AI in Software Engineering - AI models have solved coding, but not software engineering, as production speed is crucial [4] - Building AI to accelerate production is challenging due to reliability requirements and the need for multi-agent orchestration [5][6][7] - Resolve AI focuses on using AI to address the complexities of production systems, which involves more than just code [13] Resolve AI's Solution - Resolve AI provides AI site reliability engineer agents to troubleshoot alerts and incidents [11] - Resolve's agents can understand production systems from code to backend databases, offering a faster solution [11] - Customers are using Resolve AI for "vibe debugging," indicating usage beyond incidents and alerts, leading to increased product usage [12] Talent Acquisition - Resolve AI competes with companies like Meta, OpenAI, and Anthropic for AI engineers [14] - Resolve AI attracts top researchers by offering the opportunity to significantly impact the company and change software engineering [16] Future of Software Engineering - Humans will operate at a higher level of abstraction, with agents handling much of the work [17] - Underlying infrastructure and tools will adapt to be more suitable for agents [17]

Today's conversation is with Spiro Zanthos. Spiros is a serial entrepreneur who's achieved two successful exits to Splunk and VMware. Now he and his co-founder May Agaral are on a mission to rethink software engineering tools for an AI first world with their new company Resolve AI.In our conversation, Spiro shares the status of productivity gains from AI and software engineering. How Resolve was born after realizing he spent 15 years in infrastructure and observability software just creating more work for h ...

What does volume debugging look like in my mind. To perform these kind of tasks like help me understand uh what commit has landed in production or is this feature flag enabled, right. An engineer needs understanding of code but also understanding of production and and production is composed of all of these different tools that each has a silo of data but the tools don't really talk to each other, right.And so it falls upon a human to bring their tribal knowledge and also you know knowledge of how to operate ...

Software Development Efficiency - The primary goal is to accelerate the deployment of software to production [1] - The industry observes that deploying a prototype to production often takes longer than expected due to unforeseen issues [1] - A significant improvement would be achieved if deploying to production becomes as simple as creating the initial prototype [1] Agentic AI Impact - Agentic AI and software engineering aim to improve the speed of software delivery [1] Measurement of Success - The key metric for success is the increased speed at which software can be shipped to a broad audience [1]

you know, how do you know if the problem is is the underlying prototype or the application or the EVL itself? >> So, when I run an eval, um every time I do, I look at two things. Um one is what are the uh specific tests or the cases that got worse uh versus my previous eval? And when I look at that, I think, okay, great. Um are these things actually worse? Uh and um if so, let me play with them. And you know, that's great. My EVL found something that's bad. um yay, you know, and now I can improve it. Um or ...

If we zoom, you know, two years into the future, how do you think eval are going to be different particularly for agents and uh what does that mean for brain trust. >> I've been doing evals since 2016. Um and uh one of the one of the reasons I actually was excited about starting Brain Trust is from then all the way through when I was um working at Figma, uh very little had changed about EVELs.And that gives me a lot of internal conviction that eight years from now um eval will still be very relevant and the ...

If you're an enterprise sitting here, how would you answer the question of when should I think about building my own Agentic application internally versus buying. >> I think it mostly just depends on what's core to your business. The more interesting way to look at it is that everyone's agent is actually quite different.>> So, if we're deploying a, you know, customer service agent to like a large airline compared to, you know, like a car rental company compared to like a bank or something, like everyone's g ...

What do you think multi-agent interaction is. I mean, if you take a really really high level view, multi- aent interaction is basically like essential for AGI, right. Um, we're all doing multi- aent interactions now, right. Um, we're all agents, right.So, you so in in in one sense like if you want agents to be close to this level of intelligence, they need to be able to handle multi-agent interactions, right. Um, but to be a go down bit more, you know, concrete, right. I think of it as like the you always h ...

Are coding agents really just creating a bigger problem for production. I think the problem in my opinion with VIP coding is that it isn't going far enough, right. Um, and yeah, you know, you've gotten to this place where you can build code quickly.Um, and then you deploy it, something breaks, but now you don't have that deep understanding of like what you actually built, right. And somebody else has to go and fix that and then that learning doesn't sort of transfer back through, right. Um but maybe one way ...

I'm already using codegen tools like cursor. Can I just extend that to solve my production problems. >> Codegen tools are sort of, you know, designed to operate on the sort of the addressible universe of code, right.Production system is sort of like a living breathing animal, right. It's more than just code, right. It's it's really sort of emergent behavior that comes from like a bunch of these things interacting with each other, right. Like the code, the infrastructure, the deployments, the you know the th ...