Roy Swisa on $DJCO

Summary

• Roy Swisa — after research on DJCO got a consulting gig with the company and was involved in that week’s shareholder meeting — argues that Journal Technologies (JTI) has the characteristics of software he sees as more protected from AI disruption. With “Anthropic killing or annihilating sector by sector,” his rule is “the more niche you go, the more vertical you go… the more protected”: ripping out a court case-management system is a “heart transplant,” and JTI is growing ~20% with ~$13M of 2025 free cash flow against a ~$150M implied EV inside a ~$650M market cap.
• The sum-of-parts lands almost exactly at today’s price, and Andrew doesn’t let that slide. The ~$500M portfolio nets to roughly $330M after taxes and a $20M margin loan; Roy values the core tech at $250M, Andrew at $215M — Andrew’s blend gets ~$516/share versus a $500 stock: “this sounds interesting, but it doesn’t seem like there’s alpha here.” Roy’s answer is a re-rate, Veeva-style: “they did not fundamentally improve themselves, but the way that the market saw them really changed.”
• Andrew’s structural objection: post-Munger, this is “a stock portfolio with a business attached” run by people who own almost nothing. The CEO holds roughly 600 shares, the board essentially none, and even if the tech business 3×s in five years the stock isn’t a double because the ~$500M portfolio subsumes it. He nominates DJCO as “the company most likely to have a Sanborn Map situation applied to it” — a tender-out of the appreciated portfolio.
• Roy’s differentiator is proof, not narrative, that the service→product turnaround is real. An ex-data scientist, he used AI to access and analyze siloed public state RFPs and found Orange County material with milestones and budget management replacing the old “pay me when you are satisfied” model (“zero alignment of interest… I don’t even know how it could exist”). “There are words and there are words” — and here the evidence matched the CEO’s words.
• The moat is local-government inertia plus decision processes that can be less price-sensitive. Buyers make this purchase “once in their life,” upgrading is all downside for admins (“is it time to get off Windows 98?”), RFP data shows cost weighted only 20–30% and less important than reliability, and Andrew’s hypothetical grant-funded New Orleans PD budget makes a $500 vibe-coded rival unattractive against a $20K/year five-year grant: “We don’t care. We got 20,000. We got to spend it.” Roy’s correction: “They care… but they don’t care as much.”
• Roy’s AI taxonomy: UI-layer SaaS (Monday, Wix) faces build-vs-buy price wars, but proprietary-transaction-data-plus-compliance companies (JTI, Tyler, Cellebrite) are “worth more.” A cousin can vibe-code a court system in 72 hours, but “it’s good for today, it’s not backward-compatible” — legacy ticket logic, certifications, and required references block entry. He also hedges the doom: “we killed so many companies so quickly… maybe we are both in the same echo chamber.”
• The craft segment: on niche names “the real people are not there” on GLG/Tegus, while Andrew wonders whether on-network experts dial back their criticism. Roy’s edge is off-network sourcing and trust-building — “that’s an art… a skill that is disappearing and that’s an edge” — including his test of asking the CFO questions just to see if the CEO interrupts: “I’m actually asking, do you trust the CFO?”

Deep dive

1. A consultant’s-eye view of Daily Journal’s three legs

• Unusual disclosure up front: Roy’s independent research on DJCO “got me a consulting gig with them,” including involvement in that week’s shareholder meeting — the podcast was deliberately delayed because Roy was in the middle of that work (recorded Friday, Feb 27, stock ~$500/share, ~$650M cap).
• The business as Roy maps it: Journal Technologies, the court case-management system — “the thing that grew, the thing that is interesting”; the legacy legal newspaper for California/Arizona; and the equity portfolio, about $493M at the time, against about a $600M market cap when the discussion started. “An interesting name, an interesting market cap, and a very interesting EV.”
• Roy’s meta-framing for why the name earned his hours: “we are investors, we allocate money, but the better way to look into that — we’re time allocators.” Andrew’s riff on the cruelty of the job: hundreds of hours of work can return negative — he knows people who “put the wrong share count in Excel… this is a screaming buy… dude, if you use the right share count you’re dead.”

