How Cipres works

Cipres lets your marketing team build audience segments visually — using both traditional data and the unstructured text your users produce every day. Here's everything you need to know.

What makes Cipres different

Traditional segmentation tools only work with structured attributes: plan type, country, revenue. That's useful, but it misses what your users are actually saying.

Cipres combines both. You can segment on classic attributes and on the meaning behind user-generated text — feedback, support tickets, NPS responses, reviews, surveys, chat messages.

Traditional workflow

  1. 1. Marketing defines what they need
  2. 2. Files a ticket to data / engineering
  3. 3. Data team writes SQL queries
  4. 4. Back and forth until it's right
  5. 5. Repeat for every change

Slow, dependent, limited to structured data.

With Cipres

  1. 1. Marketing builds pipelines visually
  2. 2. Combines attributes, text meaning, and AI signals
  3. 3. Tests with sample data, iterates freely
  4. 4. Publishes when ready
  5. 5. Engineering just sends data via API

Autonomous, visual, works with any data type.

Four ways to define a segment

Every segment in Cipres is built from conditions. You can mix and match these freely in the visual editor.

Semantic

Meaning-based matching

Describe what you're looking for in plain language. Cipres uses AI to match users whose text carries that meaning — even if they use completely different words.

Example: "Users who express frustration with the onboarding process"
Extracted

Automatically detected signals

Cipres automatically analyzes text and extracts structured signals you can filter on. These are detected for every user, with no extra setup.

Sentiment Topics Churn risk Price sensitivity Intentions Trends
Attribute

Classic data filters

Filter on any structured field your team sends — plan type, country, signup date, revenue, or any custom key-value pair.

Example: plan = "pro" AND country = "US"
Profile

Behavioral patterns over time

Segment on cumulative metrics that evolve with every interaction — how a user's sentiment has trended, how often they write, their overall churn risk score. These build automatically from every message analyzed.

Avg sentiment Sentiment trend Churn risk score Message frequency Top topics Intentions
Example: churn_risk_score > 0.7 AND sentiment_trend = "declining"

The workflow

Cipres separates who builds the segmentation logic from who sends the data. Both teams work independently.

1

Upload sample data

Start with a CSV or sample dataset. This is what your team uses to design and test pipelines. No real production data needed at this stage.

2

Build pipelines visually

Use the drag-and-drop editor to create segmentation pipelines. Add conditions — semantic, extracted, or attribute-based — and connect them into the logic you need. Test against your sample data until you're satisfied.

3

Publish as a versioned release

When a pipeline is ready, publish it. Cipres creates a versioned snapshot, so you can always roll back or compare changes over time.

4

Engineering sends live data

Your dev team connects your product to Cipres via a simple REST API. When real user data arrives, the published pipeline runs automatically and segments are applied in real time.

5

Act on your segments

Set up webhooks to get notified when users enter or leave a segment. Push updates to your CRM, email platform, Slack — or any tool in your stack.

Ready to try it?

Cipres is in private early access. Request an invitation to get started.