← Back to Portfolio Portfolio Demo · Simulated NLP

GenAI Customer Support Analyzer

Paste a support transcript and explore a production-style analysis dashboard. Under the hood, this demo uses rule-based NLP (tokenization, keywords, heuristics) to simulate what a GenAI pipeline might surface—no live API calls.

Transcript input

Batch mode (all samples)
Running NLP pipeline
Tokenizing & classifying…

Analysis dashboard

Batch analysis

Scenario Category Sentiment Emotion Resolution Priority

Category distribution

Sentiment distribution

Trending topics & volume

Word cloud from combined sample transcripts (frequency-weighted). Issue volume below uses simulated weekly aggregates for illustration.

Simulated issue volume (12 weeks)

How it works

This demo mirrors a typical text analytics stack: lightweight preprocessing, lexical signals, and templated generation—the same stages often wrapped by an LLM in production.

Tokenization
Lowercase, split on boundaries, drop noise tokens.
Keywords
Score topical n-grams; surface key phrases as tags.
Classification
Lexicon overlap picks Billing, Tech, Feature, Account, Help.
Sentiment
Positive vs. negative word counts → score −1..+1.
Summarization
Template slots filled from category, emotion, resolution.