> ## Documentation Index
> Fetch the complete documentation index at: https://docs.getvector.me/llms.txt
> Use this file to discover all available pages before exploring further.
# How Vector works
> See how the parts of Vector connect: from one line about your product to a catalog of your market's features, and from there to the work you plan.
Vector is built as a chain. Each part reads what the one before it produced, so most of the work isn't yours: you describe your product, let AI find your competitors, and collect a dossier on each of them. Everything after that — the catalog of features your market offers, its structure, the analysis on top of it — Vector assembles from what those first steps gathered.
This page is the map of that chain. Read it once, and every screen you open later has an obvious place in it.
## Your part and Vector's part
| Step | What you do | What Vector does |
| ------------------- | ---------------------------------------------------------------- | ------------------------------------------------------------------------------ |
| Your product | Write one line about what you're building, and approve the draft | Drafts the full description and uses it as context for everything that follows |
| Competitors | Sort them into the short list and the long list | Finds competitors that match your description |
| Competitor dossiers | Start the research, one competitor at a time or several at once | Researches each one and pulls out the features they offer |
| Features in domain | Review candidates, and mark which features you already have | Collects every feature your market offers into one catalog |
| Product map | Accept or reject the structure AI suggests | Proposes clusters and areas, and analyzes each of them |
| Delivery | Plan a delivery cycle for each feature you'll build | Runs each cycle through stages toward dev-ready artifacts |
| Monitoring | Choose which competitor sources to watch | Checks them on a schedule and tells you what changed |
## The chain, step by step
On **My product**, write one line about what you're building and click **Next — AI writes the description**. AI drafts a full description from it; you correct whatever's off and approve it. That description is the context every AI step afterwards runs against — competitor search, research, comparison. Optional research steps on the same screen can enrich it further. See [Describe your product](/getting-started/describe-your-product).
**✦ AI find competitors** searches for products that match your description. You sort what it finds into two lists: the **Short list** holds the direct competitors you always watch, the **Long list** the wider field you keep on record. You can move a competitor between them at any time. See [Add competitors](/competitors/add-competitors).
A competitor's [dossier](/competitors/competitor-dossier) holds everything Vector learns about them: a deep teardown of their product, their public sources, pricing, help center, changelog, and the features they offer. Fill it in one competitor at a time, or select several in the list and pick an action — **🔬 Deep research**, **✦ Feature scan**, **💲 Pricing** and others — to run it for all of them at once.
This is the step that feeds everything after it. As dossiers fill in, the features Vector finds in them build up into a catalog of what your market offers.
**Features** → **Features in domain** in the left menu is that [catalog](/features/features-in-domain): every capability at least one competitor offers, whether or not you have it yourself. It works with two kinds of entries:
* A **feature** has a [dossier](/features/feature-dossier) of its own, which draws in everything known about it: how each competitor implements it, what their users say, and where they seem to be heading with it.
* A **candidate** is what a source said that isn't a feature yet. Nothing research finds goes straight into the catalog: it waits in [Candidates](/features/candidates) until you decide — attach it to an existing feature, turn it into a new one, or dismiss it as noise.
Every feature also carries your own status, from **not assessed** to **we have it**, so the catalog shows where you're ahead and where you're behind.
Once the catalog grows, a flat list stops being readable. The [product map](/features/product-map) groups features into **clusters** — broad domains, such as Talent Acquisition or Payroll and Benefits in an HR product — and narrower **areas** inside them. A feature can sit in an area or directly in its cluster.
AI helps with the structure: **Suggest structure (AI)** proposes new clusters with areas inside, and you accept or discard each one. When many features pile up directly in one cluster, the map suggests grouping them into areas. Every cluster and area has its own page with an analysis: how many of its features you have, where your gaps are, and how much of it each competitor covers.
## Monitoring alongside the chain
Monitoring doesn't wait for the chain to finish. In a competitor's dossier, the **Resources** tab lists their public sources — release notes, changelog, blog, pricing, help center, social accounts — and each source has a **Monitoring** switch.
Turn it on, and Vector checks that source on a schedule. The first check only records a baseline. After that, whenever something changes, Vector analyzes the change, sends a notification to your **Inbox**, and writes it to the log in that competitor's dossier. **Monitoring** in the left menu shows the whole picture: how many competitors and sources are watched, and the latest changes across all of them. See [Put a competitor under monitoring](/monitoring/set-up-monitoring).
## Glossary
| Term | What it means |
| --------------------- | --------------------------------------------------------------------------------------------------------------------------------------- |
| Description | The approved description of your product on **My product**. Every AI step reads it |
| Short list, Long list | Your two lists of competitors: the direct ones you always watch, and the wider field |
| Dossier | A page that collects everything known about one thing. Vector keeps one for every competitor, every feature, and every cluster and area |
| Feature | A capability offered in your market, with its own dossier |
| Candidate | What a source said that isn't a feature yet: you attach it to a feature, turn it into a new one, or dismiss it |
| Cluster, area | The two levels of the product map: a broad domain, and a narrower part of it |
| Delivery cycle | One pass of work on one feature — v1, v2 — that moves through the stages of Delivery preparation |
| Us | Your own status for a feature, from **not assessed** to **we have it**. The catalog and the feature's dossier show the same value |
## Where to go next
Why Vector exists, and what you need before you start.
The checklist that walks you through the chain in order.