# Product | Rippit

> Rippit is the growth data platform built to maximize the insights your people and agents can produce. Each part of it is a building block, from a secure foundation up to work that keeps running.

Canonical: https://www.rippit.com/product

Product

# Build your edge, block by block

Rippit is the growth data platform built to maximize the insights your people and agents can produce. Each part of it is a building block. Scroll to stack all twenty, from a secure foundation up to work that keeps running.

The blueprint. Scroll to lay the first block.

## Security

Everything else stands on this.

### Security and data handling

Rippit acts only within the signed-in user’s permissions and the connections an admin has authorized. The same rules hold whether you use it directly, through an AI assistant, or in a pipeline that runs with nobody watching.

-   Rippit does not train any AI models. Your data is used for inference only
-   Data integrations are read-only, and connected MCP servers start read-only
-   Credentials are encrypted at rest with AES-256-GCM

[Security and data handling in the docs](https://docs.rippit.com/security)

## Connect your data

Every conversation and the business data around it, stored in one place as tables.

### Data integrations

A data integration is a read-only connection to a system that holds your customer conversations. Rippit pulls them in on a schedule and stores them as tables you can query, build reports from, and run pipelines on.

-   15 sources, including Zendesk, Intercom, Salesforce, Gong and Slack
-   The first sync backfills up to 90 days by default, and further on several sources
-   No storage limit. Hot and cold storage plus compute is $1 per GB per month

[Data integrations in the docs](https://docs.rippit.com/plg/data-integrations)

### Salesforce and Snowflake tables

Bring in the records your conversations never carry: accounts, opportunities, orders and usage. Each Salesforce object and each Snowflake table becomes a Rippit table you can query and join to conversations.

-   Salesforce: pick up to 50 objects at a time, and cases become conversations
-   Snowflake: up to 100 tables per connection, synced hourly or daily
-   Your source data is never changed

[Salesforce and Snowflake tables in the docs](https://docs.rippit.com/plg/supported-integrations#snowflake)

### Bulk API and files

No built-in integration for your system? Send its data with the Bulk API, or upload a CSV. It lands as tables you can query and run pipelines on, the same as integrated data.

-   Conversations as JSON Lines, tables as Parquet or JSON Lines
-   Sending a record again updates it instead of duplicating it
-   `rippit upload` sends a CSV with no API token

[Bulk API and files in the docs](https://docs.rippit.com/plg/data-integrations#bring-data-from-another-system)

### MCP connectors

Rippit’s in-app agent is an MCP client. Add MCP connectors and it can reach Slack, Notion, Salesforce, Snowflake or any MCP server, in the same conversation where it is analyzing your tickets.

-   An admin chooses the exact tools the agent can use
-   Every connection starts read-only
-   Pipelines use the same connections to deliver results

[MCP connectors in the docs](https://docs.rippit.com/agentic-chat/overview)

## Generate insights

Ask a question in plain language and keep what Rippit builds to answer it.

### Ask Rippit questions

Chat is where most work in Rippit starts. Ask in plain language and Rippit picks the data, runs the analysis, and answers with the work attached. It reads whole conversations, so you get what happened in the thread, not how someone tagged it.

-   Follow-ups keep their context: “now split that by channel”
-   Drop in images, PDFs, text files and CSVs
-   Can search the web and reach your connected MCP servers

[Ask Rippit questions in the docs](https://docs.rippit.com/plg/chat/asking-questions)

### Artifacts you keep

Ask for a breakdown, a report or a table and Rippit builds it as a real artifact in your workspace rather than a wall of text. Charts, worksheets, canvases and pipelines appear beside the conversation and stay afterward.

-   Add a column that classifies every conversation by theme, urgency or root cause
-   Sort, filter and chart that column later
-   The side panel lists the sources an answer was based on

[Artifacts you keep in the docs](https://docs.rippit.com/plg/chat/asking-questions#what-comes-back)

## Build pipelines

The docs call these the building blocks: steps you combine so any analysis keeps running.

### Data source

Every pipeline begins with data that is already in Rippit, so there is nothing to export or prepare: integrated conversations, business data, or another pipeline’s output.

-   A row filter limits it to the rows that matter
-   Rows that don’t match are never processed, so they cost nothing
-   A pipeline can also start from a schedule alone

[Data source in the docs](https://docs.rippit.com/pipelines/steps#data-sources)

### AI column

Creates the column your data is missing. A model reads each row and the full conversation behind it, applies the judgment you describe, and answers in a typed column you can filter, count and chart.

-   Reads every message in order, not a subject line or a summary
-   Columns defined together share one model call per row
-   Choice, true or false, number, text and list outputs

[AI column in the docs](https://docs.rippit.com/pipelines/steps#enrichments)

### Formula

Calculates a column from other columns on the same row: a duration from two timestamps, a flag from a score. Use one whenever the answer is arithmetic or logic.