2. The thesis: regulated vertical software is where Roy sees more AI protection

• The week’s context sharpened the pitch: “Anthropic was killing or annihilating sector by sector. And the more niche you go, the more vertical you go, the more regulatory-heavy… the more protected.” Businesses fused to the customer’s core operations can’t be swapped easily — “think about it like a heart transplant” — and those, he argues, “will be future compounders.”
• He’d run the same lens on Cellebrite before: “it’s almost the same trick. Highly regulated environment. Switching the business model.” JTI is another example of the pattern.

3. Words versus words: proving the turnaround in the RFPs

• The old JTI commercial model staggered him: clients paid when they were happy. “Pay me when you are satisfied… This is no way to make business. You have zero alignment of interest. I don’t even know how it could exist.” The new CEO — whose name is rendered inconsistently in the transcript — inherited “really hard marsh land” and is moving the company from service-oriented to product-oriented.
• Rather than trust management’s narrative, Roy — a former data scientist — used AI to tap the siloed, state-by-state public RFPs for court systems and prosecutors, then studied why JTI wins and loses. Finding Orange County RFP material with milestones and budget-management emphasis was the tell: “I see the change. I believe the CEO is changing the company. He’s not only saying that… there are words and there are words.”
• Andrew’s seconding, from scar tissue: he’s lost count of CEOs professing “we are so shareholder focused” the day before announcing a deal that drops the stock 50% — “what happened to that shareholder focus?”

4. The sum-of-parts lands at roughly today’s price — so where’s the alpha?

• Roy’s math at a $650M cap: liquidating the ~$500M portfolio nets ~$330M after taxes and the $20M margin loan; the traditional business gets “a big nice X”; JTI is valued at ~$150M implied EV against ~$13M of 2025 free cash flow and ~20% growth — “extremely undervalued compared to their growth.” Competitive map: Tyler sits federal-to-state, but court management is only 15% of its revenue and not its focus; Equivant is “quite patchy,” and every new state means new regulation and compliance — a current ticket may still need to run on 2022–2023 logic, so backward compatibility blocks new entrants.
• Andrew’s arithmetic pushback: securities are worth ~$335/share pre-tax ($275 post-tax); using Roy’s $250M for the core (Andrew had $215M) gets a ~$700M fair value, ~$516/share against a $500 stock. “It doesn’t seem like there’s alpha here.”
• Roy’s rebuttal is a re-rate case: “not all software companies are the same,” and specialized vertical software can get repriced as a class — “look at Veeva right now. They did not fundamentally improve themselves. But the way that the market saw them really changed.” If growth and margins continue, “the conservative 250 is worth a little bit more… it’s a totally different game.”

5. The post-Munger portfolio: war chest or accountability sinkhole?

• Andrew’s core objection: this is “a stock portfolio with a business attached,” and the portfolio Charlie ran — which Andrew described as having grown from about $130M to about $500M, with $340M of gains — is now post-Munger. Is it still concentrated in what Charlie liked two years ago? Who pulls the trigger now? And the incentives are thin: “the CEO doesn’t own any stock — I think he owns 600 shares,” the board essentially none. “I get the downside of an equity portfolio inside a corporate wrapper with no longer… one of the best stock pickers of all time running it. Now it’s just kind of a group of guys.”
• Roy concedes it’s “a real downside… a real debate point” but reframes: the money isn’t there to be “a mini Berkshire” — according to Charlie and the current CEO, it exists to help JTI grow via M&A, buying growth, and talent, and Roy argues that it makes JTI hard to swallow without also swallowing the war chest: “You need to bring a lot of guns to the game.” Actions he’ll credit: margin debt has decreased, and the shareholder meeting featured “a real Q&A, similar to Berkshire” — questions about selling the building and the cash, with answers. Still: “sometimes you need to swallow the fish with some skeletons.”
• Andrew’s kicker on structure: even if JTI 3×s in five years, “the stock as a whole is not even a double… the equity portfolio just so subsumes the tech business.” His historical rhyme: Buffett’s Sanborn Map — an investment portfolio bigger than the market cap and a board whose insurance members owned no stock. Buffett, Andrew thinks, got them to tender out the investment. “DJCO might be the company that I think is most likely to have a Sanborn Map situation applied to it” — potentially involving highly appreciated Bank of America stock, a swap, or retiring shares.

