How sciOAISIS generates answers

Not every sciOAISIS response is created the same way. Some commands return direct, structured system data with no AI involved. Other commands first gather data procedurally and then send that result to AI for summarization, explanation, or drafting.

Why this matters

Users should know when they are seeing raw system output versus interpreted output.
Staff can judge whether precision or summarization is more appropriate for the task.
It builds trust because the system is transparent about what happened behind the scenes.

Type 1

Procedural response only

A command runs against one or more systems, returns structured data, and displays it directly. No AI transforms the answer.

Example: :amp info studentid
API request + database lookup → formatted output to the user

Best when accuracy and exact values matter most.
Useful for IDs, statuses, dates, enrollments, contact fields, and direct records.
The wording is deterministic because the app is formatting the response itself.

Type 2

Procedural retrieval + AI use

A command first gathers data through code, then passes that result into AI so the AI can summarize, explain, reformat, or highlight what matters.

Example: :amp events today
Query events → send event results to AI → return a readable summary

Best when users want a concise overview instead of raw rows.
Useful for summaries, prioritization, plain-language explanation, and drafted responses.
The source data is procedural, but the final wording is AI-generated.

The clearest way to make this understood

Every tool or command should visibly declare its response mode before the answer is shown.

Recommended UI labels

Mode: Procedural
Mode: Procedural + AI summary
Mode: AI only

Example banner:
Mode: Procedural Direct system result from AMP and local database. No AI used.

Example banner:
Mode: Procedural + AI summary Data was retrieved directly, then summarized by AI for readability.

Recommended help-page language

Add a permanent “How responses are generated” section in Help.
Link to this page from Help, Tools, or the command reference.
Document each colon command with one of the response modes above.

Command reference style:
:amp info STUDENTID — Procedural
:amp events today — Procedural + AI summary

Operational rule of thumb

If the system is returning exact fields, counts, or records, label it Procedural.
If the system retrieved exact data but the final text was rewritten, condensed, or interpreted, label it Procedural + AI summary.
If the answer is mostly generated from user language without a deterministic backend pull, label it AI.

Good trust pattern:
Show the mode label first, then the answer, and when relevant include the source system underneath.