Tatami
In the shade of cherry blossoms, there are no strangers. 花の陰 赤の他人は なかりけり — Kobayashi Issa, 1819
The Workshop
Ideas arrive unfinished. They leave as systems, architecture, and understanding.
Ways of working
Independent decision review
- For an AI, vendor or infrastructure choice: a decision memo with options, assumptions, implications and missing evidence.
Architecture and ownership review
- For unclear responsibilities or exit options: a system map, ownership gaps, tested and untested alternatives, and ordered next actions.
Build and transfer
- For a direction ready to implement: a working system, decision records, operating guidance and a usable handover.
These describe possible outputs. Scope, timing and availability are agreed around your situation.
Bring a decision or an unfinished briefThe Workbench
Active practice on the table — areas under investigation and systems being built. Not products on display.
Areas of attention
Sovereign AI
AI capabilities that organisations can operate, audit, and retain — without surrendering the control plane.
AI Strategy and Operating Models
From experimentation to reliable value: decision rights, governance patterns, and workflows that survive auditors and real operations.
Digital Sovereignty
Knowing where data lives, who can touch it, and what it costs to leave — before dependency becomes strategy.
Resilient Infrastructure
Platforms engineered to perform under pressure and priced to stay under your control.
Board and leadership counsel
Hard technology decisions: what to build, what to buy, and what never to surrender.
Product and platform development
Builder work — products, platforms, and research systems measured by outcomes.
Tatami
In the shade of cherry blossoms, there are no strangers. 花の陰 赤の他人は なかりけり — Kobayashi Issa, 1819
Stealth 02
Stealth 03
Stealth 04
Drawings and Systems
Selected public records. Working sheets, not case-study theatre.
The Studio — a working publishing system
Keep multilingual writing, evidence and editions understandable to readers and agents.
One publishing record supports EN/DE/FR reading, associative MEMEX discovery, edition histories and a read-only agent interface.
Panopticon 2.0
Leaders lacked a practical field guide to digital sovereignty under systemic surveillance pressure.
Book: Governing in an Age of Total Surveillance — frameworks across cloud, AI, infrastructure, and governance.
AI Constitution
Retain instructions across agent sessions without overwriting each project’s rules.
An MIT-licensed toolkit for shared instructions, project onboarding, model information and consistency checks. The public manual identifies v0.1.0; check the selected release before installation.
Digital Sovereignty Assessment
Cloud and SaaS adoption often outpaces visibility into data residency, vendor lock-in, and exit cost.
Interactive six-layer maturity assessment with executive-oriented scoring and roadmap framing.
Engawa — an appointed door for agents
Give a visitor’s assistant reliable public material without exposing private operating tools.
Discovery resources, canonical Markdown and a read-only public MCP used on this site. Public reading and authenticated operations have separate interfaces.
CFA Institute Research & Policy Center
Institutional audiences needed clear analysis at the intersection of markets, technology, and governance.
Published contributions via the CFA Institute Research & Policy Center author profile.
Past Work
Concluded work that still teaches something. Not current operations.
Kurz Notiert — AI financial news for Basler Kantonalbank and Bank Cler
The banks needed timely financial news that readers could navigate across markets and languages.
The MachinaLabs team built an AI-powered financial-news platform that selected and summarised news, assessed relevance, and analysed sentiment in English, German and French.
MachinaLabs AG — Former venture
Financial data markets needed practical AI under real operational and financial constraints — usable intelligence, not slideware.
Swiss applied-AI and financial-data company focused on secure, practical intelligence systems.
Workshop principles
How work is approached — not a feature list.
Observe before building
Questions are investigated until they become understanding. Speed is never the objective.
Clarify ownership
Critical capabilities stay under operational control. What you cannot operate, you do not truly own.
Reduce dependency
Replaceable dependencies and exit options before lock-in becomes strategy.
Design for failure
Systems that hold under pressure — reliability before novelty.
Make systems understandable
Complexity in; clarity out. Technology should disappear; understanding should remain.
Leave maintainable structures behind
Design for decades, not demos. What survives the pencil deserves the keyboard.
Other doors
When one of these questions has become concrete in your organisation, begin a conversation.