put us back. And if you remember, what were you doing in the '90s, right? When you were on that first Mercury event where you had the poster behind you, you were building frameworks, right? You were kind of going, let me data drive this or create keywords, right? That's what you're doing. And we've got that natural thing. It's like a, I don't know, it's like an evolution within us as automation engineers to create frameworks because we have to, right? Because we have to justify our worth when linear scripting, i.e., sequential coding, doesn't matter how you represent it. It doesn't matter if it's in TypeScript, Python. It doesn't matter what the language is. By defining it in a language, you're in essence limiting your ability to deal with complex systems. And I don't get me wrong. I tried to do this with model-based testing, and modeling's too hard to do. I tried it with digital twins. But what about neurosymbolic languages? Right? So if you you know I was chatting to Diego, and Diego's been saying to the last six years when he started writing about Turing bots back in 2019, right? He's been saying to me neurosymbolic languages, right? And I was like, okay, yeah, makes complete sense. And neurosymbolic languages are High-level English language, right? Intent, intent-based testing, right?

Is, is that the new, the new phase then? Because you've been saying, like, let's go back, you said prompt engineering, you were laughing like 2 years ago, because now it's not even the prompt, it's just like, like I have, I installed GroqBot, I created a QA engineer, I, I'm not saying anything, I'm checking in code, it knows what code I did, it's verifying, I, I didn't give it any, I didn't even give an intent. It just knows. So like, is that where we're going? And then if that's where we're going, where are we as testers?

Exactly. So I did a demo and I'm going to send it over so everyone can see it, right? For what I'm classing as intent-driven testing, right? For a second. And I just gave it an intent. Now the intent was less than 1,000 words. So when I said to it, test all of Joe's website on the test tool, and it went through and it It's saying to me, you've, you have got, which, you know, if you haven't checked this out, you need to check out your, your, your, your tool because it's not, there's no sponsored links in it. This is purely the, the podcasts you've got. You know, there is approximately, which you'll probably see in your thing, there's 306 or 305. Now it's gone through and it's checked all of those. And then because what we've been working on at the moment is Put out the confidence engineering stuff, which is the the how do we know that it's right? So it's gone off, but I've had to give it a high-level message, right? Which I kind of class as AI assurance. But you know, we call it intent for the second. I've said to it, go and have a look at Joe's website and test it. That's literally all I said. And then with that, it's gone through, and it's gone through every single one of those three hundred pages. And it's looked from a confidence engineering at, I'm just going to, I'm going to pick on, you know, Zaptest because I love Alex, but Zaptest Enterprise, right? It's got a cross-platform OneScript, right? And then we're talking exactly the same thing, right? Is he's created a capability called OneScript for web, desktop, and Android. Brilliant, right? And I loved what Alex did. And we worked on something called DesignOps, which was you take a Rocketry prototype. So let's just take. Figma, you see something and then you automate it right before the app even exists. What I'm saying with intent is you don't even need the prototype mocking up. You can just express the intent of the system and then model the entire system before a single line of code has been ever written. And this is what scared people because I think part of it is that we all got over-obsessed with shifting left being closer to, you know, the, the coding landscape. What we're really talking about is shifting really left. And this is where intent, but also this is where you spotted the confidence, the confidence asymmetry principle. If there's 2 areas, there's test verification, i.e., the halting problem, information theory, which means you have to run the system to make sure it works, right? There is no way you can get out of that, right? And if the whole point between the confidence engineering principle and also the asymmetry principle is that as code, as you've just rightly said, on your phone, you've got Groq. I don't use Groq. I do have one of my agents sat running quite nicely in Cursor, but I've, I've been a big fan of Cursor before the acquisition. But, you know, Groq will be generating slop. It's hard for you to tell what slop because the amount of code lines are going up. And therefore the complexity is going up, and therefore the only way you can bring complexity down and confidence back up is for a system to understand as much of that code as possible. So the verification era, which was what that paper was all about, was saying that now's the age of testing. You're literally— it's never been a better time to be a tester, and even better to be an automation engineer, because now The slop's getting generated at a higher rate than even DevOps could do with the whole 26 releases of nonsense to production a day. Now we've got infinite amount of generate code generation by lots of teams. Part of it is the quality that comes out, it's gonna be hard to measure, which is where confidence engineering is really important, right? Whereas on the other side, there's this missed opportunity is design verification. How do I make sure that my requirements, my plans, my designs of what I'm building are better, right? So if you think about the DevOps infinity loop, overlay that twice. When I've built something, you've built something with your mobile app, how do you learn about the blueprints of what that agent did? And I think your answer or your question around prompt engineering is not about prompts. It's about blueprints of verification blueprints. It's about improving how you build confidence and how you verify at every single stage. So for every tester who's listening, you've got another 10, 15 years of being rock stars because you'll come in and go, wait a second, you know, I might be doing, you know, test management lifecycle or incident management lifecycle. I might be doing guardrails. I might be doing, you know, UI API testing. I might be doing security testing. I might be doing performance testing. I might be doing accessibility testing. And each time I'm going to be creating better and better blueprints that are going to improve and evolve my agents. And so it's me plus my agents. And that's a difficult thing because I think we're like, the reason why we got on this call as well was people are all sat there going, oh yeah, I've got 50 agents working for me, 100 agents, but if those 100 agents are just creating more noise, then are they really adding value? Everyone's using AI, but only 5% of them are actually getting any value from what they're doing.

