> ## Documentation Index
> Fetch the complete documentation index at: https://docs.anycrm.anyreach.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Deals

> Sales opportunities, the pipeline they move through, and the AI-generated intelligence that rides alongside them

A deal is a sales opportunity — a chance to close business with an account, worth some amount, moving through a pipeline of stages toward a win or a loss. It's the central object most of AnyCRM's other features attach to: activities get logged against it, contacts join it as a buying committee, and the platform's AI features generate intelligence about it.

## Pipeline stages

Every organization defines its own pipeline: a named, ordered sequence of stages (e.g. "Qualification," "Demo," "Proposal," "Negotiation," "Closed Won," "Closed Lost"). Stages carry their own configuration — whether they're a closed/terminal stage, whether they're hidden from normal views, and their display order — and that configuration is deliberately restricted to organization admins, separate from the day-to-day permission to create and manage deals themselves. Moving a deal between stages is a first-class action with its own side effects: it's logged to stage history, can trigger notifications, and feeds the forecasting and win-probability calibration described below.

## Forecasting fields

A deal carries both a **rep-entered** expected close date and value, and several **computed** fields:

* **Health score** — a rule-based score computed from deal signals (recency of activity, stage velocity, and similar factors), refreshed on demand.
* **Win probability** — track this against actual outcomes over time via the win-probability calibration view, which buckets closed deals by their recorded probability and compares it to the real win rate in each bucket.
* **Forecast confidence** — a sales-methodology category (commit, best case, upside, pipeline, and so on) distinct from win probability, used for forecast rollups.

## AI-generated deal intelligence

Several families of generated content attach to a deal, each addressing a different question a rep or manager might ask:

* **Competitive intelligence** — who else is this deal being contested against, and how do we compare? Ranges from a cheap name-only detection pass to a full generated battlecard (SWOT, feature comparison, competitive positioning, recommended plays).
* **Sales intel** — who's actually involved in the buying decision (a stakeholder/buying-committee analysis), and what happened on recent calls (call intelligence extracted from meeting transcripts).
* **War room** — a synthesized "situation room" view of a deal's current state, plus saved what-if scenarios for exploring how a change (a discount, a timeline shift) might affect the outcome.
* **Deal brief** — an executive-readable summary of the deal.
* **Deal actions** — suggested next steps, either generated or logged manually as plain tasks.

Some of this generation is genuinely AI (LLM-backed), and some of it — despite the "generate" naming — is deterministic, rule-based scoring. The [API reference](/api-reference/deals/get-deal) calls out which is which per endpoint, since it materially affects how to think about latency, cost, and reproducibility.

## Relationships

* A deal belongs to at most one **account** (the company).
* A deal can link many **contacts**, each with a role describing their part in the buying decision.
* A deal can have **activities** logged against it directly.

## Technical reference

For the full endpoint-by-endpoint reference — this is the largest resource in the API, covering pipeline configuration, core CRUD, linking, and all of the AI-generated deliverables above — start with [Get a Deal](/api-reference/deals/get-deal) and [List Deals](/api-reference/deals/list-deals).