About

A two-person AI automation studio

LuminalQ is a two-person AI automation studio. We design and build custom systems that let small teams operate like much larger ones: outbound engines, intelligent assistants, and data workflows that run quietly in the background and do the work that used to eat entire roles.

We are deliberately small. There’s no account manager standing between you and the people building your system; the founders are the team. That keeps us fast, honest about what’s actually worth building, and close enough to the work to stand behind every system we ship.

How LuminalQ works

Jonah handles the client side and Lucca handles the technical side, but we both cross over frequently. Rather than selling a fixed piece of software, we start by finding the one or two bottlenecks costing a company the most money or time, then build a custom system to solve that specific problem. Most of our work falls into two categories.

Revenue-generating systems

Built to put more sales conversations in front of a business, automatically. A good example is a customer reactivation campaign we ran. A lot of companies have thousands of past customers who simply went quiet and were never followed up with. We built an AI “agent” (think of it as a smart, automated texter) that reached out to all of those old customers over SMS, held a real back-and-forth conversation with each person rather than blasting a generic message, answered their questions, and guided the interested ones toward buying again. Because it runs automatically and talks to thousands of people at once, it pulled in several months’ worth of revenue in a couple of weeks. The same approach applies to finding and reaching new prospects, not just reactivating old ones.

Time-saving
systems

Take repetitive manual work off a team’s plate. In practice, that means getting the software tools a company already uses to talk to each other, so information flows between them without anyone copying and pasting. From there it can take a few forms: cleaning and organizing data automatically, answering routine customer questions through a chatbot on the company’s website, or pulling information from across those tools to assemble the recurring reports someone would otherwise build by hand every week. The goal is to remove the quiet, hours-a-week tasks that don’t need a human but eat up everyone’s time.

Every build is custom and tailored to that company’s specific tools and process. Much of our work so far has been in recruiting and staffing, but the underlying capability, turning a repetitive, judgment-light process into a reliable system, travels well beyond any one industry.

What sets us apart

A lot of companies right now are rushing to put AI into everything, building over-engineered systems where it isn’t actually needed. We take the opposite view. In our experience, traditional, well-built software can handle about 90% of what most automation problems require, reliably and cheaply, and AI is really only the right tool for the remaining 10%. The real skill in the modern era is knowing what not to automate. True leverage comes from a human holding the steering wheel, designing the operational structure, and letting AI handle the mechanical grunt work underneath. So rather than selling “AI” as a buzzword, we use it surgically where it genuinely adds value and lean on proven, dependable software for everything else. That keeps our systems simpler, cheaper to run, and far less likely to break.

The same discipline shows up in how we build. Most AI automation is a black box: something goes in, something comes out, and when the output is wrong, nobody can tell you why. We build the opposite of that. Our approach is grounded in the Interpretable Context Methodology (ICM), an established approach to structuring the context an AI model works from so that every system stays legible and auditable rather than opaque. In practice, that means we keep two things cleanly separated: the stable rules that define how a system should behave (voice, constraints, conventions, the things that should never drift), and the specific, changing input each task actually operates on. Holding those two apart is what keeps an AI system consistent on the hundredth run and the ten-thousandth.

The payoff is what the methodology calls a glass-box workflow: every step a system takes leaves behind a readable artifact you can inspect. When something needs to change, we fix the source rule, not the one-off output. Every future run gets better, not just the one in front of us. For a client, that means systems you can actually trust, understand, and grow into, rather than fragile automations that break the moment conditions shift.

It’s also where the name comes from. LuminalQ is short for luminal question: a light-bearing question. A good question, framed with the right context, is what brings the unseen part of a problem into the light, and that’s where every system we build begins.

The founders

Jonah
Carvalhaes

CEO · Co-founder · Client & Growth

Jonah leads LuminalQ’s client relationships, strategy, and growth. His real job is to understand a business well enough to know exactly which part of it should be automated first (and, just as importantly, which parts shouldn’t).

That instinct comes out of an industrial engineering background. Jonah studies Industrial Engineering at the University of Massachusetts Amherst and got his start as a Quality Engineer at Draper in Cambridge, where he learned to treat any process as something measurable, improvable, and occasionally worth tearing down and rebuilding. He holds a Six Sigma Green Belt from the Institute of Industrial and Systems Engineers. But the idea that shaped him most came from a book, Donella Meadows’ Thinking in Systems: the conviction that almost anything (a business, a workflow, an entire market) can be modeled as a system. Once you see a problem that way, its leverage points begin to surface: the few places where a small, well-aimed change moves the whole thing. That is how he reads a client’s operation. Find where the system actually turns, change that, and leave the rest alone.

Jonah is Brazilian and American, a native English speaker who’s also fluent in Portuguese from a stretch of schooling in Brazil growing up. He’s a keen traveler, too: four months studying abroad in Japan, where he started picking up the language, plus trips through China and South Korea. These experiences influence how he works with clients, showing up with curiosity and adaptability.

Lucca
Siffredi

CTO · Co-founder · Engineering

Lucca leads LuminalQ’s engineering. Once a business’s real bottleneck has been found, his job is to turn it into a system that actually holds up: the AI agents, client databases, and data pipelines that quietly do the work, run after run, long after the build is done.

He graduated from Southern Oregon University with a bachelor’s degree in Computer Science and Mathematics, and got his start as a developer there: building software for the research platform FlowJo and a system that turned raw JSON into fully editable PowerPoint reports, early proof of the thing he still cares about most: taking messy, manual work and making it structured and repeatable. He builds across Python, TypeScript, JavaScript, C++, and SQL/PostgreSQL, and works just as comfortably at the newer edge of the stack: AI agents, embedding vectors and vector databases, retrieval-augmented generation, and prompt engineering. (The site you’re reading is his, built from scratch in TypeScript, React/Next.js, and Tailwind.) The mathematics background is what shaped his instinct for the work: a bias toward structure, rigor, and systems that can be reasoned about rather than guessed at. It’s the same discipline behind LuminalQ’s glass-box approach: every step a system takes should leave behind something you can read and trust.

Originally from Spain before moving to the United States, Lucca is fluent in Spanish and English.

Working with us

We take on a small number of engagements at a time, which means the work stays focused and the people building your system are the same people you talk to. If you’re looking to improve something within your business, we’d love to talk to you.

Book a Free Discovery Call