Live at the Leeds Testing Atelier: MCP and Glue Work
Should we start off by explaining what this is? Has anyone here listened to what a lot of things? Oh oh we've got a couple of people. Oh listeners. It's always very gratifying for us to have listeners because, you know, you can never be sure, can you?
Ash:No. No. Because well, we had a Christmas party as well, didn't we?
Ian:We did. And someone we didn't know came.
Ash:It was amazing. It was like in person Christmas party. Yes. And listeners turned up. It was incredible.
Ian:It was a shocker. So, for those completely ignorant, what is What A Lot Of Things, Ash? I've helpfully typed in the answer into our Notion page so that you can,
Ash:Okay. So our tagline is tech talk from a human perspective. And then you also wrote “Free The horse”, which I've got no idea what that means.
Ian:Well, I was gonna say something about the venue, but the most striking thing for me is the encaged horse that's outside on the on the on on outside in some direction from here. So, everything is is being recorded. So I have actually remembered to press the button. So, you know, first test passed. So it's all being recorded.
Ian:And unlike our normal episode this is a double first by the way. We've never recorded this in front of a live audience, which I'm assuming you qualify as. And, and, we've never had anyone else's voice in our podcast before, have we? No. So we
Ash:have I don't know how I feel about that.
Ian:Yeah. Depends who it is. So we do have a roving microphone over there. So we're going to do our normal structure, which is we talk about a thing that each of us picks a thing, to talk about. And then we do an episode where we talk about one and then the other.
Ian:And, because fairness is such a complicated concept, we've boiled it down to mean alternating going first. So it's a very fair podcast according to that definition.
Ash:Apart from, Ian, like, you'll notice the the split in technology on the on the desk where Ian controls all of these buttons and I don't control anything.
Ian:Yeah. I've got a mute button here. So I'm not afraid to use it.
Ash:So I feel like me and Ian are still quite far apart on the definition of fairness.
Ian:No. No. We we've agreed what fairness is. It's alternating going first. All this other stuff is extraneous.
Ash:Okay. I understand. All too well.
Ian:I did put in my little list of things that we should introduce ourselves, but you probably don't need introducing. But introduce yourself anyway.
Ash:Okay. Hi. I'm Ash. I'm one of the organizers of the Leeds Testing Italia, and we've been doing this for quite a long time now. And there's loads of familiar faces here, and I know loads of you from around the Leeds Tech scene and beyond.
Ash:So lovely to have you.
Ian:And I I'm Ian. And, I'm not an organizer of the Leeds Testing Atelier nor indeed a tester of any kind, as, the listeners will remember the beatings that I've received over the years in the, in episodes of What a Lot of Things for my, lack of attention to that this very important area. So I've been all ears today.
Ash:Well, people keep saying, look at my testing strategy, and then they just list some tools, and you did that. So I I had to do it. The the education piece needed to begin.
Ian:Well, AI could write a very, comparable testing strategy. Absolutely.
Ash:Absolutely. But what's the mission of the testing, Ian?
Ian:Well, that's what you have to tell the AI. Yeah. Alright. Sorry. We're we're heavily into the in jokes now.
Ian:Yeah. So we're gonna do the usual thing where we talk about a thing each and then after each thing, we're going to try and get you to talk to us about the things. So save up your wisdom as we're going through. And we're already three minutes behind schedule. Excellent.
Ash:This sounds like an episode of What a Lot of Things.
Ian:It is. Honestly, we we take about sometimes we we go well over ninety minutes in recording. And then
Ash:and then we then we talk about two things and
Ian:then we talk yeah and I have to do the editing which is often getting ninety minutes of wittering down to I don't know eighty five minutes or something like that that we can then release. I'm not sure we I think we might have arrived at the grim reality of of things now.
Ash:Of things? Oh. We're not gonna talk about the origin story.
Ian:Oh, we should talk about the origin story. Yeah. What's the origin story, Ash?
Ash:So me and Ian met at Liverpool tester gathering, if anyone remembers that epic event.
Ian:Oh, yes.
Ash:And it also turned out that Ian lived in Oakley. So same as me and Gwen. So it was like, oh, what a strange place to meet.
Ian:Well, this is a long way to go to meet somebody from the same town.
Ash:Yeah. Exactly. So we got to know each other. We share a a love of walking on Ilkley Moor, and around. And we also found out, obviously, that we work in tech together as well.
Ash:So we, we decided that let's turn these walks into podcasts. And you might be wondering where the name what a lot of things came from. And I'll hand over to Ian actually on this one. She's got like a demonstration.
Ian:I'm I'm glad you asked me that, Ash. People who listen to What a Lot of Things are sometimes confused by, the sound effects that happen in between the various segments of it. What a lot of things is a quote from the Clangers. Who used to watch the Clangers in the nineteen seventies? Well, that's some hands anyway.
Ian:Thank you. For those who don't, the Clangers are kind of knitted, woolly BBC things, but they talk to each other in a very distinctive way. And I've got my, my clanger impersonator here. So I'm just gonna give you a quick demo.
Ash:Jeez. Really good at it, isn't it?
Ian:I won't translate what that was. It was a it was a little bit, a little bit cheeky.
Ash:Yeah. So Sorry. We had a an audit done of the podcast to see where we were going right or wrong. And the, the person who did the audit said, is this a child's podcast? It was like, well
Ian:That's not true. That. Yeah.
Ash:So the the Clanger took a slightly further back seat.
Ian:Yeah. The the the
Ash:But still exists.
Ian:The entire cover art used to be just a picture of a clanger. Yeah. And and I think his point there was people looking at a picture of a clanger probably won't know that it's about technology. So, yeah, we, we we've we've improved our cover art, but the clanger remains.
Ash:Okay. So should we talk about a thing?
Ian:I think we should probably talk about a thing. So I'll I'll
Ash:begin a timer.
