The SUAMI Method

In this episode of the DevOps Sauna, Darren and Pinja discuss the reveal of an exciting new technology created by Eficode: SUAMI (Single Unified AI Metrics Indicator).
[Pinja] (0:02 - 0:11)
At least he's not talking to the other people anymore. So he's not complaining about the workload. He's not overwhelmed anymore.
So yay! TORILLE worked here as well.
[Darren] (0:14 - 0:22)
Welcome to the DevOps Sauna, the podcast where we deep dive into the world of DevOps, platform engineering, security, and more as we explore the future of development.
[Pinja] (0:22 - 0:32)
Join us as we dive into the heart of DevOps, one story at a time. Whether you're a seasoned practitioner or only starting your DevOps journey, we're happy to welcome you into the DevOps Sauna.
[Darren] (0:37 - 0:42)
Welcome back to the DevOps Sauna. I'm once again here with Pinja.
[Pinja] (0:43 - 0:44)
Hello, how are you doing, Darren?
[Darren] (0:44 - 0:52)
I'm doing reasonably well. And I think it's time we talk about a subject we don't really talk about at all on this podcast, which is AI.
[Pinja] (0:52 - 1:09)
Yes, we have not given this topic enough space here, if at all. So let's talk about something that is also close to our heart and something we have developed here at Eficode, which is a Single Unified AI Metrics Indicator, aka the SUAMI method.
[Darren] (1:09 - 1:27)
Yeah, so this is a new code analysis method which uses some esoteric ideas, things like code aura logistics and Bayesian probability distributions generated from code change volumes. We measure things like algorithmic echo resonance and intrinsic aura analysis.
[Pinja] (1:27 - 1:36)
Let's do something human. Let's break it down into human sense here. We have a new model for this, and this is, of course, complete nonsense.
[Darren] (1:37 - 1:37)
Yes, yes.
[Pinja] (1:37 - 2:07)
We all know that when we're in a job that we have people, let's take developers as an example, and they have so much of this solid knowledge. And they leave for another, a better job, usually. There might be multiple reasons for this.
Maybe we didn't pay them enough. Maybe the person was distracted by how many team-building activities they had to do, like an Agile coach coming in and making them floss on stage, for example, at a PI planning. But we run into problems here, don't we, Darren?
[Darren] (2:07 - 2:38)
Yeah, the on-stage flossing is almost universally fatal to a person's career. It mandates a company switch. And you end up with these comments in code saying things like, when I wrote this code, only God and I knew how it worked, and now only God knows.
And you end up with people wasting hundreds of hours trying to optimize routines that they don't understand. And I think this is a universal problem that happens to every company across the entire history of companies throughout all time and all industries.
[Pinja] (2:39 - 3:03)
It is. And the stage flossing as a universal problem that has infested all of the development world and the world of software has now hopefully come to an end with the left developers as a problem, because we now mentioned the SUAMI method. So we're also proud to introduce the new agented AI approach of reviving or returning the departed coders back to your projects.
[Darren] (3:03 - 3:55)
Yes, we have this new innovation called the Technical Overhaul for Reviving Innovative Legacy Language Engineers, or TORILLE, powered by the SUAMI method. And what this does is we can use this to basically, we don't want to use the word resurrection, but let's be fair, Easter is just around the corner. But when your coders depart for whatever reason, build yourself an agentic AI that works in exactly the same way.
Powered by this SUAMI method, we apply the intrinsic aura analysis to prompt engineering workflow. And what this happens is if you have an AI agent that's too robotic, you can actually instill in them, imbue in them the right level of neuroses and consternation and distraction. And of course, the important thing for software development, which is barely concealed rage.
[Pinja] (3:55 - 4:13)
We always miss the introversion of the departed developers and coders, and just the look of dismay on everybody's face when we force the onstage flossing at our event. So here's a solution for this, because we know how much we miss those people's reactions to these things.