And so why is that? I mean, no matter how much automation we had, how much testing we had, how much development we had, if our requirements were always bad, it doesn't matter how many agents we have, right? So we're saying We need to actually shift as far left as possible to get the requirements right, which has always been the case in order for any of these technologies to be successful. Is that what I'm hearing? 100%.

And also this really scary idea that the people who are coming up with these ideas might not be right. I mean, they might not have thought it through, and this could be the business, right? You, if you remember the old days, IT was a delivery unit, right? And I would be in the mindset of saying they shouldn't be IT, right? right, that are shipping bananas or they're making cans of Coke, right? They shouldn't be an AI center of excellence. They shouldn't be in automation, building automation frameworks. They should be delivering the product to the customer, right? And adding value all the way through that, making it cheaper, better, whatever, right? So if I'm told by the business to build this new feature just because I have a team of 10 developers and they have a throughput of 100 velocity, Should the idea that they've come from the business be verified before we even put it out, right? You know, should that go through some kind of— if you remember the whole lean engineering kind of principles of saying, okay, let's prove that hypothesis out before we build any software. Now, if you remember the old days when you bought your favorite vibe testing tool, you just got straight into it. The first thing you did was type in Build me my mobile app. And it did it. And then people like Cairo came along and said, no, no, no, you should build your design first. Once you've looked at your design and you've created a goal and you've created a plan, then you build it, right? But that's still in isolation. That's just Joe, one man Joe. Whereas Stephanie, Joe, and all the A-team that makes Test Guild Test Guild and a family, they all have to work together. They all have to be sharing those plans. And those goals. And the goal is to be the best test guild with the best, you know, the biggest community, which is the biggest community in the testing world, right? So part of it is that's the goal, that's the mission statement, but each team have their own plan of how they work together to add value to that goal. And that's not in isolation. And that's what devs are doing at the moment. They're planning. They're, if they are using design, if they're using what I call priming, giving them the right knowledge. And they're learning from each execution and not just going, oh, I've delivered that, now I'm going to start a new thread. And I think this is where it gets really interesting, is we end up back at square one, right? We end up doing a demo. I'm not going to say who to, and it's spending 20 minutes to the agent to eventually press H on the keyboard because it's going to type hello onto a mobile device. And we're sat there going, Wait a second, you know, in 1990 we could connect to something in Type H in analog mode in about 30 seconds. Where have we come if we're still stuck at Hello World and we're still stuck at here's the login process with the standard flow when we've actually got problems like testing cyber robotics and physical AI? You know, last week I was doing space licensing, which sounded really cool, but it's in essence how you plan your missions in space and how you coordinate it between the different authorities. With agents and autonomous agents, right? Or autonomous agentic systems, right? But I don't know nothing about space. All I know about space is what you see behind me, which is genuinely what I've watched of Star Wars.

