So, What Are We Building?

A quick introduction

I'm Wes (DaytimeLobster), one of the founders of Firespawn Studios. We're a small software company working out of rural West Virginia, and this is our first blog post.

I've rewritten this thing more times than I'd like to admit. First posts carry a weird amount of pressure. You want to set the right tone, say something worth reading, and not come across like you're performing. So I'm going to try to keep it simple and just tell you what we're about and what you can expect from this space.

Who we are

We started Firespawn in 2023 with a straightforward idea: technology should be useful. Not theoretically useful, not useful-in-a-demo useful, but useful in the way that matters - where it solves a problem someone actually has and keeps solving it after the novelty wears off. We hope to have some exciting announcements to share soon, and we're just getting started.

We're a small team working from a small rural town, by any standards. We think that's an advantage, not a limitation. There's a clarity that comes from working on a project when you're not surrounded by hype cycles and funding pressure. It helps you figure out what actually matters in a piece of work so that every decision and every line of code earns its place.

Our work in AI/ML spans memory and retrieval systems, multimodal search, alignment, agent tooling, and benchmarking, among other things. We also do cloud infrastructure, full-stack development, and digital media. It's a broad surface area for a small team, but it all points the same direction: we build systems that help humans and AI make sense of complex, unstructured information, and we study how AI agents can be made more capable, reliable, and useful. We're big believers in open source - the best tools should be available to everyone, not just teams with big budgets.

What this blog is about

Mostly: the technologies we use, the problems we hit, the trade-offs we made, and why. We'll also share project updates and milestones as things take shape. The goal is to write the kind of posts we wish we could find when we're deep in a problem - specific enough to actually help.

Here's a sample of what we have planned:

AI engineering in practice. How to build data pipelines that don't fall apart at scale. Why your RAG system's retrieval quality matters more than your choice of model. How to design AI applications that aren't locked into a single vendor. How to use one model to check the work of another, because trusting a single model's output without verification is a recipe for shipping garbage.

Autonomous agents and how they learn. Not chatbots - agents that operate in complex environments and make sequences of decisions with incomplete information. Memory architectures, hierarchical planning, spatial reasoning, and the surprisingly difficult problem of teaching an AI to recognize when it's stuck in a loop.

The infrastructure nobody talks about. Container orchestration for real applications, not just examples. Database architecture and critical application design decisions, like when semantic search is probably overkill (despite how cool hybrid semantic search sounds). Deployment strategies that let a small team ship with confidence. Security concerns that shouldn't be an afterthought. All the unsexy work that determines whether your idea actually runs in production or becomes a grand testament to that one time you tried and only got a cloud bill to show for it.

Open source and the tools we wish existed. We like to build in the open when we can. We'll share our thinking on open protocols, developer tooling, and the kind of practical, well-documented software we want to find when we're searching for solutions at midnight.

The business of building. What does it actually cost to run AI systems in production? What are the real trade-offs between cloud providers, model APIs, and infrastructure choices? We think the industry needs more honesty about what this stuff costs to operate.

Why bother

Part of the reason is selfish. Writing forces you to think clearly. When you have to explain a system to someone who wasn't in the room when you designed it, you find the gaps in your own understanding and plans. Every post we write makes us better at what we do.

But the bigger reason is that there's a gap in the conversation. There's a ton of AI content out there, and most of it is either surface-level hype or academic papers that assume you already have a PhD. There's not enough writing from small teams doing serious work - being specific about what they built, how it works, and what they'd do differently.

We want to help fill that gap. We'll be honest about what we don't know, and, hopefully, helpful enough that you walk away (alt+tab away?) with something you can apply.

What's next

We have a backlog of posts in the works - technical deep dives, architecture breakdowns, tutorials, and the occasional opinion piece. We'll be publishing regularly, and we'd genuinely love to hear from you - whether it's feedback, a question, or just to say hey.

Thanks for being here. We're just getting started.