> ## Documentation Index
> Fetch the complete documentation index at: https://support.affinity.co/llms.txt
> Use this file to discover all available pages before exploring further.

# Pipeline catch-up with AI Chat

> Use AI Chat to find stale deals and bring your pipeline current in a few minutes.

Every Monday you'd normally scroll your pipeline, eyeball which deals have gone quiet, open each to check the last touch, and fix stages by hand. This is the same routine — Affinity AI just does the finding, and applies the updates you approve.

AI Chat is a conversational way to work your CRM: ask in natural language, get answers grounded in your own data with cited sources, and — at this launch — have it take **confirm-before-execute** actions like updating a field or pipeline stage. Nothing is written until you approve it.

**Time:** \~5 minutes · **You'll need:** a pipeline list with a stage field, and edit access to that list.

## Before you start

* **Plan:** AI Chat is available on **Scale, Advanced, and Enterprise**.
* **It only sees what you see.** Affinity AI works inside your own permissions.
* **Where it lives:** open AI Chat from the left navigation (also on your home page). Chat and **Ascend** (agents) are separate surfaces.

## Step 1 — Ask what's gone quiet

```text theme={null}
Which deals in my pipeline haven't we contacted in 30 days?
```

Affinity AI returns the stale deals with each one's last-contact date and current stage, and ends with the data it reviewed. If you get “showing the first 20 of 34,” ask **“show more.”** Narrow the same way you'd ask a colleague: *“Just the deals in Active Diligence that have gone quiet.”*

## Step 2 — Get the context on one

```text theme={null}
What's the latest on Lumira Diagnostics?
```

You'll get the last interaction, recent notes, and who's engaging — enough to decide the next move without opening the profile.

## Step 3 — Act on it

Because changing a field or stage is a **confirm-before-execute** action, Affinity AI shows exactly what it will write and waits for approval:

```text theme={null}
Lumira Diagnostics — Stage
  Monitoring → Active Diligence   (visible to your team)
Update stage?   [ Confirm ]  [ Cancel ]
```

Confirm, and you get a receipt linking to the record. If you ask for several changes at once, Affinity AI confirms them **one at a time**.

## Step 4 — Clear the rest

Work down the list the same way — *“Move Greenfield Systems to Closed – Passed.”* Each request is its own confirmation.

## Step 5 — Review the suggestions waiting from your meetings

If you also use the **Data Update agent**, the deals you met with that week may already have **Suggested Field Updates** waiting — so part of “catching up” is clearing that queue. After an external meeting, the agent reads the meeting note and proposes field changes. Each suggestion shows the **field name**, its **current value** and the **proposed new value**, and a **link to the source note** (**View Meeting Note**).

These don't live in the chat — they collect in the agent's **Agent Activity** queue. Open it and **approve all**, **approve some**, or **approve none**. At this launch you **accept or reject** each suggestion as-is; editing a proposed value before approving is coming later. See [Review & approve an agent's suggestions](/s/article/Review-and-approve-agent-suggestions).

## Tips

* **Start broad, then narrow.** One wide question plus a follow-up beats many small ones.
* **Use your real stage names.** Affinity AI updates to your list's actual stage values.
* **Batch your asks, review one at a time.**

## Common questions

* **Will Affinity AI change my pipeline on its own?** No — updating a field or stage is confirm-before-execute. The Suggested Field Updates queue is the same: proposals until you approve them.
* **What actions can it take?** At launch: update a field value or pipeline stage, add to a list, create a note, create a reminder. Suggested Field Updates from meeting notes are reviewed in the Data Update agent's queue (Step 5), not as a chat action.

***

*Affinity AI can make mistakes. Review each suggestion before approving it.*


## Related topics

- [Tutorial 12: Advanced Pipeline Management](/s/article/Tutorial-12-advanced-pipeline-management.md)
- [Take actions with AI Chat](/s/article/Take-actions-with-AI-Chat.md)
- [Affinity AI Chat Overview](/s/article/Affinity-AI-chat-overview.md)
- [Prompts for Affinity AI Chat](/s/article/Prompts-for-Affinity-AI-Chat.md)
- [Meeting Preparation with AI Chat](/s/article/Meeting-preparation-with-AI-chat.md)
- [What AI Chat can do & how to trust it](/s/article/What-AI-Chat-can-do.md)
- [Prompts for Affinity MCP (Claude, ChatGPT & other tools)](/s/article/Prompts-for-Affinity-MCP.md)
- [Getting Started with AI Chat](/s/article/Getting-started-with-AI-chat.md)
- [Warm Intros in AI Chat and MCP](/s/article/Warm-Intros-in-AI-Chat-and-MCP.md)