How agents can use AI to triage their pipeline fast

Most deals you lose don’t fail in a single moment.

They slip through small handoffs.

A lender stops replying.

An inspection email gets missed.

A buyer says “we’re thinking.”

No one sets a next step.

A few weeks later, you realize the timeline moved and you weren’t driving it.

The problem usually isn’t effort.

It’s triage.

If you can’t answer “what’s at risk this week?” in under 60 seconds, you’re working from incomplete information.

This post gives you a simple workflow you can run quickly to create two outputs:

a Hot / At-risk / Stale table
a Top 10 actions list for today (with deadlines)

And you’ll end with draft messages so you’re not starting from scratch.

Why deals slip (and why you don’t notice)

Real estate is coordination and accountability across dozens of open loops: buyers, sellers, lenders, title, escrow, inspectors, and appraisers.

When any one of them slows down, the fix usually isn’t better wording.

It’s clear ownership and a dated next step.

The bottleneck isn’t writing messages.

It’s consistently deciding:

what’s Hot
what’s At-risk
what’s Stale
what to do next

The 3 buckets: Hot / At-risk / Stale

Hot

Use Hot when any of these are true:

action is due today or within 48 hours
you got a high-intent signal (showing request, offer/counter, inspection/appraisal scheduling)
you’re waiting on you and the next step is obvious

At-risk

Use At-risk when momentum is decaying:

you’re waiting on a third party and there’s no next checkpoint
there’s no owner and no explicit next step date
it’s been long enough that silence becomes the default outcome

Stale

Use Stale when:

there’s been no meaningful movement
the thread needs either helpful re-engagement or an explicit close-out decision

Suggested starting thresholds (adjust later): Hot = 0–2 days, At-risk = 3–14 days, Stale = 14+ days.

Why AI is really helpful for this

The hard part isn’t drafting the follow-up.

It’s maintaining accurate state:

what the thread is actually about
who’s waiting on who
what the next step is
what’s at risk if it slips

AI helps because it can:

extract structure from messy context (threads, notes, half-finished plans)
drive decisions by turning every open loop into a bucket + next step + by-when
reduce the time cost of follow-through by drafting messages once the decisions are clear

The exact outputs you’re building (one table + Top 10)

You want one table that’s easy to scan.

Use these columns:

Person/Deal
Stage (buyer / seller / under contract)
Last touch (date)
Who emailed last (me/them)
Waiting on (me/them/third party)
Risk trigger (why it’s at risk)
Recommended next step (one line)
By-when (deadline)
Bucket (Hot / At-risk / Stale)

Then generate:

a Top 10 actions for today (priority order)

Example: what the workflow looks like

Here’s a realistic example of the output you’re aiming for.

Pipeline triage (example)

User

Help me quickly triage my pipeline.

From my active deal threads from the last 30 days, build a triage table with:
- Person/Deal
- Stage (buyer / seller / under contract)
- Last touch (date)
- Who emailed last (me/them)
- Waiting on (me/them/third party)
- Risk trigger (why it’s at risk)
- Recommended next step (one line)
- By-when (deadline)
- Bucket (Hot/At-risk/Stale)

Then:
1) Give me my Top actions for today (priority order)
2) Draft follow-ups for the most urgent items

Assistant

## 🧭 Triage table (last 30 days)

| Person/Deal | Stage | Last touch | Who emailed last | Waiting on | Risk trigger | Recommended next step | By-when | Bucket |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Linh N - Condo buyer | Buyer | Apr 7 | Them | Me | Showing request pending; high intent | Send 2 showing windows + confirm preferred neighborhood | Apr 8 (11:00 AM) | Hot |
| Evan R - Under contract | Under contract | Apr 6 | Them | Third party | Inspection report ETA unclear; repair window tight | Get inspection report ETA + schedule 15-min repair call | Apr 8 (3:00 PM) | Hot |
| Priya P - Under contract | Under contract | Apr 3 | Them | Third party | Lender requested docs; no confirmation received | Confirm docs received + ask for clear-to-close ETA | Apr 9 (12:00 PM) | At-risk |
| Marco A - Buyer search | Buyer | Apr 1 | Me | Them | No reply after sending options; momentum decaying | Send 2 fresh listings + ask a yes/no next-step question | Apr 9 (5:30 PM) | At-risk |
| Mei C - Listing consult | Seller | Mar 20 | Me | Them | No response on next step; likely dead unless re-engaged | Send close-the-loop note with two options + deadline | Apr 10 (2:00 PM) | Stale |

## ✅ Top actions for today (priority order)
1) **Linh N:** send showing windows + confirm area (Hot) - **by 11:00 AM**
2) **Evan R:** confirm inspection report ETA (Hot) - **by 3:00 PM**
3) **Priya P:** confirm lender doc receipt + timeline (At-risk) - **by noon tomorrow**
4) **Marco A:** re-engage with 2 options + a yes/no question (At-risk) - **by 5:30 PM**
5) **Mei C:** close-the-loop message (Stale) - **by Thu 2:00 PM**

## ✉️ Draft follow-ups (top 3 actions)

**1) Linh N: showing windows (text)**
Hi Linh, I can do showings either today 4:30-6:30 or tomorrow 10:00-12:00. Which window works better? Also, are you leaning more toward “quiet/residential” or “walkable/near shops” so I send the best 2-3 options?

**2) Evan R: inspection report ETA (text to inspector)**
Hi James, quick check: when should we expect the inspection report for Evan Reed’s place on Baker Street today? If you can share an ETA, I’ll coordinate a quick review call so we stay inside the repair window.

**3) Priya P: doc receipt + timeline (email to lender)**
Hi Sloan - quick check: did you receive the updated documents for Priya Patel’s file?

If yes, what’s the next milestone and your best estimate for clear-to-close? A simple ETA is perfect so we can keep everyone aligned.

Weekly cadence (so it sticks)

This works best when you run it consistently.

Here’s a simple cadence:

Monday (weekly pass): generate triage table + Top 10
Daily (quick scan): clear Hot first
Friday (cleanup): stale cleanup (re-engage or close out)

The goal isn’t more activity.

It’s earlier intervention on deals that are slipping.