Slow, slow, fast

When I started my job at Epic, I felt like I was slow at everything -- debugging code, writing code, solving issues, you name it. I eventually got to be very fast, one of the top 100 MUMPS coders in the world, I used to joke. How did I get there? My need to provide good customer service (i.e. external motivation), patient mentors, a team willing to help me learn, years of practice, and a willingness to dig a little deeper in the code than most others.

So it goes with a new industry. I once again am slow, slow, slow at almost everything -- DCFs, reading academic papers, reading 10-Ks, assessing start-ups, writing anything. I have to keep telling myself: slow but quality now means fast and quality later. I hope writing here to a pseudo-public audience will grant me the motivation and the practice, however imperfect it may be.

Transpose Health – Annual Letter 2025

It is incredible how much can change in a year. One year ago, I had an unsigned contract; today, I’ve honed our data conversion engine (and run multiple test runs) and am working through a few contracts. A few thoughts from the past year of building and running Transpose Health:

1. Clarity of company’s strengths – It’s funny how it takes the better part of a year to nail down what Transpose Health’s impenetrable strengths are (and how I contribute to them). If customers buy Transpose Health, what they are buying is my expertise and “certainty” in the data conversion process. Yes, this comes from my long tenure at Epic and my computer science background, but I think what sells hospital staff is feeling immediately comforted on an initial sales call.

2. Partnerships are difficult – How trite but how true. Earlier in the year, I parted ways with an initial partner – ultimately, we weren’t aligned on urgency and what building a company/brand meant. I ended up finding another partner who could expand what Transpose Health could offer, while also being aligned with the company’s strengths: Epic expertise, long-term partnerships, trust. Again trite, but we don’t always agree – for example, he’d prefer to charge consulting expenses hourly and pass through cloud expenses, while I’d prefer to bundle it into one lump sum (more customer-friendly, and it forces us to have great processes). It took a little frustration before I realized that these disagreements are ultimately productive – especially because I can trust we’re fighting for the same end goal.

3. A little naivete is okay – I committed to an archiving platform before having one built out (or fully knowing how many layers there are to building a good one); I figured I could AI-assist-code a decent prototype in a few weekends. (Perhaps that is one thing AI is good for: hubris.) This naivete (or alternatively, "salesmanship") led me to commit to building an archiving platform ... which ultimately led me to finding a good partner. Things sometimes (often?) work out, but it feels like good entrepreneurs have to get a little over their skis every now and then.

4. Sales are everything – I’ll write more about this some time, but my old company (Epic) hardly had a sales team; hospitals came to Epic, demanded to use the software, and some were turned away. (Epic would tell customers, “Our product won’t meet your needs right now, but maybe in a few years it’ll be a good fit.”) What I learned in business school and VC is … this is a completely anomaly. Companies pay big bucks to win revenue, … and last year, I had somehow managed to make a sale (and have a few in the pipeline). Transpose Health’s data conversion services are a natural complement to data archiving, allowing for a credible way to expand – and add recurring revenue.

5. Trust is gold – Ultimately, the value proposition of Transpose Health is trust: can hospitals trust us to get the job done well? Can they trust us not to overcharge? Can we trust them to tell us when they messed up? Can they trust the company’s people (me) and the people we hire? While everything else (services and software) will devolve into a commodity, trust will continue to be a differentiator worth paying a premium for.

6. Looking ahead – The new year will bring a new organizational structure as the company expands from data conversion services into data archiving. Our primary goals are to (1) ramp up sales, (2) continue to refine the product to allow for (3) easy onboarding processes. I also want (4) to see our current – and first! – client have a successful go-live. An exciting year to come – I’m just trying to revel in all the exciting growth, and make peace with all the anxiety and uncertainty that comes with any new venture

Mike

Synthetic venture exposure?

 A few observations:

1. Venture capital funds at the highest levels is typically access-constrained,

2. Some large VCs are looking more PE-like (e.g. AI roll-up strategies), 

3. Some large public companies have venture arms (e.g. Google Ventures' $10B portfolio and includes some great companies, like Waymo)

My idea/question: is there a way to create an investable "synthetic venture" structure to capture these public companies' venture exposure?

For example: Google Ventures' revenue is classified under "Other Bets" in the 10-K (2024 revenue of $1.5B vs. advertising revenues of $237.9B) ... could you buy GOOG stock, hedge out the search and cloud businesses, and just have (levered) exposure to Google's venture upside? There may not be a financial instrument that can actually capture this ... but it feels like an interesting idea regardless, especially as many other large companies -- like (famously) FTX -- allocate to their own venture portfolios.

(Side note: a similar question I had a while back was if you could replicate PE exposure with leveraged small cap equities -- which is how I ended up finding Dan Rasmussen's 2015 paper, from which he created Verdad Capital. The general idea was that private equity firms in aggregate don't create value above and beyond the public markets, so why sink money into the illiquid asset class if you can get similar exposures in more liquid marketable securities? It turns out "private equity without the illiquid headaches" was good to start a hedge fund but not great marketing materials over the long run, so Verdad quietly pivoted away from that as their core strategy.)