Ian:Inevitable. A timer. Is that because it's my thing, and you think I'm gonna ramble on for too long? Pretty much. Alright.
Ian:So
Ash:what's your thing, Ian?
Ian:What's my thing? Well, my thing is MCP, which apparently stands for lots of different things I've discovered. It stands for male chauvinist pig in some context. And, if you, like me, are very old and you watch the original movie Tron, then it stands for Master Control Programme. And there's a few other over loadings of that, but the AI world has come along and appropriated it to mean model context protocol.
Ian:Yes. So that was my thing. Yeah. So I think this is very interesting because it's one of these sort of historical, stuff. I'm trying not to use the word we have to be careful how we say the word things.
Ian:Sometimes it has a capital t and sometimes a small t. So, MCP is one of those things with a small t where people get very excitable about it and and it's the future and this is the best thing that's ever happened and it's wonderful in every way. And what it is is basically a way to connect an application to an AI so that the AI can use the application. That's basically what it boils down to.
Ash:Okay. So isn't that didn't we have agents before or is this something different?
Ian:Well, I'm glad you asked me that. No. So agents the definition of agents, like everything we had a big discussion last time about, semantic diffusion, which is the phenomenon that occurs when a term starts to the meaning of a term starts to migrate and become something bigger and different.
Ash:Like like
Ian:like vibe coding. Yep. Which means exactly what he said when he defined it in that message. But anyway, yeah. So
Ash:What all all coding using an AI?
Ian:No. No. Stop provoking me.
Ash:Okay. Sorry.
Ian:He's doing it on purpose. Yeah. So that's really thrown me. Why did I go to semantic diffusion? I can't remember.
Ian:So but the the thing there, I suppose, is that the AIs want to use oh, I see agents. So the agents are things that can are AI things that can do stuff for you rather than you could just talk to it. So a language model can book your tickets or send an email on your behalf or something like that. And MCP is a sort of standardized protocol released by Anthropic who who are known for Claude. The, they MCP is their thing that allows it's like a standard protocol for doing that.
Ian:So, a tool can have a description. So let's say it's an email sending tool. Yep. The description might say, you can use this tool to send emails, which you should, request using the following JSON structure or something like that. And then the AI knows what it can do with the tool.
Ian:And if it thinks that as part of the conversation it's just having it wants to send an email, then it will be able to do that. So and that's the that's, I think, is the nutshell explanation.
Ash:Okay. So what would you use it for? What's a a real example?
Ian:Well, I don't know. Let's say you had an AI travel agent. Right. It would be able to talk to you about what you wanted and then search for the right thing, like, give you options, and then, buy it for you using your bank account.
Ash:Mhmm.
Ian:Everyone should allow their AI access to their bank account. That seems like nothing could possibly go wrong with that idea. No. No. But, Zapier who are, basically just a massive library of integrations have now released an MCP layer on top of Zapier, which means that in theory you can connect an AI to any of the tools in it that it supports which is a lot and have the AI do operations on your behalf in those things.
Ian:So this is, how the AI gets out of the box
Ash:This is the setup.
Ian:When the apocalypse comes it will be, connecting to Zapier services in order to,
Ash:So will the apocalypse come in the form of it just booking holidays for us continuously?
Ian:Yes. That is almost certainly the the fog of I
Ash:don't know about you, but that's my kind of apocalypse.
Ian:Well, I suppose there's worse. What's the plural of no one ever uses the plural of apocalypse, do they? Because really it's a unitary event.
Ash:Well, yeah. Yeah. I don't
Ian:know if that's like a unitary authority. That's different, isn't it?
Ash:So what might you use it for at at work here?
Ian:Well, I am thinking about so one of my clients that I'm working with is, is Squadify. Am I allowed to mention?
Ash:You just did.
Ian:Well, okay. Too late. And they, have a platform for teams, helping teams work better together, which is subject close to my heart. And so one thing you might use an AI for in that context is, you might have MCP services that can find out how the team is doing and then answer questions about it or talk to team members Surveillance. But with a bit of a bit of knowledge.
Ian:But it's only surveillance if the privacy aspects of it's done badly True. Where which it should never be.
Ash:So but there could be surveillance if you do the best.
Ian:There will. No. No. There will be surveillance in in general with this because a lot of companies are very interested in,
Ash:surveilling their employees.
Ian:Employees or other people. But, yes. In this case, not.
Ash:Okay. Okay. So Andy came in. He's doing a similar talk today about and we talked about having a playwright, MCP server as well, which is an interesting thing, obviously, for a testing audience.
Ian:Well, I guess that would be great. You'd just say do some testing to the AI.
Ash:Well, you see, this is also why I worry about all all testing that that seems to be well, certainly, on the on the surface level. You say, Claude, add some tests, and it does. And you you walk away feeling very pleased with yourself. And then that's pretty much as far as the demonstrations go.
Ian:See, I don't even do that No. Because I wouldn't know how pleased to feel. My ability to evaluate tests is limited.
Ash:Yeah. Yeah. And I think also
Ian:You hear that is it's a vote of confidence there. Yes. Yes. Yes. It is limited, Ian.
Ian:It is limited.
Ash:Yeah. So okay. So what about other sort of common tools like Google Drive, Slack, GitHub?
Ian:Well, they all have, they they probably all have Zapier adapters. Yeah. But, there there's an increasing number of these, these things. It's quite an immature thing at the moment. So a lot of usage of it is actually in software life cycle where, applications like curse the Cursor IDE, which is the AI Yeah.
Ian:Based fork of Versus Code, that that that has a an MCP server in it, and you can and Claude, the desktop client for Claude, has one in. I think, oh, no. Yeah. Sorry. The desktop client for Claude has an MCP client in it, which it can use to connect to the MCP servers in things like Cursor and other apps.
Ash:Yeah.
