ECONOMICS()
Cost per verified outcome
Token price is an input metric. The operating metric is the full cost of a result that survives independent verification and arrives inside the decision window.
AI CTO / KRAKÓW, POLAND
AI systems that survive reality.
I build and operate AI at the point where model capability meets cost, verification, liability and human work. CTO of Archdesk. Founder. Former white-hat hacker. Author of Proof-Adjusted Autonomy and Joule Wars.
CURRENT ROLE
CTO, Archdesk
ORIGINAL WORK
PAA · Joule Wars
BUILDER RECORD
4 ventures · 1 acquisition
OPERATING MODE
Strategy → production
01 / OPERATING THESIS
The model is not the system.
An impressive model is a component. An enterprise AI system also needs context, tools, permissions, evidence, failure boundaries and a clear owner. The AI CTO is responsible for the complete loop.
The job is not to maximize how much work a machine can generate. The job is to maximize how much machine work the organization can safely accept. NULLIUS IN VERBA // TAKE NOBODY'S WORD FOR IT
02 / WHAT AN AI CTO OWNS
Four systems. One accountability.
AI leadership becomes operational when responsibility is explicit. These four systems determine whether a deployment compounds value or proof debt.
ARCHITECTURE()
Models, data and the harness
The durable system lives around the model: context assembly, tool permissions, observability, memory, evaluation, rollback and vendor portability.
PROOF()
Independent validation
A model grading itself produces correlated confidence. Evidence requires another mechanism: deterministic tests, replay, a different model family or a human at irreversible boundaries.
LIABILITY()
Ownership when AI is wrong
Capability does not determine adoption. Settled responsibility does. Every production action needs an owner, an evidence trail and a defined recovery path.
03 / ORIGINAL FRAMEWORKS
Concepts built from operating constraints.
A vocabulary for decisions that benchmarks do not answer: how much autonomy can be trusted, what AI work costs and where durable advantage lives.
PAA / PROOF
Proof-Adjusted Autonomy
The share of completed work that is autonomous and independently proven. Measures what an agent can prove, not merely what it claims to have done.
→
JOULE / ECONOMICS
Joule Wars
The AI race moves from raw capability toward useful intelligence per joule — an economic, infrastructure and geopolitical constraint.
→
HARNESS / SYSTEMS
The Agent Harness
The control plane around the model: context, tools, permissions, memory, evidence, evaluation and recovery. The model is replaceable; the harness compounds.
→
04 / OPERATOR EVIDENCE
Built across different failure modes.
Construction ERP, regulatory intelligence, industrial monitoring, marketplaces, algorithmic trading and security research share one property: the system has to work outside the demo.
ENTERPRISE OPERATIONS
CTO of Archdesk
Technology leadership for an enterprise operations platform serving construction, manufacturing and engineering workflows.
FOUNDER SYSTEMS
Four ventures
Founder of Lextron.ai, Inclify, Rejsomat.pl and Robotero — spanning RegTech, industrial AI/IoT, marketplaces and algorithmic trading.
ADVERSARIAL ROOTS
White-hat hacker
Security research from 2004–2012, including responsible disclosure of critical flaws in Sun Microsystems' sun.com. Read the record.
05 / FIELD NOTES
The work, in public.
Selected essays on the economics and operating architecture of intelligence.
AI AGENTS / VERIFICATION
Proof-Adjusted Autonomy: The 90% Agent Is a 61.6% Agent
READ FIELD NOTE →AI ECONOMICS / LABOR
The Unit of Work Is the Agent-Hour
READ FIELD NOTE →AI ENGINEERING / PROOF
Verification Cost Is the New Bottleneck
READ FIELD NOTE →AI INFRASTRUCTURE / ECONOMICS
OpenAI's Economics: GPU-Constrained, Not Demand-Constrained
READ FIELD NOTE →06 / OPERATING PRINCIPLES
How I make AI decisions.
Start with the irreversible boundary.
Identify where an error becomes expensive, public or impossible to undo. Architecture flows backward from that line.
Separate generation from proof.
The mechanism that produces an answer should not be the only mechanism that declares it correct.
Price the accepted result.
Include inference, orchestration, retries, verification and human exceptions. Cheap tokens can still produce expensive work.
Own the harness, rent the model.
Model leadership rotates. Context, workflow integration, permissions, evaluations and evidence become organizational memory.
Measure autonomy after proof.
A task is not autonomous because a human did not touch it. It is autonomous when the organization can accept it without borrowing against a future incident.
CITE_THIS_PROFILE()
Piszczek, M. AI CTO — Enterprise AI Systems, Proof and Economics. piszczek.pl. https://piszczek.pl/ai-cto
07 / DIRECT ANSWERS
AI CTO, without the theatre.
What does an AI CTO do?
An AI CTO turns model capability into a reliable operating system for the business. The role owns the economics of verified outcomes, architecture and data, independent validation, security, liability boundaries and the organizational change required to put AI into production.
Who is Michał Piszczek?
Michał Piszczek is the CTO of Archdesk and a technology founder based in Kraków, Poland. He founded Lextron.ai, Inclify, Rejsomat.pl and Robotero, previously worked as a white-hat hacker, and is the author of Proof-Adjusted Autonomy and Joule Wars.
What is Proof-Adjusted Autonomy?
Proof-Adjusted Autonomy measures the share of completed work that an AI system executes autonomously and supports with complete, independent, reliable and timely evidence. It measures what an agent can prove, not merely what it claims to have completed.
What are the Joule Wars?
Joule Wars is the thesis that the durable AI competition moves from raw model capability toward useful intelligence per joule. As capability commoditizes, energy efficiency becomes an economic, infrastructure and geopolitical constraint.
Where is Michał Piszczek based?
Kraków, Poland. Work and speaking are available in Polish and English. For press, podcasts, panels and keynotes, use the media and speaking page.
OPEN_CHANNEL()
High-signal problems only.
Strategic AI deployment, complex systems architecture, C-level advisory, podcasts and keynotes.