[Darren] (4:14 - 4:48)
Yeah. And let me just confirm something. This is, of course, utter nonsense.
[Pinja]
Yes.
[Darren]
Great. The biggest use case I would say is spaghetti code because we all know how coders like to code as if they're creating dinner for 15 Italians.
They will create these codes that only they can read, only they can build. And I'm not saying that these methods can actually untangle; the SUAMI method can't untangle spaghetti code, but imagine applying a nice sauce to it, something like a tomato sauce, something with a little mozzarella, perhaps a little cream, some parsley.
[Pinja] (4:49 - 5:12)
And if we think of monoliths, that's another use case. The departed developers built those monoliths for a reason. That's why they're here.
So, who are we to break them down? And if in case we accidentally did, for whatever reason, the TORILLE AI agents can also rebuild these broken-down monoliths. And what more would we want here?
[Darren] (5:12 - 5:27)
Yeah, of course. Every time we talk about monoliths, I have to bring it back to the pyramids. There's a reason they're still standing.
When someone made these small houses from 8,000 years ago, they're not still around. The one thing that's still around are the pyramids.
[Pinja] (5:28 - 6:15)
Exactly. And I would like to turn our attention to the inverted pyramid as well. How can we make stable enough solutions so that they are as tangible as the inverted pyramid?
And if we think of domino tracking, if something is built on a domino piece, a single piece that can fall down whenever we want, why would we want to take that away? So let's talk about the use case of the third-party dependency domino tracking. So there was this time when the developer that you used to have in your organization used to tell you that the doomsday was approaching, and there's only one piece that domino was holding.
The one component was holding the system from falling apart and falling out of date. We still want that, especially combined with the monolith.
[Darren] (6:16 - 6:42)
Yeah, and this isn't about supporting that package. It's about identification, because it's not about stopping what's coming, more like knowing the single package that's holding up most infrastructure, just so we can make an accurate guess of when it will collapse, because we know it's going to. That's really what we get out of the SUAMI method.
We really like stating the obvious and using AI to do so.
[Pinja] (6:43 - 7:04)
Absolutely. And building upon that, the use case to use TORILLE for the targeted dependency tracking, and if we combine that with the people’s and the developer's externally perceived cognitive load reduction factor, we can actually predict when the domino will fall. Can't we, Darren?
[Darren] (7:04 - 7:24)
Yeah, yeah. We can basically pinpoint to the wheat when your company is going to collapse due to the infrastructure failure. It's a revolutionary breakthrough.
The idea, then, is that you can basically run your business purposes as profitably as possible before the inevitable collapse and then flee when everything does break.
[Pinja] (7:25 - 7:39)
Exactly. And one other good thing about this is that we know how much every single employee in every single organization really enjoys that feeling of inevitably heading towards doom, right?
[Darren] (7:39 - 7:40)
Yeah, uncertainty.
[Pinja] (7:40 - 7:47)
So it's that anxiety that we want, and that's what the studies are also showing, that this is what we want as employees.
[Darren] (7:48 - 7:52)
Yeah, yeah. We all know the saying, a nervous coder is a good coder.
[Pinja] (7:52 - 7:52)
Yeah.
[Darren] (7:53 - 8:48)
And that's what we need to progress with. One of the advantages of agentic AI using the TORILLE system is that if you enjoy making your developers nervous and requiring them to push code that's crap and not ready to meet unrealistic deadlines, this feature is actually your best friend because these developers will only work like eight hours a day, tops 12 if you emotionally blackmail them. But imagine having agentic AI that you can bully 24-7, and the great thing is they are resonant models, so you have continuous learning.
So all this anxiety just gets pushed back directly into your next generation of agents, increasing your speed at the cost of trivial things like sanity. And by the end of our testing methods, we've actually seen that we've been able to use these AI agents in such a way that just reduces them to quivering messes.