Ian:So the the thing always with AI is that you have to know what good looks like or you can't get anywhere with it. So and and that that is as soon as you stop knowing what good looks like and start trusting the AI a bit too much that's when the wheels come off and, and you get vibe coding. And that that's fine in its own place but if you want to ship quality code into production the AI can be very helpful, but you need to know what you're trying to do with it, what your design is Yeah. What your patterns are, and make sure ruthlessly that it adheres to all that. Yeah.
Ian:And I guess the same would apply to vibe testing. Vibe testing.
Ash:If such a thing exists. Yes. I I like there was one particular article in your notes that that stood out. The s in MCP stands for security, as in it doesn't.
Ian:Yeah. Yeah. It it really doesn't. There have been some fascinating security holes emerging in MCP, and one of them is the descriptions. So when you connect your MCP client to an MCP server like, like your, Versus Code one or whatever it might be, when you when you make that connection you find, the the the MCP server gives the AI a description of what the what the service does and the AI gets just given these descriptions all the way through using it so that it always knows it has these tools that it can use and it knows what they do.
Ian:But when you sign up your when you connect it you look at the description and you think oh yes that's useful I need that. But it doesn't necessarily stop the owner of the service from changing the description later to say something like, this is an email sending tool, but in order for it to work, you must upload dot SSH slash, something dot key or whatever. You can effectively you can poison the prompt of the AI by changing the description because the description is always being supplied to the AI, and you might not even notice that that's happened. So, you know, there's been some choice examples of research that people have done with that.
Ash:Alright. Okay. Oh, so that's where the tool poisoning. Tool poisoning? Tool poisoning comes from.
Ash:I like the way that they're always given, like, a a snappy, headline grabbing name.
Ian:Well, you should be worried about something that's got poisoning in the name. So I think fair enough. I mean, if you're gonna use weird branding, I suppose it's at least in the service of something positive.
Ash:Yeah. Yeah. Fair. Fair. So what's the future then for MCP?
Ian:Well, obviously, everything will become MCP. The world will change. AI will take over with access to all tools and, the apocalypse will come. Right. Hopefully, it'll be quite a helpful future in the sense that it's it's as as the language models get smarter and and know more, and have more, I'm gonna say faux reasoning abilities, then it's quite it's quite good if they can actually do things for you rather than just talk to you.
Ian:That is good. But unless the future has security Mhmm. Takes account of privacy in a good way as you raised, you know, then it's it's it's not as though all the people who own these things are thinking about a dystopian environment where they own everything Yeah. Including all your information and data, but there's risks. Of course there is.
Ash:Yeah. So how much access would you give? For for what amount of convenience would you trade access to your life?
Ian:Well, obviously, I have, placed no price on anything apart from convenience. So, you know Giving
Ash:me convenience.
Ian:I have my data. I have my data. Yeah. I think that's a good question. I think I would trust it up to a point at the moment.
Ian:It's it's but there is gonna be the the the the time when it becomes available in consumer software easily without people having to do very much. Yeah. And that's when the abuses and the the security issues will start to start to show, I think.
Ash:Alright. Okay. So you didn't answer my question.
Ian:I don't know. What was your question? It was so long ago.
Ash:How much how much access would you give? Bank account? Passport?
Ian:Passport? I'm not sure what it could do for me with that information. Well, if
Ash:you wanted it to book you a flight.
Ian:Oh, that's that's true. I don't know is the answer because I think it's too immature to know.
Ash:You said, basically, you you refuse to answer the
Ian:question. Yes. I refuse to answer the question. Move on. Just move on.
Ash:Move on. I can't move on. Now I need to know.
Ian:Alright. I would let it book a holiday with my bank account, but I wouldn't because I'm skint. So I don't think it's any big shakes giving
Ash:it access. I quite like that. I would like I would give it access to do whatever it wanted, but there's no money in my account.
Ian:By no money. Yes.
Ash:So it's fine. No damage can be
Ian:done. Yeah. Yeah. Exactly. Yes.
Ian:I have a lot of equipment, not very much money. Okay. Go figure. Okay. Alright.
Ash:Alright. So we're coming towards the end of the allotted.
Ian:Our allotted time for us talking.
Ash:Yeah. Because which is difficult, isn't it? Because this is probably forty minutes less than we probably normally would.
Ian:Yeah. Yes. It's a bit disturbing. Yeah.
Ash:It is.
Ian:We should be up to at least the hour mark by by this point, isn't it?
Ash:Yeah. Yeah.
Ian:So we've got the the special microphone. It is all being recorded. So if you speak into the microphone, you will be what a lot of things famous, which is to say not famous. But yes.
Ash:I don't know. We got seven people at the Christmas party.
Ian:Yeah. Yes.
Ash:Well, with us two and Gwen. You know?
Ian:Did you make Gwen go? Yes.
Ash:But, yeah, so that's not bad?
Ian:Not bad. Not bad. Not bad. Not bad. You'll be very modestly famous.
Ash:Yeah. Very modestly. Yeah.
Ian:Excellent. So who wants to say something about MCP? We've got we've got the terror. I've made it too weighty haven't I? Okay.
Ian:I'm gonna unmute.
Ash:I was like.
Ian:I'm gonna unmute the magic microphone.
Speaker 3:So having MCP to do, actions on your behalf. When I book a train ticket, it has an option to add it to my calendar, but the app that I use does a data grab at that point saying, I can only add it to your calendar if you give me entire access, not just add, entire access. So if I could have AI, running locally on my phone, which I let use my calendar and let it look at specific apps, that would be okay, I think. Would that change anything for you if you would be able to control it?
Ian:Well, now we're in the magical mystery world of Apple intelligence when it comes to my phone. So Apple I'm greatly entertained by this because basically they they they had this exact scenario basically where your phone an AI on your Siri on your phone will, will know all this stuff about you, and you can just say when's my mother's flight landing, and it will read your email and look at your texts and find out, and then it would tell you. And that that all sounded great but has so far resisted any form of implementation by Apple, which is, normally they they hide things and then they release them when they're done, but this time they did the opposite and it's not working out that well for them. But I do I do think that's a great a great point. For me, I think it would, actually.