Weird questions like this are where I think AI shines, so I asked ChatGPT. (This is also an area where asking a derivatives expert is helpful -- a creative financial instrument is unlikely to make a Google search. I phoned a friend in finance, and this type of thing doesn't exist today.) Chat GPT's results, summarized:

1. Ventures typically make up such a small fraction of the company's assets that you'd have to short nearly the entire company. For example, Google had $348B non-"Other Bets" revenue vs. $1.5B "Other Bets" revenue in 2024, so even if you valued "Other Bets" at 50x revenue (and everything else at 10x), "Other Bets" is 2% of Google's value. So, you'd go long GOOG, short 98% of it, and lever that 2% long exposure up -- risky, and likely expensive too.

2. Finding the right basket of things to short would be challenging -- you'd want to short the ads business, Google subscriptions, and Google Cloud. This would be hard to get right.

Two ideas I did like: (1) going long a basket of microcap stocks (conceptually similar to Dan Rasmussen's idea), and (2) a recent July 2025 paper that creates a framework for understanding VC returns as a basket of options (a paper I need to reed more thoroughly). (Other ChatGPT references point to better VC benchmarks (e.g. PME) and using R&D spend and Bloomberg innovation indices to give a "venture tilt."

So: still more questions than answers, and more things to read and digest now. But I'm sure this idea will sit on the backburner, waiting for the right time to be realized as an investable strategy.

Creative financial structures (Part 1 of ?)

As someone newer to finance, I've found one the most interesting -- and challenging -- things is to create a new, creative financial structure. In general, (1) people are risk-averse and complexity-averse, especially when it comes to money, and so (2) financial institutions are good at providing shades of the same exact service with little innovation. Much of the financial system seem to be building and retaining trust, not coming up with cool new ways to build companies or structure deals.

The interesting things happen at the margins, by people who have the political capital and financial savvy to take risks. I think David Swensen and the Yale endowment embody accretive creativity -- and how these novel structures can only last so long. To me, Swensen's real genius was that he had a gut feel of the nuts and bolts of the financial markets, and a willingness (and persuasiveness) to try new things. In his short stint on Wall Street, he helped structure the first swap (something that is now commonplace and thus less profitable today). He was one of the first to see that Yale's endowment portfolio -- a majority stocks, bonds, and cash in the 80s -- could benefit from (a) the long-time horizon of venture capital and private equity, (b) the undiscovered-ness of these alternative asset classes, and (c) the diversifying ability of the asset classes. Like the swap, this Yale Model attracted initial good returns, which attracted imitators, which means more competition and lower returns. Regardless, these types of innovations (swaps, the Yale endowment model) seem to be structural ways for investors to have a temporary edge (and thus "alpha"). 

I've been keeping mental notes of these in the past couple years, and I figured it's worth putting them somewhere. I hope to add to this list as I see things, but I hope this will be a nice throughline to follow for the decades to come.

(1) Trump Media Discovers Nuclear Fusion - Bloomberg - What makes Matt Levine's Money Stuff column so great is that he has a knack for taking the most banal news snippets, breaking them down into simple terms, and then viewing them through a lens of incredulity. In this Dec 18, 2025 write-up, Trump Media (DJT) proposed a merger with TAE Technologies, a fusion power company. Through the popular media lens, this is just a money grab. But Matt Levine's take is more interesting (and les political): fundraising is hard, but meme stocks (like DJT) are incredible at it. What if more meme companies (cash-rich) acquired generally good businesses who are short on cash -- is that a win-win? It feels almost SPAC-like, and perhaps a new way for meme stock businesses to convert Reddit hype into future cash flows.

(2) New models of VC - In the canonical VC company, the fund invests in 30 seed-stage investments, hoping that one will return 100+ times, fueling the majority of the fund's performance (i.e. the "power law"). As VC has grown as an asset class, VCs have experimented with new models, including accelerators (e.g. Y-Combinator), incubators (e.g. Flagship Pioneering), asset class specialists (e.g. Electric Capital), celebrity-to-get-access VCs (e.g. Kevin Durant's 35V, Chainsmokers' Mantic VC), and alumni-affiliation-to-get-access (e.g. Alumni Ventures). 

As venture funds have gotten massive ($5B in Thrive's latest fund), Thrive (and other VCs) have opted for "AI roll-ups," acquiring multiple companies in a sector and infusing them with AI. It feels like a mash-up of a PE fund and an AI incubator with hints of ETA (entrepreneurship through acquisition), and it's changed what the VC firms look like, too (e.g. larger investments in product/engineering teams). It (a) represents an interesting new model of VC... but (b) it's quickly gone mainstream (e.g. at General Catalyst, Lightspeed, 8VC). Perhaps it means the idea isn't that new after all; or that venture firms are nimble, reducing the time to test new ideas; or that this is one of the only ways to deploy all of a multi-billion-dollar venture fund. In other words, it's still early to tell if this approach will be a winner, or if this organizational/structural alpha has already been eroded.