[Pinja] (8:49 - 9:15)
Yes. So anxiety-induced developers, according to studies, are the most efficient developers. And if we think of how TORILLE can help in this as well, I guarantee that the SUAMI method and the TORILLE agents, can prevent you from actually making this mistake of fixing the problems that you have in the organization.
And getting those anxiety-ridden developers a better way of working.
[Darren] (9:16 - 9:26)
Yeah, we don't need support because we have AI. The age of psychological safety is over because it's now about abusive artificial intelligence.
[Pinja] (9:26 - 10:01)
Everybody's really happy about that. And if we think about the constant seesaw and the bouncing act of the co-structure anxiety versus time to deployment, and again, the nervous developers are the most productive developers here. And let's give them, the agents can handle the unrealistic timelines, the deadlines, and they can push the code with minimum security.
They can push the code with minimum testing as well. So this is also for the benefit of the reliability of business here.
[Darren] (10:01 - 10:36)
Yeah, move fast, break things. That's how it should be done across all areas of the development phase, not just in startups. And the key motivator here is actually readability.
What we want to avoid in the use of the SUAMI method and TORILLE is readable code. Because if your developers are wasting time reading code instead of writing code, that's a fundamental loss of productivity. That's just a complete loss across the board.
So one of the few areas where code is actually changed by TORILLE is if it's readable. It will obfuscate code naturally just to avoid that loss of developer time.
[Pinja] (10:37 - 10:38)
This is, of course, complete nonsense.
[Darren] (10:38 - 10:46)
Yes, yes. And going forward, let's hear a couple, because we've had a couple of success cases here with the TORILLE method.
[Pinja] (10:47 - 12:13)
Yeah, we just wanted to build some concentration around this, because I know that this is the first time you're hearing this, because this is revolutionary as Darren and I are telling you. So one of our customers who really enjoyed the SUAMI method and TORILLE agents was telling us that their rockstar developer left us after only three years. And they were baffled by their departure, because they had only been there for three years.
And honestly, they hadn't had any holidays or any salary increases, which is the standard at the moment. So they applied the SUAMI-powered TORILLE method, and they applied that into their generative AI approach. Their AI strategy was, indeed, at this time, to not use any AI-powered tools.
They just added the agents, which is, of course, the suggested method on how to implement them. And within only three weeks, so imagine this, like how short this implementation time was and application of this was, the agent was now barking obscenities to Slack. So all that psychological safety, which nobody wants, was gone.
And it took only three weeks here. So it wasn't like the agent didn't show up to any architecture meetings, and neither was it responding to any pull requests. And they just didn't know.
The agent was saying no. So it's like the developer never had left. So it felt very familiar to the whole community in this company.
[Darren] (12:13 - 12:35)
Yep. And we also had a success story where one company's senior principal primary lead super software engineer said it was great. And it didn't matter that this person hadn't coded in a decade and didn't really understand what the AI was doing.
It was working. It was doing whatever was decided was functional. So that was a very quick success story.
[Pinja] (12:36 - 12:37)
Did the organization know what it was doing?
[Darren] (12:38 - 12:40)
Oh, no. Oh, good. No, no, not at all.
[Pinja] (12:40 - 12:41)
Oh, that's brilliant.
[Darren] (12:41 - 12:42)
Yeah, that's exactly how we like it.
[Pinja] (12:43 - 13:47)
Yeah. And one company I really want to talk about, because this is like, we talk about DevOps teams all the time. And what we always recommend is to use that one single person.
And in this company, the DevOps team was also called Jeff. So they had this one person who was the whole DevOps team. And I don't really get it, but he was, for some reason, Jeff was overwhelmed.
And, well, obviously, the company did not hire any additional engineers, because that is the strategy of the company that we build upon. The goodwill of the single developers who are the superstars. So, let's overwhelm them.
And now, because of, well, Jeff has gone from this, gone from the organization, they use the SUAMI method, which is, by the way, patented. So it's available for everybody. So they generated additional AI Jeffs, which is amazing for this company now.