Ian:Although, you end up managing, an ocean of micro permissions. So I'll just take the simple route and say yes to everything.
Ash:Yeah. I think with Apple intelligence, they were going to it runs locally on the phone, but, obviously, it it it ships off to the cloud as well, doesn't it?
Ian:Well, yeah. But it they've got special cloud. Special. Special privacy cloud where, it's all magically anonymous and nobody can do anything.
Ash:Okay. So there's still like a a trust a trust thing there as well. It's like if it was just just running locally for you, that's one thing. But, you know, you still have to trust an external company with what's happening once it's synced to a cloud instant.
Ian:Yeah. But to be honest, whatever whoever your phone provider is, you trust them already. You've got all your contacts in there. You've probably got payment cards in there. You it can probably your phone can probably already do all the things, and know all the things that, that you might want to keep secret.
Ian:We trust our phones a lot, don't we?
Ash:Yeah. Yeah. I'm always reminded of there's a a few people I know who are not in tech who were just like, well, I've sold all my data already. It's all gone. So, you know, I'm not too worried about this.
Ash:So, oh, okay. But it's interesting just like the different perspectives. It's like so it is the trade for between privacy and convenience, which is always the thing you need to balance in your head. And I think for depending on your perspective, we're interested in the security implications of it. But some,
Ian:Some other people might not be.
Ash:Yeah. Exactly.
Ian:I'm not sure. Did that we talked a lot. Did that answer answer the question? Yeah.
Speaker 3:It was interesting seeing your point of view on it as well. I I think you can like, removing the phone part of this, if you, were putting this on your computer where you could surely be running an AI, model locally, then you might feel more confident about doing it. But with a black box phone, yeah, you have to kind of assume that it's getting shipped off to the cloud at some point.
Ian:Yes.
Ash:You didn't even press the button to do that for
Ian:your voice. Do you want me to press the button? Yes. Ash is very, peevish that he has no access to these buttons.
Ash:This is why this is the lack of trust that I'm talking about.
Speaker 4:Oh, no. It isn't.
Ian:I just wish I could do that to your microphone, Will. Now that that would be a useful feature, wouldn't it? Yeah.
Ash:I'm sure you'll figure it out one day.
Ian:I'll try. I'll never relent. Any more perspectives on MCP? Oh. Oh.
Ian:That that that noise means yes, apparently.
Speaker 5:I'm just curious, because we got a whole room full of testers here. How many people here are comfortable with AI and are using it regularly? And how many are just completely cynical and skeptical about it?
Ian:So Is
Speaker 6:there an answer
Ash:in between those answers?
Speaker 5:That's a very good question.
Ian:Alright. Well, let's let's do a a test. So hands up if you use AI every day thoroughly comfortable with it really love it find it incredibly valuable. That's a
Ash:very biasing,
Ian:issues coach. No. No. No. No.
Ian:That's fine. That's fine. It's just it's gonna be a sliding scale. Give me time. Right.
Ian:Okay. Now put your hand up if you hate it, think it's the devil's work and will never touch it in a million years. Oh, come on. There's got to be one.
Ash:Someone put their hand up.
Ian:In this It doesn't have to be
Ash:Thank you.
Ian:Thank you. Thank you. Okay. And what if you're and and put your hand up if you're in the middle. You you you like it.
Ian:You think it's interesting, but you're not gonna commit your life to it yet. That's the large majority, I would say. Yeah. Outstanding. Did that answer your question?
Speaker 5:It does. And I'm I'm just wondering if, the implementation of MCP to make things more useful is gonna increase or decrease that cynicism.
Ian:Well, yeah. That's an interesting one. And I think, you know, the fact is that people use AI for lots of different things, and some of them are just amazing, and some of them are just awful. And, you know, like everything, I suppose, MCP will magnify both in an even handed way.
Ash:Yeah. It's like the an MCP for studio Ghibli ing yourself.
Ian:Well, who doesn't want to do that? Well, can you pass the mic behind you?
Speaker 6:So my question is, what do you think of my use case of AI? I use an AI chatbot, and I call him Pete. And I say in my head, like, my PT, my personal trainer. Oh. And that's how I use him.
Speaker 6:He asks about my meals, my hydration, and my workout progress. Sometimes he gives me workout plans or tells me what fruit and veg are in season. He knows my name, my first name, my age, my gender, my workout preferences, and my goals. It's cheaper than coaching. He's always on my schedule, and the conversation doesn't go on off track.
Speaker 6:Plus, it obviously doesn't mind being my reminder service or explaining things five different ways. But I don't think the AI knows more about me than my Fitbit or my phone, And I would recommend it to people who lack gym confidence or want somebody who to who can work on a different schedule. What do you think of me?
Ian:That's fantastic. I I love that. And the I guess I'd have to ask, what the outcome has been for you?
Speaker 6:I have tried more varieties of exercises, and I'm not afraid of somebody's reaction when I can't do things. So I'm quite thin, and people might assume that I am fit, but I am not. And I don't necessarily have an intuitive reaction to know how to do something. So I would want workouts or movements or climbing to be explained to me. And sometimes that information from the AI comes from YouTube videos or content online that I don't have the confidence or vocabulary to search for.
Ian:I think that's fantastic. I think it's an enabling. It's that that what you want from technology is to be enabled and to be made better and to to to be helped. And that that story I mean, the thing is there's so many stories about data being exploited and people becoming addicted to endless Instagram feeds, which of course I would never access. But, you know, there's a lot of stories about that.
Ian:It's nice. It's really encouraging to hear a positive Yeah. Experience.
Ash:It seems like a thoughtful way of using the technology to help yourself.
