Project summary

My work starts with a question: What is happening that we still cannot see?

For more than three decades, I have worked with sales, operations, people, processes, negotiation and business decisions. I learned that visible results are often produced by dynamics that do not appear clearly in reports or indicators. This became the basis of VISÃO 360°, an organizational investigation method that I am developing.

While working with AI, I saw the same problem in a new form. AI can give a confident answer before the organization has checked whether the evidence is sufficient, what assumptions support the conclusion, who remains responsible, or when a human should stop the process. I do not believe the challenge is choosing between human and artificial intelligence. The challenge is building a process in which both continuously validate each other.

I want to test a limited public research protocol derived from selected VISÃO 360° principles. It will require the user to state the decision, record competing hypotheses, separate evidence from inference, mark uncertainty, identify disconfirming evidence, assign human responsibility and define escalation or STOP conditions.

This is a pilot. I am not claiming that the method is scientifically validated. The purpose is to discover where the protocol works, where it fails and whether a larger independent study is justified.

What are this project's goals? How will you achieve them?

Over four months, I will prepare and preregister six organizational decision scenarios involving AI-assisted recommendations. The scenarios will cover automation bias, unsupported confidence, missing evidence, suppressed alternatives, unclear responsibility and failure to escalate.

Each case will use the same bounded evidence packet. A human evaluator and fresh AI sessions will review the case separately. I will compare their decisions, disagreements, unsupported inferences and escalation choices. The rules, expected warning signs, scoring criteria and removal conditions will be fixed before the scored tests begin.

External reviewers will challenge the scenario design and the conclusions. I will publish the usable scenarios, anonymized decision records, errors, negative results and limitations. At the funding goal, I will add cases, second reviews and one end-to-end run conducted by someone outside the project.

The main question is simple: can this structured evidence-and-escalation process detect predefined failures in AI-assisted organizational decisions consistently enough to justify a larger study?

How will this funding be used?

The minimum of US$12,000 funds a complete six-scenario pilot. The US$20,000 goal adds up to four more scenarios, second reviews and independent replication.

Minimum / goal:

- My work on design, testing, analysis and documentation: $6,000 / $9,000

- Independent reviewers and second checks: $2,500 / $4,000

- Scenario preparation and participant support: $1,000 / $2,000

- AI access, secure records, storage and archive: $1,000 / $1,500

- English editing, translation and publication: $1,000 / $1,500

- Administration and IP/legal review: $500 / $1,000

- Independent end-to-end replication: $0 / $1,000

The public output will include a preregistered study plan, the test scenarios, an anonymized dataset, a limited public research edition of the protocol, a pilot report and a replication guide in English with a Portuguese companion version.

The VISÃO 360° Kernel, OIPS, proprietary scoring engines, commercial diagnostic instruments, books, client processes and confidential organizational data are outside this grant and will not be disclosed.

Who is on your team? What's your track record on similar projects?

I am Fabio dos Santos Borges, a Brazilian strategic advisor, author, independent researcher and third-semester Technology in Business Management student at FATEC Praia Grande.

My experience comes from more than three decades of practical work with sales, operations, people, negotiation, processes and organizational decisions. I created VISÃO 360° to connect human, structural and strategic evidence and investigate what isolated indicators or specialists may not reveal.

My comparative advantage is not frontier-model engineering. It is understanding how organizations interpret evidence, distribute authority, hide or reveal disagreement and make decisions under uncertainty.

VISÃO 360° has a developed conceptual architecture and documented instruments, but it has not yet been independently validated. That is why this project includes preregistration, external review, negative-result publication and replication.

AI has supported research, comparison, drafting and translation. Authorship, methodological decisions, publication and responsibility remain human.

What are the most likely causes and outcomes if this project fails?

The protocol may miss a serious warning sign, produce unstable results across AI sessions, create too much disagreement between human reviewers, or prove too slow for practical use. My own choices as the method's creator may also bias the cases or scoring.

These are not reasons to hide the result. I will preregister the materials, preserve version history, use external reviewers and publish the failures. A negative result can show where the protocol should be narrowed, changed or stopped.

If a real organizational case cannot be shared, I will use synthetic or public retrospective scenarios rather than expose confidential client information. Conclusions will be limited to the tested cases. I will not claim adoption, university sponsorship, peer review or general effectiveness without evidence.

How much money have you raised in the last 12 months, and from where?

I have received US$0 in grant funding during the last 12 months. VISÃO 360° has been developed through self-funded work.

I submitted a separate US$18,000 application to Coefficient Giving for a January 2027 career-transition and foundational study period. It is pending and is not money raised. This Manifund proposal funds a later empirical pilot, from 18 May to 17 September 2027. The same work or cost will not be charged to two funders. Any overlap or budget change will be disclosed before funds are accepted.