Data Concept Studio is the trading name of Data Concept Studio Michał Szarek, a software company registered in Poland (Kraków), founded in 2025.
We are a product company — we build, own, and operate our own AI software, not client projects — running everything on Google Cloud. We currently ship two products: Briefcaster and MATHS.
Building AI systems at the intersection of finance, machine learning, and engineering.
I have spent the past decade working across financial services, software engineering, and AI — first as a practitioner inside institutions, then building tooling for them. That dual perspective shapes everything about how Data Concept Studio is designed: systems that are rigorous enough for real money, but fast enough for a startup.
On the finance side, I have worked across trading operations, risk frameworks, and capital markets technology — understanding not just how markets work, but how decisions are made under uncertainty and time pressure.
On the engineering side, I have designed and delivered full-stack systems end-to-end — from data pipelines and APIs to cloud infrastructure and production deployments. I have led teams, shipped products, and maintained systems that cannot afford downtime.
The convergence of large language models with real-time data is the most significant opportunity I have seen in my career. Data Concept Studio is my vehicle for building products on that frontier — AI that does real, daily work for the person using it.
Trading operations, risk management, capital markets technology, and quantitative analysis across multiple asset classes.
Multi-agent systems, LLM integration, retrieval-augmented generation, ensemble modelling, and production ML pipelines.
Cloud-native architecture, APIs, data pipelines, and infrastructure for our own products — designed, built, and run in-house.
Leading engineering teams, defining technical strategy, and delivering complex projects from concept to production.
The best AI capabilities have historically gone to those with the most infrastructure: institutional trading desks with quant teams, media companies with editorial staff. Individuals get generic products built for everyone — which means built for no one in particular.
Data Concept Studio builds AI that works for one specific person at a time. Briefcaster reads your sources and produces a briefing that exists for a single listener: you. MATHS brings institutional-grade trading decision support — systematic, auditable, risk-aware — to individual traders and smaller funds.
Both share the same engineering DNA: production systems with real accountability, human oversight where it matters, and infrastructure rigorous enough to trust with your money or your attention.
Data Concept Studio is a small, independent software company. There is no holding structure, no outside shareholders, and no client-services arm — the whole company is the products it builds and the person who builds them.
We are a digital-native business: every product is software we own and operate, delivered over the internet, sold directly to the people who use it. There are no billable hours in the model and no physical goods — our main running cost is cloud infrastructure and AI inference, which scales with usage rather than headcount.
Briefcaster is free to download and free to use today — there is no paywall in the app. Our planned revenue stream is a consumer subscription that raises usage allowances and unlocks longer briefings, more sources, and additional voices. It is not live yet, and we publish that state honestly rather than implying revenue we do not have.
MATHS is in private beta and is not sold today. When it opens up, access will be offered as a subscription to the platform.
See pricing & business modelBoth products run entirely on Google Cloud. A one-person company can only ship two production systems by leaning hard on managed infrastructure and by automating the operational work that would otherwise need a team.
Serverless containers on Cloud Run, managed PostgreSQL on Cloud SQL, and object storage for generated audio. Infrastructure is defined as code and applied through review, not clicked together in a console.
Vertex AI for language models and speech synthesis. Briefcaster scripts and narrates every briefing through this pipeline; MATHS uses models to reason over market data before a decision reaches a hard risk gate.
Automated tests gate every change, and both systems watch themselves — health checks, reconciliation between our records and external sources, and alerting that pages a human when a pipeline goes quiet.
Briefings produced by Briefcaster are generated by AI, and the app says so — we treat AI transparency as a product requirement, not a legal afterthought.