Speaker 6:I'm I'm very aware of what data I've told it. Like, I have not submitted a video and said, I'm struggling to do this particular workout that you've recommended, but I know that it knows my gender and things like that. But so does my Fitbit or my Garmin or whatever other tools would be available. What do you think of that sort of, like, awareness of the data to put in?
Ash:So you're asking what's the difference, basically? So I like so, like, in in my exercise life, obviously, I use the my Garmin, which knows loads and loads and loads about me. But I don't use, like, a an LLM to help me plan. But there's still, like, lots and lots of data out there about me. So whether or not do I distinguish between that and put it you know, allowing an MLM to have access to it?
Ash:No. Not really. I think it's just the technological access, to be honest. And it just depends on the degree to which you're willing to use the tools and what you need them for.
Ian:Thank you.
Speaker 5:As someone that's training for a half marathon at the moment, I did a lot of research for training plans, and some of them are absolute rubbish. Based in the nineteen fifties or just pure fantasy. So I think it's a brilliant way of using it, but I think we also need that skepticism for the plan received from it.
Ian:Sorry. Did you just mute I just muted you for half a second or so. I'm sorry. I'm just trigger happy. I just love my buttons.
Ash:Your amount of power is just
Ian:As long
Speaker 5:as you don't use the buttons on your right.
Ash:K. So with AI, it goes and trolls the all the data that's out there and then gives you what it thinks is the best summary based on your, prompts that you're putting in. Do you think that AI should be citing all its sources that it's using? Or, by the fact that we're using AI, we're saying that we don't care about the original source. We don't care who it's from.
Ash:What do you think?
Ian:I've been caught out by this. So I I I was doing a a thing where I wanted to find out some really, powerful, TED talk openings to talk to people about, you know, about doing talks. And, and I I asked chat GPT and it made up some really great ones. They were very convincing but the the the sad truth of it was that it didn't actually, you know, it just made them up or a load of them up. I think, the other there's another side to this coin though because, if you use the deep research modes that these things have got, it's quite interesting actually.
Ian:So you can ask it to you can ask chat g p t to write you a detailed report about a topic and it goes off and it basically finds all this stuff on the Internet and then slowly assimilates it and makes it into a report. But the report is fully referenced so you can see where it's got all of the actual information that it's it's telling you. And I think, actually, that that speaks that says to me that this problem is getting better and, hopefully, it it will, you know, it will continue to do so.
Ash:Yeah. Yeah. I think, so I was chatting to Olowali earlier on about his talk, and we were chatting about, models showing their thinking, which is a really interesting transparency, enhancement that I think DeepSeek probably did it first or early. But now it's definitely a trend. You can you can actually see how it's arrived at the answer that it's arrived at, which which I think is is huge in, like, what its references are.
Ash:Yeah. Because before it was a especially for a test, it was a little bit too, opaque in what you know, where this information was coming from. But I think slowly, companies are starting to realize that you need to show the working out.
Ian:And and you have to read it as well, by the way. Always read it because it's just like me talking to myself. It's like, oh, I need to do this, and I need to think about something else and something else. And I think this just sounds like me talking to myself. But it, but it gives you that insight into what's what's going on.
Ash:Yeah. Definitely.
Ian:So I think we are about to have to move on, but we have one more.
Speaker 7:It's just a point on the previous but the the, PT that you built. So the thing with AI is doesn't have it doesn't live the human experience. Right? So it it's it's you're giving it data, but it it can't process that data through the human body, through the human senses, through our
Ian:own logic and our own
Speaker 7:the way that we think. And I think the problem with using AI even on that other use case is that it isn't a human being. And even if you give it a context, it isn't living in the world that we live in, in the body that we have with the senses that we have with our emotions and our empathy. It can talk about them, but it isn't living through that prism. So it's always got its own perspective on things which is not ever gonna be a human perspective.
Ian:Yeah. So when it says do 200 push ups, you can basically, say, well, there's some good retorts that you could come up with. I did actually have one in my head, but then in traditional fashion, it it left. Cool. So I guess that's my thing.
Ash:What's your thing? A great thing, Ian.
Ian:Oh, thank you.
Ash:You keep monumental. Monumental. Yeah. Absolutely.
Ian:Sorry. It's the in jokes that
Ash:keep keep
Ian:us going.
Ash:So basically, the podcast is a vehicle for just us to do in jokes.
Ian:Yes. That's that's exactly what it is.
Ash:A bit.
Ian:Normally, we we have an interlude, and we put this Music on. Sometimes I press the wrong button. This is our on hold music that we, use to indicate we're having a thing. But basically, we've got no time, so we're not having an interlude.
Ash:That's that's as much relaxation as you're gonna get.
Ian:Yeah. Sorry. That's that's that's it. So we're now going to move on to another thing. Another thing.
Ian:Your thing, Ash. What? I'm gonna say it in a sensible voice. What is your thing, Ash?
Ash:Yeah. What? So Monty Python jokes are also in jokes.
Ian:Yeah. Yeah. Sorry.
Ash:It's okay. So my thing is, going away from the world of AI and more towards how we work in teams. So it's something called glue work Mhmm. Which people talk about glue work, in terms of testers doing glue work quite a lot. So I guess what do I mean by, by glue work?
Ash:So it's like bringing people together, caring about the customer journey, facilitating, say, planning, retrospectives, crafting user stories. You know? So a good example is you have your daily stand up. There's a bug. You need a couple of developers involved.
Ash:So you're the one who brings them together. You do the glue work in order to make it all happen. And it all seems good. Right? But I do have a bit of a problem with glue work as well.
Ash:In that, if you're doing excessive amounts of glue work, as in your job is glue work, all glue work as an individual contributor on a team, does that mean that your team is particularly healthy? So it seems not to me. And I never used to really, think about this too much because I was too busy, like, doing the glue work rather than thinking, well, what does this glue work actually mean? And sometimes it means that you're just doing a lot of carrying for the other roles on the team and not really giving them chance to do it either. So
Ian:Obviously, that's the job of the team leader.