6. Local-government inertia is the moat — and grant money can reduce price sensitivity

• Andrew’s picture of the buyer, which Roy calls “not so remote from the truth”: part of the nuclear-weapons program still running on floppy disks, offices asking “is it time to get off of Windows 98, or can we let this go another year?” Upgrading “has no upside and all downside” for an admin with no bonus at stake; Roy adds these buyers make the purchase “maybe once in their life” and outsource judgment to Gartner and consultants.
• On price: Andrew’s hypothetical — imagine a New Orleans PD running Cellebrite on a $20K/year, five-year grant and meeting a $500 vibe-coded rival: “We don’t care. We got 20,000. We got to spend it. If we spend 500, we’re just giving 19,500 back.” Roy’s calibrated pushback: “They care, okay?… but they don’t care as much” — reliability comes first, cost is “second or third.”
• The evidence is in the documents, not opinions: “look at the RFP. Cost is 20% to 30%. Hence, it’s not as important as reliability… that’s a data point, you know? It’s not an opinion.” Andrew’s clincher on the JTI scenario: mission-critical, regulatory-nightmare software costing ~$50K a year against $50M of billings — “I’m probably not going to go vibe-code that.”

7. Not all software companies are the same: Roy’s AI-disruption taxonomy

• First, the hedge on the consensus: “we are a little bit in an echo chamber. We killed so many companies so quickly… the terminal value right now is dead. It takes time for the decision-maker to make a change” — procurement processes move slower than the narrative.
• The loser pool: mainly-UI companies — he names Monday — where agents may do the job and buyers gain build-vs-buy leverage. Andrew extends it: enterprises can now threaten “we’ll hire two engineers and vibe-code our own monday.com” to force price down, and Wix-like products face a commodity war like Instagram ads — early adopters won until “everybody sees that and there’s just no barrier.”
• The protected pool: companies sitting on proprietary transaction data created inside the product, reachable through their API, wrapped in compliance. The vibe-code cousin who builds a court system in 72 hours fails the test: “it’s good for today, but it’s not backward-compatible” — and certifications plus required references from previous deployments gate government buyers regardless of balance-sheet size. These, Roy says, are “worth more.”

8. The disappearing art of primary research

• Roy’s honest read on expert networks: they are “amazing,” but the expert is “highly incentivized to give you this 15-minute block… reuse the same information over and over. There is alpha there — but in these niche spaces, sometimes the real people are not there.” For Cellebrite, he says you could not source a CIA agent through GLG when they last checked, though that may have changed: “you need to go, you need to earn trust” — with compliance front of mind. From Columbia’s Value Investing Program, the lesson that stuck: “knowing how to tap into people… there is no replacement for that. That’s an art… a skill that is disappearing, and that’s an edge.”
• Andrew’s corroboration: on an off-network call a friend’s expert “blasted the company CEO” in terms no on-network expert ever matches — like comedians who “don’t drop any of the F bombs” on network TV — and his broader view that “alpha is increasingly outside the filings” (the sweating CEO at a conference), weighed against the real cost, which is time.
• Roy’s field technique, worth stealing: in a management meeting he asked the six-months-in CFO direct questions solely to watch whether the CEO let him finish. “I’m actually asking, do you trust the CFO?… And then I can see — I’m wasting my time here. There’s zero trust in the management team.” Andrew, laughing: “Do not interrupt Roy.”
• On Gartner’s survival: for once-in-a-lifetime decisions, sleepy buyers still need hand-holding. Roy’s humility about the AI discussion: “we think the whole world has the same access to this information. But maybe we are like this 3% to 4%, maybe 1%” — with Andrew’s caveat that “maybe we are both in the same echo chamber.”