Disclosure

Effective date: July 4, 2026

Disclosure

This Disclosure page explains how OCO publicly describes AI-assisted work, automation, examples, generated content, and system limits. It is written for public website context and does not disclose private client systems, private datasets, internal prompts, security-sensitive controls, or unreleased architecture.

The disclosure is meant to make OCO’s public position understandable without oversharing implementation details. AI may appear in internal workflows, product concepts, specialized models, research assistance, software delivery, security review, and public examples, but the level of disclosure depends on whether the system is public, client-owned, private, experimental, or approved for release.

AI-Assisted Work

OCO may use AI-assisted tools for drafting, research support, code generation, analysis, translation, testing support, documentation, workflow automation, and system review. AI assistance is treated as a tool within an accountable engineering process. It does not replace ownership, review, judgment, security controls, project scope, or approval requirements.

OCO may use multiple AI providers, models, local tools, agents, or automation workflows depending on the task and approved boundaries. The use of AI does not mean that every output is final, that a model has independent authority, or that private data is automatically eligible for model processing.

Data and Confidentiality

OCO does not use public website visitors’ confidential information to train public AI models. Public examples may be illustrative, simulated, generic, or simplified. Private client material, private datasets, credentials, security-sensitive details, and unreleased architecture should not be submitted through public channels and are not disclosed on the public website.

When sensitive data is involved in a scoped project, OCO expects the handling path to be defined before processing. That may include what can be sent to a model, what must stay local, what must be redacted, what can be logged, who can review output, and what material must be excluded from public examples or future training workflows.

Human Review

Outputs that affect public release, security posture, client work, system behavior, policy, commercial decisions, or sensitive data handling require appropriate human review. Review may include source checking, code review, test execution, security review, translation review, factual correction, and approval by the person responsible for the work.

Human review is not a single checkbox. Depending on the work, review may include technical review, factual review, translation review, security review, legal or compliance review by qualified parties, owner approval, or comparison against deterministic system output. The review method should match the risk of the use case.

No Independent Advice

AI-assisted output shown on the public website is not independent legal, medical, financial, investment, security, compliance, or professional advice. It may summarize, illustrate, draft, classify, translate, or support analysis, but important decisions require qualified human review and the proper professional context.

If a public example resembles a professional workflow, it should still be treated as an illustration of engineering structure, not as a recommendation to act. Any regulated, high-risk, safety-sensitive, financial, legal, medical, or security-critical decision needs its own qualified review and written context.

Automated Systems

OCO may design or operate automation for data collection, normalization, classification, routing, reporting, deployment, monitoring, security review, API delivery, or deterministic software workflows. Automation should have purpose, limits, logging, review paths, and a way to correct or stop behavior when conditions change.

Automation may connect data sources, APIs, queues, build systems, deployment targets, notification paths, dashboards, or model calls. OCO’s position is that automation should be observable and reversible where practical, with logs, approvals, limits, and failure handling appropriate to the consequences of the workflow.

Public Examples

Public examples, animations, diagrams, JSON samples, interface text, data records, venture descriptions, and technical scenarios may be simulated, reduced, anonymized, or illustrative. They are used to explain the engineering approach without exposing private data, live systems, client records, proprietary thresholds, or security-sensitive implementation detail.

A public animation, diagram, card, or sample record may show the shape of OCO’s work without representing a live client dataset or production system. Viewers should not infer that sample identifiers, values, confidence scores, security conditions, or record states are real unless OCO clearly says they are public production data.

Limits

AI and automation can be incomplete, incorrect, biased, stale, overconfident, or unsuitable for a particular context. OCO’s public description of AI models, engines, protocols, APIs, evidence layers, or software products does not guarantee performance, availability, accuracy, regulatory suitability, or fitness for a specific use without a written scope and evaluation plan.

Limit disclosure is part of trust. OCO does not want public descriptions to imply that an engine, model, protocol, API, or software interface can replace the governance, evidence, testing, security, and commercial evaluation needed for a specific deployment. Public pages describe the approach; scoped work defines the actual commitments.

Contact

Questions about public legal, privacy, security, governance, or disclosure information can be sent to info@oco.io. Do not send credentials, private datasets, exploit details, production secrets, or confidential third-party material through public email.

Public email is a triage channel. If OCO decides that a matter requires sensitive exchange, the next step may require a separate process, verified identity, written scope, secure transfer path, or owner approval. Until that exists, public email should be treated as unsuitable for secrets, regulated data, exploit material, or confidential project files.