Explore NV360

Giving AI a task leaves important questions unanswered.

Which information should guide it? What may it change? What needs to be checked? When should it stop? Who decides what happens next?

NeuroVision360 is developing CSD/IBP to address those questions as part of how the work is designed and carried out.

The aim is to expand what people can accomplish while keeping their judgment, authority, and accountability central.

An AI response can sound convincing while using the wrong source, overlooking a requirement, or changing an approved decision. When AI can use tools, those errors can affect files, messages, and other systems.

CSD/IBP addresses the conditions around that work: its purpose, information, sequence, permissions, checks, and human decisions.

Cognitive Systems Design (CSD) defines the operating structure. It establishes what governs the work, what the technology may do, and where people must decide.

Instructional Build Protocol (IBP) puts that structure into practice through stages, checks, records, and controlled changes.

The names reflect the method’s origins in instructional engineering. Its broader development asks how that discipline could support other kinds of AI-assisted work.

Human direction starts before a result is produced.

People establish the purpose, decide the boundaries, and identify what matters. Their approved decisions should carry forward. Missing information and conflicting requirements should be made visible.

Reviewing a file is different from rewriting it. Drafting a message is different from sending it. Proposing a change is different from receiving permission to make it.

CSD/IBP is designed to preserve those distinctions throughout the work.

We are examining how the method could support:

Complex workflows. Keeping work within its agreed purpose and stage, with required checks before it advances.

AI that uses tools. Defining which actions are permitted, which require approval, and when execution must stop.

Work that depends on evidence. Distinguishing supported information from assumptions and unresolved questions before a result is used.

Projects that continue over time. Preserving sources, approved decisions, changes, and unfinished work so the next stage begins from an accurate record.

These are development directions. Each requires its own design, implementation, and testing.

CSD/IBP emerged while developing GRACIE, where complex work exposed recurring problems with sources, decisions, sequence, and verification.

GRACIE remains its instructional-engineering application.

NV360 Labs examines applications beyond that field. The Human-Directed Email Agent is a bounded test in a familiar setting: a personal inbox where choosing the right messages, respecting permissions, and stopping when necessary matter.

The result and the process both matter.

Did the intended work happen? Were the required checks followed? Did the system stay within its authority?

CSD/IBP’s development uses records of the work to investigate failures, propose specific corrections, and test whether those corrections improve performance without breaking what already works. People decide which changes are adopted.

The larger opportunity is a repeatable way to govern and improve AI-supported work across different applications. That possibility is what we are investigating—not something we assume one successful result has proven.

Technology should work for people. Not replace them.