Why I Stopped Writing Checks and Built an Instrument
Four bets across twenty-nine years turned out to be one bet. The scarce input in venture was never capital — it was judgment. So I stopped writing checks and built the instrument.
I have made four serious bets in twenty-nine years. I did not understand until recently that they were the same bet.
The first was SilkRoute, in 1996. Singapore had no commercial internet. We built one and discovered the actual product was not the storefront but the trust — the small set of guarantees that let strangers transact across a wire they did not understand. The second was Reciprocal, a few years later in the US. Digital content was leaking everywhere and nobody had built the rights layer that would let it move under its own rules. Microsoft bought the infrastructure.
The third bet stretched across two decades and a portfolio. 3PAR. Twilio. Cloudflare. Red Hat. I called it angel investing at the time, and the cheques were small, but the pattern was consistent enough that I have stopped pretending it was luck. Storage. Communications. Network. Open source. Every cycle I watched the application layer rewrite itself and the infrastructure layer compound.
askOdin is the fourth bet. The pattern is the same. The layer is new. Judgment — the work of weighing, deciding, and standing behind a call — is the asset the rest of the stack has always assumed someone else would supply. It is also the layer no one has built as infrastructure, only as features inside other things.
I am building it now because I have spent twenty-nine years watching what happens to people who tried to operate without it.
The same shape, four times
At SilkRoute the lesson arrived as a correction. We had sold ourselves as a marketplace, and for a while I believed the marketplace was the company. It was not. The thing customers actually paid for was the assurance underneath — that a counterparty they had never met would deliver, that money would settle, that the wire could be trusted before either side could verify the other.
Reciprocal taught the same lesson in a different register. Content was moving faster than the rules that governed it, and the gap was not a content problem. It was a rights problem — who could do what with which file, enforced at the layer below the file itself. Microsoft did not acquire a product when they acquired Reciprocal. They acquired a piece of plumbing they had decided they would otherwise have to build.
By the time I was writing angel cheques, I had seen the shape twice and still did not have a name for it. 3PAR was not a storage company to me; it was a bet that the way data was held would have to change before the things built on top of it could. Twilio was not a messaging company; it was the recognition that communication would become something you called rather than something you operated. Cloudflare looked like security and delivery and turned out to be a position on the entire path a request travels. Red Hat was the oldest and the strangest — a bet that something the industry treated as a hobby was in fact the substrate the next twenty years of software would be assembled from.
Four companies, four layers, one shape. In each, the visible thing was the application and the durable thing was the infrastructure beneath it. The market priced the application. The value accrued to the layer.
What I was actually doing
It took me an embarrassingly long time to see that the cheque was never the contribution.
In each of those bets, capital was abundant. There was money for storage and money for messaging and a great deal of money, eventually, for everything with a network attached. What was scarce, every single time, was the read — the willingness to look at a layer everyone treated as plumbing and decide, before consensus, that the plumbing was the company. The cheque was the artifact of a judgment. It was not the judgment.
This is the part the venture industry has organized itself to avoid admitting. The scarce input is not money. Money is the most abundant thing in the system; in most cycles there is too much of it, chasing too few good reads. The scarce input is the quality of judgment applied per dollar — the capacity to weigh an unfamiliar claim under real uncertainty and stand behind the answer when it is still unpopular. Every good outcome I have been near came from getting that one thing right early. Every bad one came from outsourcing it to a room, a brand, or a consensus that felt like safety.
Why I stopped writing checks
So I stopped.
Not out of disillusionment — the bets worked. I stopped because writing cheques had become the lowest-leverage expression of the only thing I was reliably good at. A cheque participates in one company at a time. The judgment behind it, if it could be separated from the cheque and made repeatable, could participate in all of them.
Awesome Ventures is a private office now. It invests proprietary capital and does not deploy other people’s. That structure is deliberate: it removes the part of the work that was never the work — the raising, the committee, the performance of conviction for an audience — and leaves the part that was. Reading. Deciding. Standing behind the call. Most of what I learned over twenty-nine years was not which sectors to like. It was how to interrogate a claim until the load-bearing assumption fell out of it, and how to tell that assumption apart from the one that merely sounds good.
That skill does not have to stay locked inside one person writing occasional cheques. That was the realization that ended one practice and began another.
The instrument
askOdin is judgment built as infrastructure rather than smuggled in as a feature.
The questions I would ask a founder in a first meeting were never secret and were never personal. They are the questions any disciplined reader of a company asks: where is the market claim actually sized from, which assumption carries the whole story, what does a skeptic see first, what survives if the most flattering number is wrong. I asked them by hand, one company at a time, for most of my working life. There is no reason they should remain artisanal.
Crucible is the part of that a founder can use directly. It runs the same interrogation on a deck before the deck goes into a room where the cost of the gap is real — the first meeting you do not get a second version of. It is not a verdict and it is not a substitute for the meeting. It is the scrutiny moved earlier, to the point where it is still cheap to act on. That is the whole idea: apply the standard before it costs you the shot, not after.
I spent twenty-nine years learning that judgment was the layer no one had built. The four bets were the same bet, and this is what it was always pointing at. The first three I made with a cheque. The fourth I am making by building the instrument itself — so the read no longer has to depend on whether the right person happened to be in the room.
If you came to Awesome Ventures looking for a cheque, this is the more useful thing I have to offer. Run your deck through Crucible at crucible.askodin.app before your next real pitch. The rest of what we are building is at askodin.app.