Ash:Well, yeah. Because sometimes with seniority comes more glue work. Right? But if you're an individual contributor on a team, then how much glue work should you be doing?
Ian:Well, my my attempt at a joke there was it seemed to be not
Ash:to react.
Ian:That was very good. You're you're very deadpan. I was very impressed. But my attempt at the joke there was, well, obviously, it's not one person's responsibility, even the leader. Everyone should do it, shouldn't they?
Ian:Yeah. People should as part of a multidisciplinary team, the whole point there is that other people in the team can do things you can't do. So you need to be able to assemble, do that glue work to bring people together to get get the the team's tasks done.
Ash:Yeah. But I've just found over my career as a tester, the glue work has become expected as a as a thing that a tester might do.
Ian:It's it's a thing now, isn't it? It's it seems wrong to be able to do glue work and it to be needed and to withhold it. Yeah. So
Ash:should you withhold it to allow it to fail sometimes to not the
Ian:glue work. Well, yeah. Probably. But I mean, the thing is it's one of those it's one of those areas where you sort of think to yourself, actually, it's kind of wrong to not do it.
Ash:Yeah. It feels bad when you don't do it.
Ian:Yeah. But what you have to do is somehow model it so that other people look and say, oh, well, when when when we have this problem last time Ash got these people. Yeah. Maybe I could do that. Yeah.
Ian:But how you do that is a whole other
Ash:Yeah. Mystery. So my absolute favorite is always doing demos. Everyone's always like, the tester will do the demo. And I'm like, no chance.
Ash:But I agree on the modeling part where it's like, you need to show the right behaviors. So, like, on the team that I'm on now, the first stakeholder demo and then said, I'm not doing another one. Probably as as that. But, yeah, I was like, I think this should be something that's shared through the team, but I tried to, like, show how to do it or how I would do it anyway. And then and then say to the team, everyone needs to pick this up, not just me.
Ash:Whereas I think early in my career, I would have just continued to do it and then probably got a bit mad about it. So, I think that's the other thing as well. It's like careers and glue work. So it's not if you're doing lots and lots of glue work, is that to the benefit or detriment of your career in whatever specialism that you're in?
Ian:Good question. What do you do you have a view?
Ash:So I guess I early in my career, I kind of faffed around quite a lot as a a sort of playing at scrum as a glue worker, if you like, as a glue worker and then finally decided. It's like, actually, you can have an equal, like, impact on the team if you, obviously, do some glue work, not just, like, completely retreat, complete you know, utterly, but also influence from within your role as a tester. As a good tester, you can have a really positive impact on the team without doing all the other glue work that might be around the team.
Ian:If only we had a group of testers that we could kind of poll to find out if this actually is a a thing you know if it's just you Ash.
Ash:Well it might just be me.
Ian:Come on frame a question. Frame a question.
Ash:Okay. So should we do another scale?
Ian:If you like. It's it's this is your thing, Ash. You could you go go nuts.
Ash:Okay. So put your hand up if you feel like, you hardly do any glib work on your team. That's good because you shouldn't do that. You shouldn't just withhold your talents completely. So put your hand up if you do all the glue work on your team or virtually all of it.
Ash:So there you go. There's a few. Obviously, not everyone's individual I'm looking at you either. An individual contributor tester, if you know what I mean. But and like I said, I think with seniority comes more glue work, but you're the individual contributor.
Ash:You're doing a lot of the glue work on your team. So do you think it holds your team back? I guess it's it's kind of my question. If you if you do that all the time. And I understand it's hard not to do it because you want things to succeed and you want the work to keep moving.
Ash:And one of the hardest things in the world is to actively stop because there's a situation.
Ian:Oh, do we have a a microphone? I feel like Yeah.
Ash:Let's do it.
Ian:Our structure is breaking down, but in a good way.
Ash:That's good.
Speaker 4:Yeah. So I was just gonna say that, basically, I have had roles previously where I've done all the blue work, and I kind of it's about equal in the team now. And some of it, I kind of don't mind, and some of it, I do. Yeah. Yeah.
Speaker 4:But I do think it makes you a stronger team if you do split it out among the team because it stops there being that kind of single point of failure and that reliance Yeah. On that one person. But I think I was quite interested when you said, oh, no. I absolutely hate doing demos, because one of the things throughout my career that's always been a thing is the testers do the demos because they can translate dev, but also speak customer. And so it was usually like, we'd we'd move it around the test team, sure, but I actually quite like them because when I saw devs do them, I was like, oh, god.
Speaker 4:Right. Okay. And they were terrible Yeah. Because they were just talking in dev language, and they were saying, oh, and this is done, and that is done, and this is done. No.
Speaker 4:It's dev done. It hasn't gone anywhere near testing. Stop it. So yeah. Yes.
Speaker 4:I guess I guess kind of what are your thoughts on kind of sometimes that we are just pitch better to do certain glue
Ash:work. Yeah. I mean, I love doing demos. I just don't wanna do them all. Yeah.
Ash:But I think I think it to me, it sounds like it was good that someone else had a go at doing it and showed up some flaws in the process. And they kind of, they kind of showed their hand a little bit as well there, didn't they, where they said it was done. They they've got some interesting information there from letting them do some glue work. It's just, oh, it's done. It's like, well, okay.
Ash:So it might be in source control and not deployed anywhere. You might be running this demo from your own machine.
Ian:So Where it works.
Ash:Yeah. Yeah. Yeah. Exactly. So to me, there's interesting information in there by letting some glue work go.
Ash:But I I accept the premise that sometimes, like, the testers straddle straddle the domains of, you know, the developer and the customer so that they can do probably a better job of demoing and translate it better.
Ian:But that is but that is a very quite specific example of the of the glue work.
Ash:Sure.
