# Quickstart

> Sign up, load one file or a sample, say what you want to know, get your first read from the agent, then run a backfill so your file gets Dimensions.

You can go from sign-up to a first answer without talking to anyone. Plans start at $299 a month with no contract on monthly plans ([pricing](https://www.dimensionlabs.io/pricing), checked October 2026).

## 1. Create your workspace

[Sign up](https://next.dimensionlabs.io/first-read) and give your organization a name. That name is how your workspace appears in the app.

## 2. Load one file, or use a sample

Drop in one file of customer language: a reviews export, survey responses, a chat or ticket export. If you'd rather look around first, pick one of the sample datasets offered at sign-up.

Your file needs:

- **A format of** .csv, .tsv or .jsonl. There's no size limit: Dimension Labs splits a large file for you.
- **Five required columns:** `text`, `incoming`, `userId`, `sessionId` and `dimensionlabs_timestamp` (a 13-digit time in milliseconds). A `conversationId` is recommended. See [Upload a file](/data/upload-a-file) for what each one holds, and the [conversation template](/templates/conversation-template.csv) for an example.
- **No personal information as an id.** Don't use email addresses, phone numbers or names as `userId` or `sessionId`.

Run the file through the [file checker](/data/check-a-file) first. It finds the format problems that cause most failed uploads, the [converter](/data/convert-a-file) fixes most of them, and your file never leaves your computer.

A file of conversations goes under **Data Sources > Unstructured data**, and gets Dimensions once you run a backfill (step 5). A file of business data (revenue, orders, accounts) goes under **Data Sources > Structured data** instead: it's queryable within about a minute and is never enriched. See [Choose your path](/data) if you're not sure which you have.

## 3. Say what you want to know

Write the question in your own words, for example *"Which stores get the most one-star reviews, and why?"* or *"What are customers asking our chatbot that it can't answer?"*. This becomes the agent's first task and the goal of your project.


## 4. Read your first answer

You land in a chat with the agent already at work. It lays out a short plan, inspects what your file contains, runs the queries, and writes back with a headline, the numbers behind it, quotes from the underlying records, and what to look at next.

This first answer reads the raw text and the columns in your file, because your file has no Dimensions yet. The agent says so in its answer. When it doesn't have enough data to answer well, it says that too instead of guessing. [Your first chat](/agent/first-session) walks through an example answer and an example refusal.


## 5. Run a backfill to get Dimensions

An uploaded file is not enriched on its own. To give it Dimensions:

1. Open the **Backfill** tab.
2. Start a backfill for your new source. The standard Dimensions (**System Prompts**) are available in every workspace, so you can backfill them before you design any prompts of your own.
3. Wait for it to finish. A backfill typically shows up within the hour, depending on how much data is being enriched.
4. Ask your question again. The agent now answers from the Dimensions.


See [Backfill history](/dimensions/backfill) and [When enrichment runs](/dimensions/when-enrichment-runs).

## What happens next

- **Data you connect later.** A live integration is enriched automatically every day. Every file you upload needs its own backfill.
- **Keep adding data.** See [Choose your path](/data).
- **Get better answers.** See [Ask good questions](/agent/asking-good-questions).
- **Build a report or anything you'll repeat.** See [Quick answers and analysis plans](/agent/modes).
- **Get help.** The **Help** button in the app's left navigation opens these docs. For anything else, email support@dimensionlabs.io.