Do the testing. How do you know it's going to be testing correctly then as a tester?

I sat there on a call with your NASA's, your, you know, your SpaceX, whoever else who are all sat there trying to understand the requirements, the specifications of this new agentic system that will need different authorizers and evidence providers to look through mappings of space, you know, different regulations and different flight plans in essence, and also mission plans for what they call line of sight. So beyond line of sight, BLL. But each one of those has its own terminology, right? The CAA and the MAA, which is the Military Aerospace Authority, have different terms, different languages. So part of it is you start off with a mission I took specifications from all these different agencies, put them into a system, shook it up, and generated about 37,000 test case scenarios. And these are all, you know, boundaries, right? Is that the evidence for the certificate off the, you know, the side of the aircraft's gas certificate, which is a picture which someone's got on a desk with a signature on it, on the date on it says that it was only tested at this temperature threshold and said, okay, is this a legitimate boundary value analysis equivalence partitioning? Yes. Okay. And I continued down and I sat there with the MEA and NASA and I said, is this what you want? And they were like, we would have never looked at creating that payload through that agentic system with certain evidences. This would all have to be done with humans. And they keep on saying the same thing, which is humans are special. And only humans can do it. But the humans are the ones that created the system in the first place, and therefore it's their knowledge, it's their context understanding what we're talking about. And we're talking about how do we bring that context understanding of the industry plus testing as a domain, which is why we created OpenTest, as well as how NASA, how SpaceX do their special source. The endgame for all of these companies is to create new sciences, new materials, and use it for their, you know, to have 100 astrophysicists available on demand because they're able to, you know, synthesize that through agents and then discover stuff which they can patent, right? So the endgame for them is to create new sciences, new patents, new biotech, new, you know, cures for cancer, which came up this week, right? That's what they're trying to do. What are we trying to do as an industry? Well, we're trying to take some of that power. Some of that capability and apply testing to it, which means we have to share openly what the special source of all of us is, which is what I call transcendence, which is really horrible because it's when you're no longer valuable for your automation cyborg because you've given all your patterns away instead of you know being a contractor and being able to hold on to them building your special frameworks. You're kind of giving all of your knowledge and context understanding for your. special sauce and passing it to the industry so that everyone can use it, right? My last podcast was with Mike Lawson, right? And he had the exact same thing. He said, I got fired from my job because I gave my agents so much of my understanding of accessibility and how to test these complex systems. They no longer needed me.

But is that true? Was he really not needed? He was able to be replaced by an AI agent. You've actually seen that in real life.

Yep.

What are some actionable things people can do then? Because it's a big mind shift. And are we going to be replaced? You said it's a good time to be a tester, 10 to 15 years is going to be awesome. But then you just said you're going to be replaced by agents. So what are we talking about?