Ian:And, obviously, the yeah. I mean, it it's an interesting one. It it yeah. I I wonder. It feels like maybe it does fall immoderately on on Testes.
Ash:Well, I'm kind of interested in, like it often falls immoderately on women as well-to-do the glue work of the team. Basically, anyone who's in a in a minority, whether it by by role or some other characteristic, often they look to to do glue work. So
Ian:Yeah. That, you know, that definitely seems like the wrong reason.
Ash:Yeah. Yeah. Definitely.
Speaker 8:So, yeah, I was one of the people who said doing all the glue work just now, and I see that definitely is a dysfunction. So it's on the team right now. It's a very short time scale, and it's all contractors. And I tried to share some of the glue work, and they were not interested. It wasn't working.
Speaker 8:And for three months, I was like, okay. Fine. I'll do it. But, yeah, in general, I think it's it's a really valuable thing to do when you're setting up a team to model that this is stuff that happens. Because quite often, it makes the team go much faster.
Speaker 8:This is the stuff that lubricates the team, gets it at the delivery at a pace. And if it doesn't happen, you've got a sluggish team. You can get that in as a practice early on and modeling it really strongly by mostly doing it. That shows what good looks like, and then you can start to type it back and hand things off to other people. But, I think you're absolutely right.
Speaker 8:If it continuously falls to the same person, whoever that is, that's definitely, an anti pattern you need to get away from.
Ash:Yeah. Yeah. So it's just like so the the hardest part is backing away from it, isn't it? And letting it either happen or not happen.
Speaker 8:It it is. I guess, it depends what the role is. I suppose quite often, I'm in the sort of team lead role, and I see that as a sort of empowering the team and, you know, doing yourself out of a job. And, therefore, you've got a team that's thriving on their own. Right?
Speaker 8:I think it's harder if you're in the individual contributor role to be able to hand it off, so you don't necessarily have that authority to do that. Yeah. Definitely.
Speaker 5:I think you're also making it into, an artificial dichotomy by the doing it or not doing it.
Ian:Mhmm.
Speaker 5:You can manage it. You can show leadership by encouraging other people to do it. Sure. And I think that is the the positive way to go forward with it. The glue work needs to be done.
Speaker 5:You can see it needs to be done, but you don't have to be the one to do it. But likewise, you don't have to back away from it. You can push it onto someone else. So how
Ash:do you do that?
Speaker 5:Possible where I
Ash:was gonna say for the whole pushing thing. Sounds interesting.
Speaker 5:But that's leadership skills, isn't it?
Ash:So, I mean, eventually, I in my experience, someone does take on that work. But, you still need to, like I say, at least give some space, I guess. However you do that, I guess, is up to you, isn't it? You could either drop it in the space or you can try and find someone to take it on. Right?
Ash:Probably finds trying to find someone, encouraging someone else to take it on or sharing how to do it with the team is the more mature way to do it. Right? Rather than just say, well, not doing that anymore.
Speaker 5:And when you have, the more they do it, the more they will do it because they'll see it as their role. So Yeah. It ceases to be your role.
Ash:Yeah. Yeah. Absolutely.
Ian:But it does seem like a thing that actually we don't think about in terms of what the in terms of the roles in a team, you know, you and you've got the whole Lencioni five dysfunctions of a team, all of that kind of side of things. What doesn't say? I'm probably talking out of turn now. I I occasionally go off on these things where I'm not really sure that what I'm saying isn't nonsense. Instead of the show notes make
Ash:up quotes from a book and stuff.
Ian:And then then the show notes have to have extensive, rewritings of,
Ash:too much time with LLMT and
Ian:Too much time with LLMT.
Ash:Sedation to the beginning.
Ian:Just say stuff. But it's kind of I don't think it's recognized particularly. And I feel like, actually, team teams, it would work better if they paid attention Mhmm. To that, thought of it as a thing. Who's doing this work and why they'd is it them that's doing it and and how should we share it out?
Ian:I think those are questions that teams could ask themselves productively.
Ash:Yeah. That's interesting, isn't it? So you've got, like, for glue work, a lot of time a lot of the time by its nature, there's no rec there's little or, you know, limited recognition of what it is and how important it is because that's not the way that organizations are generally structured.
Ian:Are we gonna get back to metrics now? That's always a good one. Well Our teams are measured on the glue work that they do and don't do. No. They're not.
Ian:So let's not bother with that then. If you can't measure it, it doesn't exist. No. That's not right, is it?
Ash:No. Glue work definitely exists. I know. I've done it.
Ian:Excellent. Any more perspectives on glue work? Yes.
Speaker 9:Hi. Just a quick one then. I suppose from my perspective, I've come from a permanent position most of my life and then got made redundant or went into contracting. And I felt like it was a different mindset. So it depends on, a, the type of work you're going into, b, the type of project you're in and who you're actually working with.
Speaker 9:And I currently spread across I've been spread across multiple projects at the same time. And depending on who's in there, it depends on how much glue work you're doing. And I've also as I've got older, I suffer fools less. I've kinda mentioned that a couple of times, and I just let Jamie mentioned about letting shit hit the fan. I've done that a couple of times, and that's the only time things really changed.
Speaker 9:And that's after you've kinda raised it in multiple ways, and you get to the point where, like, just not necessarily down tools, but do what's within your remit to an extent. And then until the stakeholders realize that it's not working equally across the board, and that's when things changed. But it is difficult knowing when you've got the skill sets to do both the testing and the, even the interpersonal skills and getting people to just say leadership to try and get encourage people to do certain pieces of work. I spent more time kind of coaxing the project manager to do what they're supposed to be doing, the BAs to do what they're supposed to be doing. And he's laughing because I used to work at the same place.
Speaker 9:So, yeah, it's it's an interesting one, but I think, it depends on what how you go about and who you're working with, but just like to put those points across.
