Data use rules for source-grounded AI workflows.
VexASI separates public evidence, inquiry data, private project material, and training candidates so AI-assisted work stays inspectable and permission-bound.
Teams need the source trail without losing control of the source material.
VexASI workflows are built around explicit data boundaries: what can be inspected, what can be stored, what can be shown in a deliverable, and what cannot be reused.
Public sources stay traceable
VexASI Signaling uses public evidence such as company pages, job postings, announcements, procurement clues, and other public sources. Each usable record keeps the quote, URL, observed date, and confidence context visible.
Private source material stays bounded
Project files, technical documents, client material, and review packets are handled only for the scoped workflow. They are not public examples, marketing assets, or reusable training records unless explicit approval exists.
Training use requires approval
Private review material is not promoted into model training data by default. Lessons can become generic operating guidance only after private identifiers, source text, screenshots, and project-specific details are removed.
Inquiry and Contact Data
The public contact page is a static intake surface. Draft text can be remembered in the visitor's browser so the form is easier to finish, but nothing is submitted to a VexASI server by the page itself. The form opens an email draft and the visitor decides what to send.
Inquiry details are used to understand the requested workflow, service lane, evidence sources, urgency, and next action. Do not send confidential project files through the public contact form.
Client and Project Material
Scoped Use
Source material is used to produce the scoped workflow output, such as a signal record, review queue, issue log, source register, or advisory review package.
Review First
AI-assisted outputs are not treated as final just because a model produced them. Review status, evidence support, and boundary language matter before use.
No Public Reuse
Private examples, project names, client-identifying content, source PDFs, screenshots, and derived confidential text are not published as public examples without approval.
No Default Training
Private work does not become training data by default. Any training-candidate path must be explicit, clean-room, and approval-bound.
Need a workflow with stricter data boundaries?
Scope the source types, review gates, retention expectations, and handoff constraints before building the AI workflow.