So now he, the Jeff, can talk to them and call them his preciousness. And at least he's not talking to the people anymore. So he's not complaining about the workload.
He's not overwhelmed anymore. So yeah, he thought it worked here as well.
[Darren] (13:48 - 13:54)
And it's important for platform engineering to really abstract the people away from the platform so you don't have to talk to any of them.
[Pinja] (13:54 - 13:57)
Yeah. Communication is what's killing the whole industry at the moment.
[Darren] (13:57 - 14:44)
Yeah, definitely. Yeah. And this kind of siloing helps the transfer of secrets.
You don't leak information if everyone's in a silo. But this, we've talked about this from a code perspective, but I think we probably will have some managers listening who want to know how this affected them. And we do have a use case.
Management of one company wanted to know why developer productivity wasn't up even after they'd hired a productivity manager, a positivity manager, a positive productivity manager, and a productive positivity manager, all to address the positivity and productivity of a three-man coding team. And the application of the Profitability Paradox Engine, which calculates the ROI using randomly generated values, didn't give any answers, but it did print out a really nice chart that told management that they were good boys. Yes, they were.
[Pinja] (14:44 - 15:43)
Good. Management needs to be told that they're being good boys and good girls. They need to be patted on their heads.
And there is an extension. We can also use an additional app with TORILLE agents, where it actually is patting the management first on the head, telling them that they're either being a good girl, a good boy, or a good them, and also patting them on the back and just stretching. Because that's what we need, really.
And if we think about the productivity managers and positivity managers and positive productivity managers and productive positivity managers, it all comes down back to the flossing on stage, right? So just to get that feeling of discomfort and think about starting the meeting by going through the endless number of postcards of telling how we feel and trying to find that perfect postcard that describes our feeling of the day. So I think that's the key here.
And managers, this is really a groundbreaking tool for you as well.
[Darren] (15:43 - 15:54)
Yep, yep. Just even a simple integration with something like Twilio that sends out a text message to every manager just saying good job randomly is key to productivity among management.
[Pinja] (15:55 - 16:11)
And we can also think about this, the TORILLE agent. We have this POC, a proof of concept going on with one of the companies, and it is trying to also replace the departed manager as well that doesn't send any emails back on the good job being done by anybody in the organization.
[Darren] (16:12 - 16:28)
Yep, yep. Basically a system that just drops all attempted communication. Basically it creates, at the C-level, a black hole that nothing passes through.
And those are actually incredibly powerful when applied correctly or incorrectly in business.
[Pinja] (16:28 - 16:51)
That is true. And there is one company and their representative who we talked to about this because they implemented this list last week and they started using the SUAMI-powered AI agents, so the TORILLE agents. So they compared this to the...
It's like ChatGPT on antidepressants and ADHD meds mixed all together. So they said it's five out of seven, so they would use TORILLE again.
[Darren] (16:52 - 17:12)
Yeah. And we can't say much more than that in our customer success stories. If those haven't convinced you that you need this SUAMI method and TORILLE inside your development loop today, I don't know what will.
So ,if these customer cases seem like they could be applied to your development process, contact Eficode's experts to see if TORILLE and the SUAMI method are right for you.
[Pinja] (17:12 - 17:13)
This is, of course, complete nonsense.
[Darren] (17:14 - 17:21)
Yes. We'll now tell you a little bit about who we are.
[Pinja] (17:22 - 17:25)
I'm Pinja Kujala. I specialize in Agile and portfolio management topics at Eficode.
[Darren] (17:25 - 17:28)
And remember, if you like what you hear, please like, rate, and subscribe on your favorite podcast platform. It means the world to us.
[Pinja] (17:28 - 17:30)
Thanks for tuning in. We'll catch you next time.
[Darren] (17:30 - 17:36)
And remember, if you like what you hear, please like, rate, and subscribe on your favorite podcast platform. It means the world to us.
Published: