November 20, 2026 · 7 min read

Delegate, Describe, Discern, Diligence: The 4Ds on One Listing Launch

Anthropic teaches AI fluency as four skills. Here they are on one sample listing launch: what to hand to AI, how to ask, what you must check yourself, and the step before you post.

Delegate, Describe, Discern, Diligence: The 4Ds on One Listing Launch

Anthropic's 4D Framework fits a listing launch in four moves. Delegate the first drafts of the ten posts and the emails. Describe the listing with facts only. Discern price, beds, baths, square footage, agent name and fair housing words by hand. Show diligence before anything posts: an MLS check, one approver, and care about what you paste in.

Anthropic's AI Fluency course names those skills Delegation, Description, Discernment and Diligence. They are written for anyone using AI. This post puts them on one made-up listing so you can see where each skill earns its place.

The working tool is our free One Listing, Ten Posts Prompt. It turns one fact sheet into ten pieces, from coming soon to sold. The prompt carries part of the 4Ds. The rest is your job.

The 4Ds in plain words

The framework comes from Claude Academy's course "AI Fluency: Framework and foundations." Its lesson on the 4D Framework says it "consists of four core competencies (the 4Ds)." We read them this way for a brokerage:

  • Delegation. Decide which work goes to AI and which stays with a person.

  • Description. Tell the AI what you need, clearly: the facts allowed, the words banned, the shape wanted.

  • Discernment. Judge what comes back with a critical eye. Is it true, in your voice, and allowed?

  • Diligence. Use AI responsibly: what you share, what you sign off on, who answers for the post.

The course also names three ways of working with AI: automation, augmentation and agency. A listing launch uses the first two.

The sample listing

Everything below is a made-up sample. The slips are examples we wrote to show the check, not the output of one run.

The fact sheet the agent pastes in:

  • 4 bedrooms, 3 baths, 2,150 square feet (source: the MLS sheet)

  • Listed at $749,000

  • Listing agent: Taylor Brooks

  • Three features to show first: the wraparound porch, the kitchen renovated in 2022, the detached workshop

  • Open house Saturday, 1 to 3 PM

  • Best photo: the porch at dusk

  • One named park nearby, supplied by the agent

  • Networks: Instagram and Facebook

That is all the AI may know. Every claim must trace to one of those lines.

Delegation: hand off the drafts, keep the decisions

Decide the split before you open the chat.

Hand to AI: the first draft of all ten pieces (coming soon through sold, including a Reel script and a Story), a caption trimmed for each network, the property email, and a first pass at the calendar.

Keep with the agent and team lead: the list price and any change to it; which three features lead, since the agent walked the house; the call on when a post goes out; and the sold post, because its days on market and outcome do not exist yet.

In the course's terms, the ten drafts are mostly automation: you specify the task and the AI carries it out. The hooks and the Reel script are closer to augmentation. You read options, pick one and ask for a change. Neither includes setting the price.

Your agents are already doing this work. The National Association of Realtors' 2026 REALTORS Technology Report says "The top AI use case among REALTORS® is writing listing descriptions, cited by 75% of AI users." If you have not decided who checks it, each agent is deciding alone.

Description: the prompt does half the job

A vague request, "write a just listed post for my new listing," gets a vague post with invented detail. The prompt asks the way a good brief does. Its ground rules read: "Facts only from me. Never invent a feature, a number, a distance or a claim." It tells the AI to describe the home, never who it is for or who lives nearby, and it bans hype words such as "nestled," "boasts" and "dream home."

It also asks for what a fact sheet leaves out: the three features to show first, the open house time, the best photo and what it shows, a named park, and the networks. Those inputs are why the porch post is about the porch and not about "outdoor living."

What you add as the agent:

  1. Paste the MLS sheet, not a summary. Retyping from memory is how 4 bedrooms becomes 5.

  2. Name the voice, or paste two past captions.

  3. Tell it to leave the price out of coming soon and the sold post's numbers blank.

Discernment: what you check by hand

The prompt runs its own check for unproven facts, steering, off-limits topics, hype and duplicates. Run it, read the flags, then read every piece against the sheet with a pen. The AI marked its own work, so a person must too.

The slips we look for, on the sample listing:

  • Price. The just listed post says "$750K." The sheet says $749,000. A rounded price is a number the seller never approved.

  • Beds. A feature post says "five bedrooms." The sheet says 4.

  • Square footage. The Story says "over 2,200 square feet." The sheet says 2,150.

  • Agent name. The email is signed "Tyler Brooks." The listing agent is Taylor Brooks. Check this twice. Every reader sees it.

  • Fair housing words. The porch caption says "perfect for a growing family." That is who the home suits. Describe what is there: a porch deep enough for a table that seats eight.

  • An area claim. The Reel says the park is "a short walk." The sheet gave a name, not a distance.

Check numbers against the MLS record, not the pasted sheet, because the sheet is where an error can start. Then read the ten pieces in reverse so repeated lines stand out.

Captions deserve this care. Metricool's 2026 State of AI in Social Media Study found that 86% use AI to write posts, captions and copy, while 84% use it to generate content ideas. The most common use is also the most common place for an unchecked line to go out.

Diligence: the step before you post

Diligence covers what happens around the draft.

What goes into the tool. Paste facts that are on the public listing. Leave out the seller's name, their reason for selling, showing codes and anything you would not say at the open house. If your office has no rule, write one line today: what may go in an AI tool and what stays out.

Who answers for the post. Name one approver. In a solo practice it is the agent, reading the set the day before launch. In a team it is the team lead, before anything is scheduled. They write "checked" and the date on the note that holds the ten pieces.

What the rules say. Your MLS, state and brokerage have advertising rules, including disclosure lines. The AI does not know your office's wording. Add it.

What the seller sees. Send the seller the first three posts before launch. A seller who has seen the price in the post has confirmed it.

Written rules are rare. Metricool surveyed more than 700 social media professionals and reported "Only 12% of respondents have a clearly defined AI policy."

By the numbers

  • 75%: AI users who cite writing listing descriptions as their top use, per the NAR 2026 REALTORS Technology Report. The page also says 56% of REALTORS use AI to create social media posts and 52% to draft emails and follow-up.

  • 86% and 84%: respondents in the Metricool study who use AI to write posts, captions and copy, and to generate content ideas.

  • 12%: those respondents with a clearly defined AI policy.

Agents already use AI on launch copy, and few offices have written rules.

How Urban Marketing Edge handles this

We use AI to move faster on drafts and variations. Strategy and taste stay human. On our plans that include listing marketing, every listing gets the same launch set, written from the MLS facts, designed in the office brand, and sent through the post approval system so nothing goes out unseen. The team lead approves once, and the posts land on the office calendar. That is the 4Ds as a standing process: AI drafts, a person checks, one person signs off.

To run it yourself first, the Listing Launch Checklist covers the timing from coming soon to sold, and the One Listing, Ten Posts Prompt covers the writing. Both are free. See how the system ran for Rocks Realty in our case studies.

Questions brokers ask

What is the 4D Framework?

It is the core of Anthropic's AI Fluency course: Delegation, Description, Discernment and Diligence. In plain terms: what to hand off, how to ask, how to judge the result, and how to use AI responsibly. The course says the four skills apply across automation, augmentation and agency.

Which part of a listing launch should AI never do?

Setting the price, choosing which features lead, deciding to publish, and signing off on the numbers. AI can draft around those once a person has made the call.

How do I check AI listing copy for mistakes?

Compare every number, name, date and year against the MLS record, not the pasted sheet. Then read each piece for words that describe who the home suits instead of what it has. Read the set in reverse so repeated lines stand out.

What should agents never paste into an AI tool?

Seller names, reasons for selling, showing codes, lockbox details and anything not on the public listing.

Who should approve the posts?

One named person: the agent in a solo practice, the team lead or managing broker in a team or brokerage. One approver is faster than three and leaves a clear owner.

Book a 30-minute strategy call

If your agents use AI on listings and nobody owns the check, close that gap first. Bring one recent launch, the posts and the MLS sheet, and we will compare them line by line and map the simplest approval step for your office. No pitch deck, no pressure.

Book a 30-minute strategy call and we will build the plan around your listings. To see how we work month to month first, start with our plans.


Written by

Casey McClintic, founder of Urban Marketing Edge. Ten years of real estate marketing for brokerages, luxury teams and agents. About Casey

Next step

Ready for a brand that shows up every month?

30 minutes. We map the simplest plan for your content, themes and monthly cadence. No pitch deck.

Book a 30-minute strategy call
Book 30 minutes