AI Market Research: How an Apartment Operator Has Claude Read 500 Podcasts Every Morning

What He Built: AI Market Research That Reads 500 Podcasts Before the Workday Starts

Barrett Linburg is co-founder of Savoy, which builds, owns and manages Texas apartments, with 8,000 units under management according to his X profile. His Oct 30 post starts from one idea: "A problem you are working on today has already been solved. The person who solved it is explaining how they did it 34 minutes into a podcast you don't have time to listen to."

So he had Claude do the listening. "Every morning, Claude reads any newly published episode from about 500 shows we track. In one week in October, that was about 365 episodes, start to finish." That week, "the first read flagged 197 short passages for deeper analysis. The second read cut that to 12 worth acting on." His team calls the result the "Podcast Brief."

What the AI Market Research Brief Turned Up for an Apartment Company

Linburg gives three examples from the same stretch. Each one connects to something his company was already working on.

AI Market Research Find 1: Half the Phone Leads Were Junk

The marketing VP at a large apartment operator said on an industry show that her ad vendor was celebrating a 50-70% jump in phone leads. Her team checked the calls against their CRM: "About 54 of every 100 were spam, vendors, and bankers."

Claude knew Savoy was building a cost-per-lease report by channel that counted calls with nothing to screen out junk. "A channel that looks cheap (per inquiry) might just be a channel full of spam." About 90 seconds of her conversation came back written up with who said it, why she's credible, where it sits in the episode, why it applies to the report, who owns it and what to check. "Nobody had to listen to anything."

AI Market Research Find 2: A Tax Credit Deadline the Notes Had Wrong

Savoy's own notes said a tax credit it had used was still live. A cost segregation CEO said on a podcast that the 179D energy deduction is closed to buildings that start construction after June 30, 2026. Linburg's team believed him but checked the IRS guidance anyway. 45L, the credit they had actually used, has the same cutoff.

AI Market Research Find 3: How Owners Judge a Property Manager

Two operators on investor podcasts told owners to judge a property manager by one document: the before-and-after T12 from a recent takeover. Savoy manages apartments for other owners, so it can now prepare for that question "or even highlight our performance proactively based on that metric."

The first time Linburg shared the brief with his executives, one of them forwarded an item to their trade association's general counsel.

How to Build AI Market Research From Podcasts: His Six Steps

AI Market Research Step 1: List What You Care About

"Write down what you care about, and go wider than feels right." His list: apartment management, construction, fundraising, opportunity zones, wealth management, local politics, zoning, liquidity events and interesting people.

AI Market Research Step 2: Build the Show List

Podscan, which he says tracks about 600,000 active podcasts and carries the transcripts, was the source. He started with the 117 shows he already followed. Claude found similar shows and searched transcripts across every line of the business, reaching about 7,000 candidates. It pulled recent episodes and guests on about 1,100 of them and rejected two-thirds. With his starting shows, that came to about 500. His rule: "Judge on the guests and the content, not the audience size."

AI Market Research Step 3: Write the Context Document

"This is the whole trick." The document says who the company is, what it believes, what it's working on, what it has already done and who owns what. It's fed by Linburg's daily work in Claude and weekly updates from his executives, "so it moves when we move." If the company is chasing a niche problem that month, the agents hunt for it across all 500 shows.

AI Market Research Step 4: Read Every Episode in Full

"Every new episode, every morning, in full. No keyword filters." The best material, he says, is a passing remark in the middle of an episode about something else. Sonnet does the first read; Opus goes deep on what gets flagged.

AI Market Research Step 5: Check the Work

"The second read goes back to the full transcript, tests each claim against our own notes and traces anything big to a primary source. It keeps what the speaker said separate from what we concluded."

AI Market Research Step 6: Correct the Context Document

When the team tells it something it got wrong about how the company operates, "that goes into the context document right away."

What Went Wrong in the First AI Market Research Runs

Linburg's review of the first run: "Surfaced some really good stuff but didn't go deep enough to be useful." He rebuilt it to pull the actual passages, with timestamps, names and numbers.

• It got facts wrong that were already in Savoy's own notes. It pinned a Houston default on the wrong sponsor.
• Transcripts have errors. One machine transcript had the buyer and seller reversed.
• The check matters. In one big run, the second read caught 79 errors from the first.
• Don't let it judge shows too early. The agents tried to rank shows after two episodes and called some of them noise. He stopped that: a national construction show usually has nothing for Savoy, then an episode three weeks later has "a two minute blurb that is solid gold."

Why This AI Market Research Setup Works

• The context document does the filtering. There are no keyword filters. Claude decides what matters by comparing each passage to what the company is working on, so the same 500 shows produce different finds as priorities change.
• Two reads, two jobs. A first read casts a wide net (197 passages). A deeper second read checks against the transcript, the company's notes and a primary source, and keeps 12.
• Every item is ready to act on. Each write-up names the speaker, why they're credible, the timestamp, why it applies, who owns it and what to verify.
• A person still checks. On the tax credit, the team confirmed the guest's claim against the IRS guidance before acting on it.

Linburg's advice for anyone starting: "The agents are only as good as what they know about you. So start with one page about your business. The podcasts are just the input." For another way to keep an operator's knowledge in one place, see our article on a multifamily acquisitions knowledge base.

How to Try It: A Starter AI Market Research Prompt

Linburg didn't publish his prompts. The one below is a CRE AI Studio add-on based on the steps in his post. Start small: one page of context and the transcripts of this week's episodes from a handful of shows, pasted or attached. Scale to a scheduled run once the brief is useful.

[CRE AI Studio prompt, written for this article]
You are reading podcast transcripts for my company's morning market research brief.

Context document (who we are, what we believe, what we're working on, what we've already done, who owns what):
[PASTE YOUR ONE-PAGE CONTEXT DOCUMENT]

Topics we care about: [LIST, e.g. apartment management, construction, fundraising, zoning, local politics]

Transcripts: [ATTACH OR PASTE NEW EPISODES, with show name, episode title and date]

First read:
1. Read every transcript in full. Don't skip episodes that seem off-topic; the useful part is often a passing remark.
2. Flag every short passage that bears on something in the context document.

Second read, for each flagged passage:
3. Go back to the full transcript and quote the passage exactly, with the timestamp.
4. Check the claim against the context document. If it conflicts with our notes, say so.
5. If it's a fact we would act on (a law, a deadline, a number), name the primary source we should check. Don't state it as confirmed.
6. Keep what the speaker said separate from what you think it means for us.

Output: only the items worth acting on. For each one: who said it, why they're credible, show, episode and timestamp, the quote, why it applies to us, who on our team owns it, and what to check before acting. If nothing clears the bar, say so.Copy prompt

When your team spots a mistake, add the correction to the context document before the next run. That's Linburg's step 6, and it's what keeps the brief accurate over time.

What to Check Before You Rely on AI Market Research From Podcasts

• Primary sources. Treat a guest's claim as a lead. Linburg's team checked the 179D claim against IRS guidance before using it.
• Transcript errors. Machine transcripts can swap names or roles, as in his buyer-and-seller example. Check the audio or a second source for anything important.
• Speaker incentives. The cost segregation CEO sells that work. Note who benefits from a claim.
• Your own notes. The system got facts wrong that were already in Savoy's notes. A second read that checks against them is part of the design.
• What goes in the context document. In Linburg's words, it "holds your deals, your people and your plans." Decide which tools and accounts it's allowed in.
• Cost and volume. Hundreds of full transcripts a week is a lot of reading. Start with fewer shows and grow the list once the brief earns its keep.