Chief Technology Officer · 2024–present
Citrus Analytics — Fusion
AI-powered survey analytics platform
THE CHALLENGE
Hours of manual analysis. No path to scale.
Market research agencies spend hours on manual analysis. Citrus Analytics needed a platform where non-data-scientists could run advanced statistical methods with full transparency and confidence.
The problem wasn't just automation — it was automation that could be trusted. Results had to be reproducible, methods had to be explainable, and the process had to be accessible to analysts without deep statistical training.
THE SOLUTION
AI agents that analyze survey data — without exposing respondents.
An AI-powered platform with autonomous AI agents that analyze uploaded SPSS survey data. The system runs regression, segmentation, clustering, and latent class analysis. Martin invented the method for automatically analyzing SPSS data while maintaining the privacy boundary.
Every analysis plan is shown to the user before execution. Every result is backed by documented statistical methods. The AI operates on structured analytical methods rather than free-form generation, eliminating the risk of hallucination.
Agent architecture
Autonomous agents plan and execute multi-step analyses using a structured task graph. Each agent is responsible for a defined analytical method and reports results in a standardized format.
SPSS analysis pipeline
Uploaded SPSS files are parsed to extract variable metadata, value labels, and response distributions. Statistical summaries — never individual records — are what the AI operates on.
Privacy model
Individual respondent data never leaves the server. LLMs receive only aggregate statistics and variable definitions. A novel design constraint that shaped the entire architecture.
Full transparency
Every analysis plan is shown before execution. By constraining the AI to a defined set of statistical operations, the platform delivers automation speed with expert credibility.
KEY NUMBERS
3hrs → 30min
Analysis time reduction
Zero
Respondent data shared with LLMs
Full
Analytical transparency
TECH STACK
REFERENCES