foundation models for AGI & Robotics | @fdotinc
San Francisco, CA
- Tweets 188
- Following 248
- Followers 197
- Likes 760
Every vertical AI startup this year:
"We're proud to support the launch of [insert whatever OpenAI did to kill a vertical]"
Harvey is proud to support the launch of Astra for Law.
Harvey’s plugin within ChatGPT Enterprise gives lawyers a new way to access Harvey directly from ChatGPT.
With the plugin, Harvey will also be able to build on the new legal configuration, bringing its capabilities into Harvey's products.
2
5
1,049
Walking around Oxford rn, and the way they approach AI research seems fundamentally different to US universities.
A lot of the most interesting Oxford research starts with “What should an intelligent system actually be trying to achieve?” instead of “how capable can we make AI?”
That leads you into human agency, uncertainty, alignment, coordination, and limits on autonomy.
Here's one of my favorite papers:
arxiv.org/abs/2607.26654
9
383
CITYMAXXING is an underrated hack for market validation if you're b2c.
Go to a city near you, bring two jobless friends, and talk to ~300 people a day.
This works because ppl love to hate, and if they hate on ur idea, they'll tell u exactly what parts they hate and what they would do.
you can iterate your mvp based on market feedback on the spot hundreds of times in a day.
1
9
381
With a lot of talk about AGI and AI regulation, I've been thinking about who stays the observer w/ AI, and who judges which future's worth building toward.
I found a parallel between a quantum physics theory and an AI lab (Anthropic) building that loop:
9
643
The tension in Dogpatch is so palpable now that it’s the last weekend before YC S26 Demo Day.
I literally just witnessed a founder sobbing in the corner of a cafe & another pair of founders so confident that they’ll raise that they decided to take the weekend off.
I’ve learned so much from working directly with some YC founders this batch, and the crazy ways they attained 10% growth every week has been so interesting to watch.
Good luck to all the founders on Sep 10!
📸 pic of just the YC sign cuz I won’t take a picture with it until I get in
9
62
6,477
If you ever think your startup idea is too "late," remember everyone thought OpenAI was too late at first.
1
9
525
Dharshini Maru retweeted
the openai huggingface incident, from an agents pov.
(part 1)
269
1,483
12,484
1,804,417
when I see a freshman solve a two sum for the first time but I genuinely laugh at him for using a double for loop instead of a hashmap
5
281
“It doesn't make sense to continue wanting something if you're not willing to do what it takes to get it” - Atomic Habits
1
5
251
(I give away a pretty good physical infra startup idea in here)
Hot take: building reliable physical systems requires a deep understanding of biology.
If you truly think you can build better system architecture than what nature has already provided the framework for, then you haven't researched enough micro-level biological systems.
I think a lot of engineers inherently know this, but don't want to put in months/years of work studying computational biology, which is why I wrote a TLDR version of some important ML lessons I learned from biology that I discovered over the last few months:
dharshinimaru.substack.com/p…
4
631
Hot take: building reliable physical systems requires a deep understanding of biology.
If you truly think you can build better system architecture than what nature has already provided the framework for, then you haven't researched enough micro-level biological systems.
I think a lot of engineers inherently know this, but don't want to put in months/years of work studying computational biology, which is why I wrote a TLDR version of some important ML lessons I learned from biology that I discovered over the last few months:
dharshinimaru.substack.com/p…
7
1,120
Dharshini Maru retweeted
Ship with an expert by your side @vovy_ai
Made with AI
Paid partnership (ad)
16
7
49
24,888