Initial priority01
AI in business
What does AI actually improve in cost, quality and productivity? Where is human intervention still needed?
Rovemark Research
What changes when intelligence goes to work?
AI is entering businesses. We want to investigate how it changes productivity, work and everyday life — and who shares the gains.
Field of studyAgents, memory and robotics
Conceptual image generated with AI01 / First publication
Studies and references
Universal basic income, shorter working hours or concentrated wealth? A society with less human work is a scenario to investigate, with choices and costs to discuss.
Evidence analysisIf AI does the work, who gets the income?
AI, productivity and universal basic income: what evidence supports, and the choices technology cannot make for us.
Read the analysisProposed study · applications open
Does AI save time — or move the work into review?
An applied company study of production, review, rework and cost per approved deliverable. Open protocol; no original results published yet.
Protocol and application in Portuguese.Publications and tools
AI models
Compare observed results, complete answers and API-reported costs for six models on synthetic tasks.
ExplorePublications and toolsOccupational exposure
Search ILO occupations, compare task exposure and export a selection.
ExplorePublications and toolsLibrary
Browse studies, versions, data and revision notes.
ExploreWe bring together analyses of external sources, a technical test using synthetic data and proposed company studies. Each publication identifies the origin and scope of its evidence. Citing a source does not imply a partnership.
02 / Research
Fields of study
Eight connected fields. Our starting point is AI in business and its effects on work and income.
Initial priority02
Work and income
Which tasks change, which jobs might emerge and how can gains be shared? Where does universal basic income fit?
Field of study03
Economics and power
Who captures the value of automation? How do platforms, model ownership and market concentration change its distribution?
Field of study04
Education and knowledge
Does learning with AI mean learning more? How can we measure autonomy, retention and access to knowledge?
Field of study05
Health and well-being
When does AI help or harm? How can we assess benefits, dependency and limits in care settings?
Field of study06
Governance and safety
Who is accountable for an automated decision? How can we test privacy, bias, human control and agent security?
Field of study07
Energy and sustainability
What is the material cost of intelligence? How can energy, infrastructure and usefulness be compared across applications?
Field of study08
Agents, memory and robotics
What makes an agent reliable beyond a demo? How do memory, tools and physical interaction change outcomes?
03 / Research
Research agenda
Three proposals to guide our first studies. These are planned questions and approaches, not completed or ongoing research.
01Can a business grow without hiring?
Proposed agenda
Proposed approach
Compare processes before and after AI adoption, measuring time, total cost, quality and human intervention. Distinguish simulated tests from real operations; task exposure alone does not establish job losses.
02When productivity rises, who gains time and money?
Proposed agenda
Proposed approach
Investigate how gains may reach wages, profits, prices and working hours. Compare sectors and adoption patterns, making external factors and limits to causal conclusions explicit.
03If AI does the work, who gets the income?
Proposed agenda
Proposed approach
Analyze universal basic income, taxation and shorter working hours as scenarios. Examine funding and distribution against evidence from cash transfers, without treating a temporary trial as proof of an economy without work.
04 / Research
Data and methods
We start with available evidence. We measure when a specific question remains unanswered.
- 01
Consult public sources first. Record the author, date, version and limitations; attribute third-party findings to their publishers.
- 02
Propose a new test only for a specific question that existing sources cannot resolve. Define the question, criteria and budget before running it.
- 03
Distinguish external analyses, synthetic technical tests and company studies. Proposals are not findings. Publish methods, limitations, negative results and commercial interests.
References to start the debate
05 / Research
Debates
Questions to open the conversation. These titles are proposed topics, not published articles.
Will your next boss be an AI?
Should AI pay taxes?
Does control of agents mean control of the economy?
Have a question worth investigating?
Suggest a topic, critique or collaboration through our contact channel. Do not send personal data or confidential client information.