Human intent, right? Humans have so much value. They always have, right? Their domain knowledge, their industry knowledge, their experiences. You know, it's the same thing we've been saying on every call. You know, the difference is humans have got wisdom and understanding of where things have gone wrong. Which is why when you get rid of Lawson, you're losing out because you need the humans involved. Humans should be part of the process. And human intelligence is where I, as an overseer, which is the best thing I can describe if you're a Fallout fan, is I'm overseeing my 30 agents every day. And I, it's exhausting. I get up at 6, I finish at 9, I run 16-hour days, and I have done that for like the last month, but I've never been so busy in my entire life. And I've also never been so challenged mentally with difficult problems than I ever have been, because I'm trying to solve issues that are so complicated, which is why it got me so excited when I became an automation engineer in the first place, because every day was a different problem. It didn't matter if it was a healthcare company, it was an IoT problem. I came at it with a, okay, I'm a human, how would I try and solve this problem and then automate it, right? So if you're an automation engineer or a tester, you're not going to be replaced by AI. You're not going to be replaced by automation. You're not going to be replaced by a tool. You're going to have a really important role, which is bringing confidence to the business around building and testing these systems that are going to get increasingly complex. And you're going to be the person, the gatekeeper, that's going to be looking at the evidence and making decisions and calls that aren't black and white anymore. There isn't a wall of green. I've not seen a wall of green for 5 years. There's complex problems that require human overseers to be involved. So it doesn't matter what level you are. It doesn't matter if you've just come into this career. Testing is the way forwards, right? You should be getting super excited. If someone tells you about classification trees, it's another pattern you can apply and you can teach your agent to do it. If someone tells you about accessibility, it's another bow that you can now understand because you now know how WCAG and VPAT reports work. Plus, you know, the new accessibility standard, European Accessibility Standard, and everything you learn, you're getting more skills from. So part of it is it's super exciting. Just don't sit on the edges, which are either everything has to be automated, everything has to be done manually. or everything needs to be done by AI. They're all very extremes. You need to understand, and this is why it's been so hard for me, is what is your new position in this new world? Because you've got a really big career in front of you, right? We're not having to worry about being replaced. We just need to learn how we add value into it. And add value isn't by running 10,000 regression tests which don't prove anything because no one can remember what the requirements were, and they've all changed since. And what we're doing is fixing them so that we get a line of green and someone gets what used to be confidence but checking. We now have to every day use our nugging and start adding value in by asking the hard questions, asking the why, how do you know that's right? And once you're able to get into this new role, which is more of a detective, is an evidence gatherer, evidence engineering, approving Over time, things are shifting, data drift's happening, you know, uh, non-deterministic systems are working in different ways. Oh, I need to learn to do some red teaming. Oh, I need to understand how guardrails work because, you know, how do I put safeguards in to make sure that that evidence from NASA around military stuff doesn't get through to this other customer? How do I know how to do— get the right test data that I need? How do I, you know, utilize all the tools that's on your 300 database, but use them as logistics, right? Be like developers, dive in, go and grab Whoopi, which is on the list there, right? And then use Whoopi stuff to learn how to use that skill because that's another capability. You might use computer vision and learn something new, and then you might add some value, but no one tool is gonna solve all problems. You can't use the same approach and expect the same answers the day after. Everything's continuously changing. Everything's continuously evolving. The challenges are evolving. The intent's changing. Just now be a first-class citizen, right?

Thanks again for your automation awesomeness. For links to everything of value we covered in this episode, head on over to testguild.com/a604. And if the show has helped you in any way, why not rate it and review it in iTunes? Reviews really help in the rankings of the show, and I read each and every one of them. So that's it for this episode of the Test Guild Automation Podcast. I'm Joe, and my mission is to help you succeed with creating end-to-end full-stack automation awesomeness. As always, test everything and keep the good. Cheers. Hey, thank you for tuning in. It's incredible to connect with close to 400,000 followers across all our platforms and over 40,000 email subscribers who are at the forefront of automation, testing, and DevOps. If you haven't yet, join our vibrant community at testguild.com where you become part of our elite circle driving innovation in software testing and automation. And if you're a tool provider or have a service looking to empower our guild with solutions that elevate skills and tackle real-world challenges, we're excited to collaborate. Visit testguild.info to explore how we can create transformative experiences together. Let's push the boundaries of what we can achieve. Oh, the Test Guild Automation Testing Podcast. With lutes and lyres, the bards began their song. A tune of knowledge, a melody of code. Through the air it Spread like wildfire through the land, guiding testers, showing the secrets to behold.