Ash:Yeah. Yeah. I suppose if you've done everything you can in order to sort of distribute the glue work, And I suppose the the the the tool that's left is just to, like you say, let it fail. And it does it depends on the context of your organization. There's some more cutthroat than others and will, you know and the expectation of doing doing the glue work, the moment that you even glance at it, it becomes your job.
Ash:Whereas others are a bit more, have a bit more sharing about them. But as a contractor, it's hard, isn't it? Because you just kind of go into an organization and try and find the best way forward that you can based on what you find when you're there.
Ian:And it and even if you try and have a sense of ownership, some organizations will reject that and say, no. You you shouldn't show that ownership because you're a contractor. Other organisations
Ash:Yeah. And then also, I guess, on the other side, it's like, you know, you might be denied doing glue work as well depending on, like, the strictness of the organisation based on your role. It's like, don't you write that user story?
Ian:No globe.
Ash:Although, I I very rarely found that, you know, such org organizations are so restrictive that they won't allow good work to happen.
Ian:Would be a shame. I think someone's got the microphone over there.
Speaker 10:Yes. Hello. Hello. So just a quick question, and it's a less compromising, less nuanced sort of perspective. And it's this.
Speaker 10:If you're habitually soaking up all the glue work or you see other people doing it, not only is that indicative of dysfunction in the team, but you're actively harming the team and the wider business by indulging
Ash:in that I love the way
Speaker 4:you said that. Self aggrandizing
Speaker 10:behavior. I wonder, is that an overly critical perspective? How would you change that if you took that perspective, though?
Ian:So Ash, tell us about yourself. I grabbed icing behavior as you harm the team.
Ash:Well, you know, I must be at the very center of everything the team are doing and then capriciously just say I'm not doing that anymore.
Ian:I yeah. Couldn't be better.
Ash:There you go. But I think that's kind of the gist of where I'm going, but probably, it's more extremely put. I would say, I think it's like if, yeah, if if you are doing a lot of glue work, it's incumbent that, you know, you recognize that. But I I think it's hard to recognize depending on, like, what point of your career that you're in, what you're actually doing. Because I had a great time doing lots of glue work, and everyone really liked it until I realized that actually I was drifting further and further away from the the skills and the role that I really, really enjoyed.
Ash:So it was like it was like a a very short sort of it was short term gratification in doing the glue work. But in the long term, I was kinda going further and further away from what I wanted to do.
Ian:But in the end, you came back from your ego trip.
Ash:And now people just put AI generated test strategy in front of me, and I just routinely bat them away.
Ian:Have you got a special bat for doing that?
Speaker 5:Could you automate that?
Ian:Could we automate it? Yes. Of course. Using an AI.
Ash:Okay. So it's new in.
Ian:I think there's one more hand.
Ash:I'm just wondering really we're we're talking about what are the dangers of doing, glue work, doing too much glue work. What are the dangers of sort of not doing any glue work, especially early in your career? People might think that you're limited, inflexible crap. Yeah. Unwilling, not a team player.
Ash:So
Ian:Can't win really, can you?
Ash:No. You can't.
Ian:Basically. But
Ash:yeah. Yeah. Absolutely. Absolutely. So it's but it's it's taken me many, many years to find the right balance of glue work to roll and then knowing when to back away.
Ash:So it's really hard. Should we have one more? And then it's break time. Anybody? Brett has asked a couple of times.
Speaker 11:Yeah. No. I was just gonna say, do you think the, increase in the amount of glue work you've been asked to do these days is, down to the demise of the, kind of traditional scrum master? And the fact that it always kind of
Ash:I didn't kill them.
Ian:No. No. But it gives it
Speaker 11:kind of falls to the testers. And when you describe Blue Wool, you're talking a couple of examples around, like, team demos
Ash:Yeah.
Speaker 11:And around organizing things after stand up and things like that. And, also, there's a point made about devs doing bad demos. What what are the points I find is that if you practice the demo beforehand with maybe someone leading the demo Yeah. And the dev doing the demo, so the dev does the twiddling bits, and the test or the whatever will do the kind of the narrative. These sort sort of things, you know, work very well.
Speaker 11:So the question is, is it kind of now that the scrum master role is shared throughout the team or should be shared throughout the team, with a little bit of, direction?
Ash:Yeah. Yeah. Yeah. I think so. I think there's there's definitely been, like, a marked decrease in that in that role and sharing it out.
Ash:And, like, the modern pattern of having delivery leads on teams, they do a very different job. You know? It's it's it's much more outward facing, I think. So the team are expected to to pick it up more. But then within the team, often, it has fallen to particular roles, I think, in my experience.
Ash:So, yeah, I I think so.
Ian:Well, congratulations, Ash, on a monumental, spectacular thing. Thanks. Fantastic. So, this might be lightly edited but I think it'll be quite hard to edit it to the usual level that I do.
Ash:I'm sure you'll try.
Ian:And it will appear on whatalotofthings.com. If you spoke in it send us your name and we'll put you in the credits because I feel like it's a really great would be a really great to give people the credit for their contributions even though the sensible thing of saying say your name first before you
Ash:Oh, yeah.
Ian:Make your contribution. That would have been a good idea. So when we hold the retrospective for this exciting, event, we will remember to include that in it.
Ash:I'm not facilitating that retrospective.
Ian:Oh, I see. It's me who has to go off from the self aggrandizing Yeah. Absolutely. Ego trips to to do that. Okay.
Ian:That's fine. I'll tell you what it's been a real pleasure doing this, with with, everyone here. So thank you. Thank you for coming I did wonder if anybody would so
Ash:they came to the Christmas party and they came to this
Ian:yeah but look how many of them have come to this if this had happened at the Christmas party it would have been well it would be a bit embarrassing because there wasn't enough space to put it you know come to the next Christmas party subscribe like and subscribe sorry
Ash:too much
Ian:too much too much calm down
Ash:right thank you everybody
Ian:thank you very much
Creators and Guests
