
Gemchanger·Barcelona
Founder
Research & markets
Finance graduate with energy-sector experience in oil, plus time in venture. Covers how markets, liquidity, and narratives meet institutional reality.
Profile on X ↗Why crypto, fintech, deep tech, and anything in between rewards obsessive due diligence.
@CapitalADHD · Research collective
We start with public, rigorous work—protocols, infrastructure, and technical stacks where evidence beats narrative. Capital is secondary and selective.
No abstract lattice—this is the work product: research artifacts first, everything else follows.
Primary outputs
Long-form memos, scenario tables, and dashboards you can reproduce.
Domains
On-chain and fintech rails, L1/L2 economics, DeFi & custody risk, wallets & infra, frontier AI, robotics.
Methods
Chain-native metrics, adversarial reviews, agent-assisted synthesis, and open methodology notes—we invest in how the desk uses AI without hiding the reproducibility bar.
Capital
Optional and selective—only when research conviction is already explicit.
We work alongside
Why we exist
We bias toward teams who want reviewers that read code, cite sources, and write the uncomfortable questions into the appendix—whether or not a check ever follows.
At a glance
One bar, shared definitions of proof, and artifacts you can stress-test. We default to memos, models, and traceable work—roadshows and vanity decks are the exception.
Collective
One research bar
Three seats · one standard
Three seats covering markets, systems, and publishing—one standard of rigor, no siloed “opinions without work product.”
Geography
Singapore ×1 · NYC ×1 · Barcelona ×1
Built across Asia, North America, and Europe—remote-first review cycles with on-site time when hardware or lab work demands it.
Mandate
Crypto, fintech, deep tech, and anything in between
L1/L2 and DeFi plumbing, fintech and custody rails, wallets and infra, frontier AI, robotics, and adjacent hardware. If it falls outside that perimeter, we are not the right desk—saying no is part of the product.
Research thesis
We are a research collective first—Crypto, fintech, deep tech, and anything in between. When we deploy capital, it is early, concentrated, and always downstream of work you can read and stress-test.
AI on the desk
We spearhead research workflows with AI agents—custom pipelines for ingestion, stress-testing, and publication—and actively support that development so rigor scales without becoming opaque.
Approach
We publish memos, models, and datasets you can trace—protocol economics, security assumptions, frontier compute, and explicit what-has-to-be-true statements. We spearhead large parts of the desk with AI agents (retrieval, drafting, adversarial review loops) and we support that tooling so the method stays inspectable—not a black box.
Crypto networks, fintech rails, L1/L2 and DeFi plumbing, wallets and infra, AI systems, robotics, and the messy layers in between—we say no when the work doesn't match the bar.
We are not a spray-and-pray fund. The team stays small; deploying capital is rare and only after the research work is already standing on its own.
Team
The people behind the desk—operators and researchers who publish before they pitch.

Gemchanger·Barcelona
Founder
Finance graduate with energy-sector experience in oil, plus time in venture. Covers how markets, liquidity, and narratives meet institutional reality.
Profile on X ↗
Feuter·Singapore
Founder
Graduate training with VC exposure—deep on analytics, modelling, and company analysis. Former owner of a contracting business; brings operator judgment to diligence.
Profile on X ↗
Kafka·NYC
Founder
Computer science background; hands-on engineering with projects spanning YC- and Paradigm-style stacks. Codes, reviews, and stress-tests assumptions.
Profile on X ↗Contact
Share the problem, what is already built, and what you want pressure-tested. We reply when the research can add real leverage—capital is downstream and optional.
Operating note
Frontier tech rewards teams that can show mechanism, not just momentum. Our default output is research you can fork—not a deck you admire once.
When we allocate, it is narrow, repeatable, and always late in the process—never a substitute for the work product.