Joseph Mocanu, PhD

A very rare seed stage healthtech impact venture capitalist -

[rygo6 @_rygo6]

I still believe a correct framing on LLM usage is that they can get you 80% of the way upfront, but that last 20% starts to become impossible to overcome. The proper way to use an LLM is to do the difficult manual 20% upfront. Once you have some decent structure, decent API surface area, decent data structures laid out, decent base libraries. Then you can largely prompt your way through the last 80%. While the first 20% which you did manually, intentionally and fully comprehend. That 20% essentially becomes part of your development harness, almost a prompt itself. Where you can steer its overall architecture, or data design, or other critical pieces by keeping a grip on that first manually done 20%.

Something real concerning with this. Knowing what to do in that first 20% requires advanced knowledge of the problem domain you are going to deal with. It is a staff or principal level engineer domain. Which most simply can't do.

Which means most people are going to run the gauntlet heavily relying on prompts, and are going to cover some incredibly impressive ground very quickly. Because 80% of the way is quite a long ways there. It looks like it could almost be done. But then they will get stuck in that last 20% and will never be able to make it perfectly polished. Nor perfectly sable. Nor remove all the odd quirks. Nor overcome the last pieces of performance or memory issues. It will basically hit a plateau where they lose control of it. Or it becomes unwieldy to take it in a new complex direction, or add some novel complex functionality to it. It will be very hard to tell from the outside if any given company or project is aimed at getting stuck in that last 20%. As they could appear to be kicking some major butt in that first 80%. But then it will reach a plateau where they can no longer do anything substantial without breaking something else, or seem unable to fix every little quirk or issue. A lot of companies will probably get stuck here. With dozens of very impressive things 80% of the way, but seem unable to get over the last final 20% to really perfect it and polish it.

[Joseph Mocanu, PhD @jmocanu]

@gandamu_ml As an ex-demoscener (and not elite like Future Crew), I remember killing myself over weeks to even get a mediocre result... just jaw dropping. And.. seeing your prompt revealed the AI genuinely has 'taste'.

[Joseph Mocanu, PhD @jmocanu]

What the....

[gandamu @gandamu_ml]

Opus 5.5 just one-shot this 90s-style demoscene demo for me. It's pretty good.
It's a genuine one-shot (in colloquial X post sense - single prompt) and my first attempt asking it to do such a thing. I just supplied Purple Motion's music from Second Reality. C/C++ and OpenGL.

[Joseph Mocanu, PhD @jmocanu]

@michuk @Streak Not really - but fully collaborative within and maybe to certain extent with portcos. I've found full context agents have been too single player

[Joseph Mocanu, PhD @jmocanu]

Awesome video!!

[RoboStrategy @RoboStrategy]

Endiatx - A Short Film

An inside look at @endiatx and the team helping build the future of robotic medicine

[Christina Farr @chrissyfarr]

Horrible indictment of the NHS, although not necessarily of government-funded healthcare overall.

I'm a huge fan of Singapore's model and think it could work in America. Strong government subsidies, meets personal savings & insurance. A best of both "hybrid" that works.

[Joseph Mocanu, PhD @jmocanu]

First Github PR... and with rclone at that. Oh down the deep rabbit hole I go.

[Velina Tchakarova @vtchakarova]

If you have time for one video today, please watch how Rishi Sharma has interviewed countless veterans from WW2 over the last ten years. And in moments of weakness and indecisiveness, you will recall their legacy, moral clarity and courage to take action.

[Joseph Mocanu, PhD @jmocanu]

We are at the Cylon in the bathtub moment of AI

[Joseph Mocanu, PhD @jmocanu]

Watch out for Gemini Flash 3.5's default thinking level.... my credit card learned the hard way.

[Joseph Mocanu, PhD @jmocanu]

It is incredibly frustrating to see so-called AI experts spout their opinions when it's clear they've never developed anything in, or using, AI before

[Joseph Mocanu, PhD @jmocanu]

In the age of cheap agentic AI, non-targeted / off-targeted cold communications is just lazy.

[Joseph Mocanu, PhD @jmocanu]

@michuk Next time run Claude Code on your VPS or home server so you can ssh into it

[Joseph Mocanu, PhD @jmocanu]

I bought two books today, both authored by people I personally know and call friends. Neat :).

[Isaac Bogoch @BogochIsaac]

Step 1: Insert bacteria ("Wolbachia") into mosquitoes.

Step 2: Release mosquitoes.

Step 3: Watch Dengue rates plummet.

Phenomenal results from Singapore.

Link: https://t.co/6TZZf5EdIs, by Lim et al.

[@jason @Jason]

🤌

[Joseph Mocanu, PhD @jmocanu]

In this day of cheap agentic AI, a generic spam solicitation is just lazy.

[Joseph Mocanu, PhD @jmocanu]

Claude is starting to refer to me by my name... I'm not sure how I feel about that.

[Joseph Mocanu, PhD @jmocanu]

This is so neat and so terrifying at once

[snav @qorprate]

something very strange is going on with Gemini 3