-   Exact, with no AI cost
-   The same formula language as worksheets
-   Recalculates when a column it reads changes

[Formula in the docs](https://docs.rippit.com/pipelines/steps#formulas)

### Lookup

A conversation knows who wrote in, but not what they pay you or who owns the account. A lookup adds columns from another table’s matching rows, so every later step can use them.

-   Matches on a shared value, such as an account ID or an email
-   Returns one row, or several up to a limit you choose
-   Matching uses no AI

[Lookup in the docs](https://docs.rippit.com/pipelines/steps#lookups)

### Rollup

Turns many rows into the few you want to act on. On a schedule, it folds a window of rows into one row per account, agent, topic or period.

-   Runs hourly, daily, weekly or monthly
-   Exact counts, averages and totals
-   A threshold keeps an alert silent when there is nothing to report

[Rollup in the docs](https://docs.rippit.com/pipelines/steps#rollups)

### Agent run

Puts an agent to work on each rolled-up row. Where an AI column answers one question, an agent can investigate: it reads the row, uses tools, and writes a result.

-   One agent session per row
-   Can read and aggregate your tables and run external research
-   Up to 20 rows at a time by default

[Agent run in the docs](https://docs.rippit.com/pipelines/steps#agent-runs)

### External research

Some answers aren’t in your data at all. An agent can search the web, find people, and ask AI models such as ChatGPT a question, and save every finding as a row.

-   Web search uses Exa unless you ask for another provider
-   Work email and phone lookups run only when you ask
-   Agent search records each model’s answer and citations

[External research in the docs](https://docs.rippit.com/pipelines/research)

### Action

An insight nobody sees changes nothing. An action puts each result where the work happens: a Slack post, an email, a new ticket, a CRM update, or any tool on a connected MCP server.

-   Runs once per row, so one row per account means one message per account
-   No double sends: each row is recorded as soon as it is sent
-   Can hold messages until a delivery time you choose

[Action in the docs](https://docs.rippit.com/pipelines/steps#actions)

### Build, test, and run

You build a pipeline by describing it, in Rippit chat or from your own AI assistant. Rippit drafts the steps, and nothing runs until you have tested it, checked the cost and put it live.

-   A draft processes no rows and sends nothing
-   A test runs every step once on a sample, 200 rows by default
-   Changes to a live pipeline stay drafts until you make them live

[Build, test, and run in the docs](https://docs.rippit.com/pipelines/build-and-run)

## Connect your AI agent

Use every block beneath this one from the assistant or terminal you already work in.

### Rippit MCP server

Connects your AI assistant to Rippit. Rippit hosts and manages the server, so there is nothing to deploy, and your assistant can only reach what the signed-in user could already reach.

-   One endpoint: `mcp.rippit.com/mcp`
-   An admin adds it once, then each user signs in individually
-   Tools to query, classify at scale and build pipelines

[Rippit MCP server in the docs](https://docs.rippit.com/mcp-server/overview)

### Supported AI agents

Rippit works with any client that speaks MCP over Streamable HTTP with OAuth. The docs have setup steps for each one that has been tested.

-   Claude, ChatGPT, Codex, Copilot Studio, Gemini, Cursor, Notion and more
-   Editors and terminals need no admin step
-   An unlisted client can register itself with just the server URL

[Supported AI agents in the docs](https://docs.rippit.com/supported-clients)

### Connect data through your assistant

Authorize a data provider and start its first import from a conversation with your assistant, then carry on with your question once a table is available.

-   Zendesk, Intercom, Salesforce, Slack and Gong
-   Passwords and keys stay out of the conversation
-   Authorization links work only for the person who asked

[Connect data through your assistant in the docs](https://docs.rippit.com/mcp-server/connect-data)

### Rippit CLI

Connect integrations, upload files and query your data from a terminal or an assistant that can run commands. An MCP connection is not required.

-   One-line install: `curl -fsSL https://rippit.sh | sh`
-   `rippit start` creates a temporary account you claim later
-   Add AI columns to a worksheet with `rippit data enrich`

[Rippit CLI in the docs](https://docs.rippit.com/cli)

### Your Edge

Rippit reads the full content of every conversation, not a sample or just tags, to produce more accurate insights for your business. With pipelines built on top, any analysis becomes work that keeps running.

[Try Rippit free](https://app.rippit.com/signup)

[or request demo](https://www.rippit.com/request-demo)

## Getting started from an AI agent

There are two ways to connect. Either works.

- MCP server: `https://mcp.rippit.com/mcp` (Streamable HTTP, OAuth). The person signs in or signs up in the browser that opens. No client ID, secret or API key is needed.
- CLI: install with `curl -fsSL https://rippit.sh | sh`, then get an account one of two ways:
  - `rippit login`: the person signs in or signs up in a browser.
  - `rippit start`: creates a temporary account with no email or browser, for a person with no Rippit account. `rippit claim --wait` lets them keep it; an unclaimed account is deleted after 14 days.

Then bring in data: connect a provider such as Zendesk, Intercom, Gong or Salesforce, or upload a CSV.

Full steps: https://www.rippit.com/setup.md