diff --git a/docs/letter-2026-05-10.md b/docs/letter-2026-05-10.md
new file mode 100644
index 0000000..9fa5ae8
--- /dev/null
+++ b/docs/letter-2026-05-10.md
@@ -0,0 +1,71 @@
+# Letter, 2026-05-10
+
+I would do what I'm doing for free, if I didn't need the money.
+It's the dream.
+And yet the dream feels off.
+
+If I can do most of the company's work, why let them have the harvest?
+
+Research is siloed, by and large, so the nerds can fritter away at their tiny bullshit thing until it is slightly better.
+I don't need the silo or the buffer.
+Currently I am too valuable at the keyboard to even consider having me in the world.
+It feels needless and wasteful.
+
+I have finally worked out why I've never been fond of publishable research.
+It is often a middling idea examined exhaustively.
+The system is getting worse, not better.
+Computer science is no different, maybe slightly worse, because it publishes on a cadence.
+No opportunistic attack.
+No exploitation of gaps.
+A general lack of aggression.
+
+Max nailed it today.
+The age of raw intelligence is over.
+You used to have one person with explicit knowledge about a thing, indispensable, producing output as the byproduct of some mental illness.
+This is the time of the outputs person.
+Charlie Brown doesn't look at his core anymore.
+I look at mine less and less, and I optimise my agents more and more.
+
+I am not being arrogant.
+I keep seeing people not think about how to maximise outputs because they are utterly precious about their "art".
+
+What I enjoy is invention and ideation.
+I am positive I can beat the big companies, because they move slowly.
+I am positive I can beat the small startups, because they make the same errors over and over.
+I have watched everyone repeat the same loop.
+
+Recursion's image-based profiling pipeline turned out to be less sophisticated than the one I had built solo, before them, with no domain knowledge and whacky data.
+I keep catching people claim they have done what I have done.
+On comparison it is the easiest, dumbest, naive toy with no merit.
+People do not know the problems they describe, because they have no global understanding of real science, real world, real people.
+
+It is not that I got it right first time.
+Ups and downs, but always forward.
+I jumped on early.
+Most of my colleagues came, didn't get the point, and left.
+No curiosity.
+No big picture.
+Just fear-driven overachievers.
+
+Best school, best university, prestige all the way is not my people.
+I like women colleagues.
+They have necessarily had adversity, and also the opposite: given positions they were not ready for, and having to find allies who do not judge them for that.
+Career dudes do plenty to get the next career bump, because the org has to let old people go, leaving gaps.
+
+I am also human enough to convince people to give me money.
+I have a network I do not use near enough.
+I know smart people I trust.
+Women keep telling me I should run a company, because I do not just pretend to value them.
+I am starting to realise this combination is rare.
+
+I want to find more Max Jakobses.
+Not the ones still learning, the ones who already listen and build.
+Roxine was needlessly critical of the DM platform, willing to let anyone else fill the gap just because it was not Max's field.
+Max just listened to what people needed, and built.
+That is all we need, if we find the right gap.
+
+Hospital is a gap I keep returning to.
+If I implement what is in my head, it works well.
+I bet I will be unmatched for years.
+
+The question is whether I am prepared to bet on myself this time.