Scope
This policy represents me, Logan R. Ramos. It applies to professional work I perform for myself, for an organization, or as a participant in a project. It covers engineering ideation, tool and technology decisions, technical documentation, code, generated assets and datasets, analysis, and communication.
When an organization, client, contract, or project has stricter requirements, those requirements control. This policy describes my own practice and the responsibilities within my authority; it does not replace an organization's governance, security, legal, or disclosure obligations.
Core principles
- Human accountability. I remain responsible for every conclusion, decision, and deliverable.
- Verification. AI output is treated as untrusted assistance until I review it against primary evidence and project requirements.
- Proportional use. AI should reduce low-value effort or broaden analysis, not obscure ownership or replace judgment.
- Transparency. Material AI assistance is recorded at the project level for the relevant stakeholders.
- Data minimization. Sensitive information is excluded from model context and indexing.
Permitted uses
The following categories define where AI assistance may be used and the boundary applied to each use.
3.1
Engineering ideation
AI may be used to generate proofs of concept and demonstrations that expose scope, dependencies, and project impact with minimal effort. These artifacts turn speculative exploration into evidence that can challenge assumptions. AI-generated proofs of concept are not delivered to production unchanged.
3.2
Tool and technology decisions
Given appropriate context, AI may help organize tradeoffs, produce pros-and-cons analyses, and develop counterarguments to my position. It is a supplemental check against omission and bias—not the source of the decision. I own the reasoning and validate material claims before relying on them.
3.3
Technical documentation
For research intended for an audience, AI-generated prose is not used as the substance of the work. AI may assist with structure and the delivery of my own reasoning. For production documentation, AI may be used deliberately to check factual consistency, reduce staleness, and document version changes.
I favor standardized, application-level documentation that serves people and software agents. Repository
instructions—such as .cursorrules, CLAUDE.md, and AGENTS.md—should
direct agents to authoritative sources, reduce unnecessary context, and constrain automated maintenance.
3.4
Code
AI-generated code is limited to proofs of concept, testing, technical exploration, asset-generation workflows, and operational changes. I review that code line by line. Production acceptance additionally depends on the project's normal testing, security, approval, and release controls.
3.5
Asset and dataset generation
AI may generate images, video, visual components, synthetic data, or other assets where manual creation is intentionally deferred or outside the project's intended investment. Generated assets must be fit for purpose, lawfully usable, appropriately identified to stakeholders, and reviewed for misleading, infringing, or harmful content.
3.6
Analysis and incidents
AI may assist with reverse engineering for ideation or integration work. During production incidents, it may be used with approved, sanitized evidence from application logs and persistence layers to form hypotheses or accelerate investigation. Its output is not treated as a finding until it is verified against the underlying system and observable evidence.
3.7
Communication
AI may help structure, edit, or clarify professional communication. I review the final message, remain its author, and am accountable for its accuracy, tone, audience, and commitments. AI is not used to impersonate another person or manufacture evidence, endorsements, or consensus.
Mandatory human review
Any and all AI-assisted work must be reviewed and analyzed by me before I accept, publish, deliver, merge, deploy, or rely on it. Review is proportionate to risk and may include line-by-line inspection, testing, source verification, security review, license review, and comparison with project requirements. The use of an AI system never transfers accountability to the system or its provider.
Project-level disclosure
I disclose material AI assistance on a per-project basis. Project documentation should identify the relevant use categories, the level of assistance, and any project-specific restrictions or review requirements in a standardized form.
These records belong to the project's stakeholder and are not normally public. They may be published when the stakeholder chooses to do so or when the project is open source. This public policy explains my baseline practice; it is not a substitute for a project's own disclosure record.
Data and confidentiality safeguards
I do not submit API keys, credentials, personally identifiable information, confidential information, or other protected material to AI systems I operate, nor do I permit that material to be indexed for AI use. Files that may contain protected material are placed on explicit deny lists in agent configuration.
Deny lists are a supporting control, not a substitute for judgment. I minimize context, sanitize diagnostic material, use only approved tools and access paths, and follow the stakeholder's security and retention requirements. If material cannot be used safely, it is excluded from the AI-assisted workflow.
Rights, rules, and prohibited conduct
AI use does not lower my obligations. I comply with applicable law, contracts, organizational policies, licenses, attribution requirements, platform terms, privacy duties, and intellectual-property rights.
I do not knowingly use AI to:
- evade a law, contract, license, security control, or governing project rule;
- deceive an audience about authorship, evidence, identity, endorsement, or system behavior;
- expose, reconstruct, or improperly infer protected or confidential information;
- produce discriminatory, abusive, malicious, or unsafe outcomes; or
- replace required human judgment, approval, or professional accountability.
Governance and updates
Project-specific questions, corrections, or concerns should be raised through that project's established stakeholder channel. I will revisit this policy as AI capabilities, professional standards, and applicable obligations change. The date above identifies the current version.