Day three of the daily show. If you missed the first two — I did quick daily deep dives on Earnings Week and FIS Earnings/Anthropic vs Pentagon
Today the thread continues: legacy infrastructure across financial services is stuck in a pre-agentic world. And the companies that recognize this — and the ones that don’t — are diverging fast.
Three stories today.
Workday: When AI Becomes a Cost Center Before It’s a Revenue Engine
Big credit to App Economy Insights and their How They Make Money series on Substack. If you play in the public markets, I’d recommend them over any of the more mainstream trade analysis outlets. They did a deep dive on Workday’s earnings — go read the full numbers there (it is a premium read).
Here’s what stood out to me as a company builder and investor.
Workday’s subscription business is still the key revenue driver, but here’s the shift: all of their new AI products are consumption-based. And I think that’s where everything is heading. One of my mentors — guy who sold his company for a little over $300 million in the data and information services business — gave me this advice years ago: he never charged a subscription, never charged a platform fee, never did a monthly minimum. Everything was transactional. Simpler for accounting, simpler for everything. Pre-AI advice that’s becoming the default model in an AI-first world, no matter what Marc Benioff says.
Workday has been on an acquisition spree — Sana (conversational AI), Pipedream (agent connectivity), Paradox (AI recruiting) — all within the last 18 months. And the results are showing: AI revenue has grown to $400 million. 75% of their engineering teams are using AI coding assistants. 50% of new code is AI-generated. Engineering output up 22%.
But here’s the most interesting thing their founder said on the earnings call — everyone thinks AI is going to kill companies like Workday. His response? Anthropic is a customer. Google is a customer. He name-dropped top AI model companies that are using Workday for payroll processing and HR services. The companies building AI still need someone to run their payroll.
Still, their stock is getting crushed. Up 2% today, but the market isn’t rewarding the spend yet. It’s a company I track closely — we have HR customers at ModernTax, so the payroll world matters to us.
PayPal: $1.75 Trillion in Transactions, $43 Billion Market Cap
This is the story I teased at the end of yesterday’s show. It’s gone full throttle today — every major outlet is covering it, and I’ve been in group chats all day with people talking about it.
PayPal. A company we grew up on. I’m 38 — you grew up on PayPal, you grew up using it. Most people still use Venmo. But the numbers tell a brutal story: up 15% over the past five days on acquisition speculation, but down 18% YTD, down 36% over the past year, and down 82% over five years. $43 billion market cap.
The new CEO coming in is Enrique Lores from HP. One of my friends on the PayPal team says Lores runs a performance-based playbook, which makes sense coming from hardware. But it’s an interesting hire — you’d think they would have gone after someone more startup-centric or emerging for this role instead of a 35-year hardware veteran.
There was a funny moment this week — Sheel Mohnot, part influencer, part creator, full-time VC at Better Tomorrow Ventures, posted a spoof of Apple acquiring PayPal. Totally fake. Got me — I clicked the link. My friend said he’d actually love for Apple to acquire PayPal. But the real frontrunner, if moves are made, is Stripe.
Stripe just yesterday announced a $159 billion tender offer backed by Thrive, Coatue, and a16z, and has expressed preliminary interest in acquiring PayPal — all of it or parts of it. That caused the 7% pop. Bloomberg, CNBC, Reuters, Payments Dive — everyone’s reporting it.
John Collison basically said PayPal has had “a tough time over the past few years and the landscape has changed quite a bit with Apple Pay and Google Pay.” Nobody’s denying the interest.
Here’s why this is massive: Stripe processed about $1.9 trillion last year. PayPal reportedly processed about $1.75 trillion. Combined, that would be the largest payment network on Earth outside the card networks. And this is the administration that would say that’s not a monopoly. So the timing is interesting.
Product Spotlight: Reducto and the J-Mail Story
Last segment — there was a great piece on The Verge about a company I’ve been tracking called Reducto. YC 2024 batch.
This is relevant to me because I’ve been in three different tax businesses — billing and tax documents live in PDFs, and parsing PDFs is still genuinely hard. Reducto has cornered a market in making that work.
Here’s the story that got my attention. When the DOJ released millions of pages of Epstein documents under the Epstein Files Transparency Act, officials claimed making them searchable was “technically infeasible.” Two engineers — Riley Walz (who just announced he’s joining OpenAI) and Luke Igel of Kino AI — built J-Mail in five hours. It’s a Gmail-clone interface where you can browse Epstein’s actual inbox. I went into it yesterday — you can type in anyone’s name and see their emails with Epstein. It’s like you’re logged into his account. Fascinating and deeply unsettling.
They used Reducto to parse the poorly scanned PDFs, handwritten notes, redacted documents, and complex layouts. Plus Anthropic’s Claude for rapid feature development. The site has served 500 million page views to 50 million unique visitors. They’ve expanded to J-Flights (flight logs on a Google Flights interface), J-Drive (photos), J-Amazon (Amazon orders), and J-Wiki (Wikipedia-style reference).
When the DOJ dropped three million more documents in January, dozens of engineers — many from Reducto — gathered at Igel’s house in SF and processed the entire dump. Reducto took about three days to crunch through it all.
I actually pass by their office when I go play my men’s basketball game on Thursday nights. Very low-key spot in SoMa. They have Airtable, Scale, Vanta, and a Fortune 10 company as customers. I’m assuming they’re going to get a lot more after this project.
This is what I love — a company going after simple things. There are so many complex, shiny problems people chase, but parsing PDFs is still not easy. And tools that make the simplest things easier are often the most valuable. I’d love to get the Reducto guys on for an interview. Shout out to that team and that project.
What I am Watching
Before I sign off — what do you think? What would happen if Apple bought PayPal? If Stripe closes this? These are companies we all know going through very interesting times. Reply and tell me your take — best answers get featured on the next show.
If you like this content, check out the ModernTax Daily Beat on Substack where we cover these topics framed around tax and the SBA lending industry — targeted to our core customers but same energy, same depth. Different audience, same me.
I see all the opens, all the clicks, all the replies. If something resonates, hit the comment line. Share this with people — people who know me, people who don’t. I’m in San Francisco, always love to connect, grab coffee, grab lunch, hang out. That’s how this community grows.
And if you’re building something around financial infrastructure, payments, AI document processing, tax — any unsexy thing — reply to this email. I’d love to hear what you’re building. See if there’s opportunities to partner, invest, angel, whatever.
Talk tomorrow.
— Matt
