Turn AI conversations into
structured intelligence

Extract sentiment, risk signals, intent, and 10+ dimensions from every conversation. Powered by Claude and GPT-4o.

pip install analyxa

$ analyxa analyze support_call.txt --schema support { "sentiment": "negative", "satisfaction_prediction": "dissatisfied", "issue_category": "billing", "risk_signals": ["frustration", "repeat_contact"], "session_outcome": "resolved", "customer_effort_score": "high" } ✓ Analysis complete — 16 fields extracted in 3.2s

“Your AI agents handle thousands of conversations.
But what’s actually happening in them?”

👁

Invisible Patterns

Sentiment trends, recurring issues, and risk signals buried in thousands of unstructured conversations.

⏲

Manual Review Doesn’t Scale

Team leads review 5% of conversations. The other 95% go unanalyzed. Critical patterns are missed.

📈

Zero Structured Data

Conversations go in, nothing actionable comes out. No metrics, no trends, no alerts.

How It Works

One conversation in, structured intelligence out.

Conversation In

File, Redis queue, or API. Any text conversation between a user and an AI agent.

Analyxa Extracts

N dimensions via LLM + YAML schema. Sentiment, intent, risk, entities, and more.

Intelligence Out

Structured JSON + 1,536-dimensional semantic vector. Searchable, filterable, actionable.

Configure What You Extract

YAML schemas define your extraction dimensions. Inherit from universal, add your own fields.

Universal 10 fields

The foundation for any conversation

sentiment topics risk_signals session_outcome user_intent action_items

Support 16 fields

Customer service intelligence

satisfaction_prediction issue_category effort_score escalation_needed resolution_quality

Sales 16 fields

Sales pipeline intelligence

buying_stage objections budget_signals decision_urgency competitive_mentions

Coaching 18 fields

Therapeutic & coaching insights

emotional_valence behavioral_patterns coping_strategies therapeutic_momentum growth_markers

Simple API, Powerful Results

Python, CLI, or full production pipeline.

Python

CLI

Pipeline

from analyxa import analyze

result = analyze(conversation, schema="support")

print(result.fields["sentiment"])          # "negative"
print(result.fields["issue_category"])     # "billing"
print(result.fields["risk_signals"])       # ["frustration"]
print(result.fields["effort_score"])       # "high"
            

$ analyxa analyze call.txt --schema support -o result.json

$ analyxa batch ./calls/ --schema support --output-dir ./results/

$ analyxa search "frustrated customer" --limit 5

# Start infrastructure
$ docker compose up -d     # Redis + Qdrant

# Push conversations to queue
$ analyxa redis push call.txt --schema support

# Process all pending
$ analyxa redis process

# Search by semantic similarity
$ analyxa search "billing complaint" --limit 5
            

Deployment Packages

Open source engine, professional deployment. We set up Analyxa on your server, tuned for your use case.

Standard

$297

Get Analyxa running on your server

Universal schema (10 fields)
Redis + Qdrant infrastructure
CLI + Python API
1 test analysis
3-day delivery

Order on Fiverr →

Turn AI conversations intostructured intelligence