Current core product · Research-first GEO analysis
GEO Positioning System: understand how AI describes your brand.
GPS is REALI’s GEO research system. Repeated questions across AI models help teams examine answers, visible citations, recurring narratives and tone—and identify what deserves a closer look.
Your team’s questions set the direction. The system helps run tests and organize material; professionals confirm the conditions, interpret findings and decide what to do or track next.
Research outputs
What can the research help me understand?
Findings should be traceable to original answers and test conditions. The materials and report format are agreed before the work begins.
- Answers
What AI says
Read the original answers to see how your brand is mentioned, described and compared.
- Sources
What AI cites
Examine the pages and domains visibly cited in answers. These citations do not reveal every source behind a model’s response.
- Narratives
What keeps coming up
Compare recurring keywords, topics and narratives to investigate patterns and changes.
- Tone
How the answer is framed
Review answer-level tone classifications and scores under the study’s rules, alongside original wording and human-review status.
Recorded product screens
From a research question to answers you can check.
These frames come from GPS operation recordings: reviewing the brief and questions, checking progress and examining citations. They show different stages and are not one continuous record of a single study.
Recording source and conditions
GPS Copilot 操作(1).mp4 · 01:06
Apple demonstration. Clips were captured under different recording conditions and must not be treated as one research run or customer outcome.
Recording source and conditions
GPS Copilot 操作(1).mp4 · 02:36
Apple demonstration. Clips were captured under different recording conditions and must not be treated as one research run or customer outcome.
Recording source and conditions
GPS Copilot 操作(2).mp4 · 00:10
Apple demonstration. Clips were captured under different recording conditions and must not be treated as one research run or customer outcome.
Recording source and conditions
GPS 完整操作結果-Apple為範例.mp4 · 01:55
Apple demonstration. Clips were captured under different recording conditions and must not be treated as one research run or customer outcome.
The research workflow
How does a brand question become research your team can use?
Repeated tests reveal recurring patterns and differences. We call this “answer distillation” as a way to describe the analysis—not model training or access to a model’s internal reasoning or complete source history.
- 01Human direction
Confirm the questions
Bring the brand, audience and purpose. The team reviews candidate questions and agrees the scope and test conditions.
- 02System execution
Run repeated tests
Test across the agreed models and rounds, retaining answers and visible citations.
- 03System support
Organize differences
Compare sources, narratives and answer-level tone, keeping the basis for calculations and classifications.
- 04Human review
Check and interpret
Professionals check the original answers and brand facts, separating findings, interpretations and unresolved questions.
- 05Human decision
Choose what to track
Select meaningful questions and agree how to retest them over time.
AI executes. PR decides.
Which decisions stay with people?
Research helps a team understand AI answers. Communication choices, improvements and resource decisions still require professional judgment.
- AI · Execution
The system supports
Repeated testing, answer and citation records, and material for comparison.
- Human · Review
Professionals review
Test conditions, original text, classification rules and limits. Tone scores are read under the method stated for each study.
- Human · Judgment
The team decides
Brand facts, priorities, content approval and the questions worth tracking.
Research → Monitoring
How do research and ongoing monitoring work together?
First compare answers and visible citations across models. Then choose important questions to observe over time, checking the model, timing and test conditions when interpreting change.
Research: investigate the question
Explore differences in descriptions, sources and narratives, creating material for review and interpretation.
Monitoring: follow selected questions
Retest an agreed question set to observe changes in answers and cited sources.