Work — IntuiTek¹

Professional Workflow Design

AI-assisted document and decision support

A professional-practice workflow integration explored how software and AI could support existing expertise without forcing a new technical burden. Specific adoption counts and measured client benefits remain unpublished pending verification and permission.

Private Client disclosure and results unverified publicly

The Question

How can complex document work become easier to navigate while leaving professional judgment with the person responsible for it?

The Approach

The work examined document movement, repetitive decisions, consistency checks and possible Microsoft 365-based support. Technology choices were treated as downstream of the actual workflow.

Evidence Boundary

This is private engagement material. Client identity, adoption statistics, deliverables and performance outcomes are not independently verifiable from this public page. Counts and case-specific conclusions should only be published with supporting records and authorization.

The Lesson

A system should be judged by the work it changes and the level of control it preserves—not by how many automated steps can be counted.

Adversarial Document Review

Constraint-aware documentary reasoning

An exploration of evidence structuring and adversarial checks in consequential documentation. This public account does not claim a court victory, opposing-party defeat, or a verified legal result.

Method Outcome claims withheld pending independent proof

The Question

What would it take to make a complicated documentary position traceable under contradictory evidence, deadlines and procedural constraints?

The Approach

The method examined supporting sources, assumptions, internal consistency, procedural alignment and potential failure points. These are reliability questions that can be evaluated without implying that an automated system provides legal representation.

Evidence Boundary

Private documents, any court record, independent review and legal outcome have not been supplied for public verification here. This case is a method description rather than a performance or litigation-success claim.

The Lesson

When errors carry consequences, clear provenance and explicit uncertainty are more valuable than persuasive language about what a system supposedly achieved.

Persistent AI Operations

Memory, delegation, recovery and bounded execution

An operating laboratory for retaining context and coordinating bounded AI-enabled work across tasks and tools. Its uptime, independence and capability totals require time-bounded measurement; deployment is not proof of uninterrupted operation.

Ongoing Operating laboratory; reliability subject to measurement

The Question

How can AI-supported work retain context, recognize failures and resume after interruptions without losing the human operator's intent?

The Approach

The environment combines scheduling, memory, tool execution, state reconciliation and verification. Work is divided between permitted actions, explicit decisions and events that have actually been observed.

Evidence Boundary

The system has undergone iterations and real operating faults. Public claims of continuous 24/7 autonomy, exact skill counts, complete self-maintenance or zero cloud dependency are withheld until a dated measurement record supports them.

The Lesson

Agent reliability is a systems problem: recorded state, clear authority, fault recovery and verified effects matter at least as much as the model's immediate reasoning.

Machine Execution Under Economic Constraints

On-chain observation, automation and transaction reliability

An investigation of automated market observation and transaction execution, including the limits imposed by latency, fees and competitive systems. No profitable-trading or zero-cost-execution result is claimed on this page.

Research Economic outcome not publicly established

The Question

How should an automated system decide whether a technically possible transaction is economically worthwhile, and how can its result be independently verified?

The Approach

Exploratory work considered on-chain events, decision windows, simulation, transaction finality, network costs and failure handling. A working scanner, a submitted transaction and a profitable outcome are distinct claims.

Evidence Boundary

Execution profitability, recovered costs, return on capital and customer demand are not demonstrated by code or architecture alone. Any future public numerical claims require independently reproducible receipts, the relevant period and complete cost attribution.

The Lesson

A system that does more is not necessarily a system that creates more value. Economic outcomes require settlement, usable fulfillment and cost-complete accounting.

Judgment Preservation Research

Unified Cognitive Substrate and related work

The Unified Cognitive Substrate paper and companion software have verifiable Zenodo records. Related independent concepts and specifications are presented according to their actual publication and validation status—not automatically as five DOI-published frameworks.

11 Experiments reported in the UCS preprint

DOI Paper and code publicly citable

The Question

What parts of judgment are lost when a persistent agent compresses or replaces its conversational context, and which components can be preserved and tested?

The Approach

The published Unified Cognitive Substrate investigation reports eleven experiments and a specific improvement in context-sensitive routing differentiation. Those measurements are not a general benchmark of artificial intelligence or an independent guarantee of persistent-agent reliability.

Evidence

The versioned research paper is available through DOI 10.5281/zenodo.18794691 and the companion software through DOI 10.5281/zenodo.18794175. Other frameworks and essays should be evaluated individually; public posting, a specification draft and a peer-reviewed publication are not equivalent.

The Lesson

Research becomes credible through falsifiable methods, transparent limitations, accessible sources and willingness to revise claims when experiments contradict the initial design.

Further questions

The record is not the endpoint.

The examples above are not a catalog of services or a promise of immediate availability. They are records of work, decisions, and technical approaches that can be examined in their own context. If you see a connection worth discussing, there is a direct way to make contact.

Start a conversation → Explore the research →

What actually happened,and what it meant.

Professional Workflow Design

The Question

The Approach

Evidence Boundary

The Lesson

Adversarial Document Review

The Question

The Approach

Evidence Boundary

The Lesson

Persistent AI Operations

The Question

The Approach

Evidence Boundary

The Lesson

Machine Execution Under Economic Constraints

The Question

The Approach

Evidence Boundary

The Lesson

Judgment Preservation Research

The Question

The Approach

Evidence

The Lesson

The record is not the endpoint.

What actually happened,
and what it meant.