atharva
@atharvaraykar.com
i write at atharvaraykar.com
i work @nilenso.com
yes-anding the world.
i work @nilenso.com
yes-anding the world.
I think this is the correct take if one is short-horizon AGI-pilled. But since I don't expect the shape of agent intelligence to fully be like human intelligence anytime soon (say ~10y), it helps to have a giant doorknob for agents, geared towards their style of processing information.
August 21, 2025 at 2:39 PM
I think this is the correct take if one is short-horizon AGI-pilled. But since I don't expect the shape of agent intelligence to fully be like human intelligence anytime soon (say ~10y), it helps to have a giant doorknob for agents, geared towards their style of processing information.
This is a good example of the weirdness of AI.
August 14, 2025 at 7:15 AM
This is a good example of the weirdness of AI.
Reposted by atharva
I loved it! simonwillison.net/2025/Jun/10/...
AI-assisted coding for teams that can’t get away with vibes
This excellent piece by Atharva Raykar offers a bunch of astute observations on AI-assisted development that I haven't seen written down elsewhere. Building with AI is fast. The gains in …
simonwillison.net
June 11, 2025 at 1:36 PM
I loved it! simonwillison.net/2025/Jun/10/...
indeed, prompting is still extremely sensitive and brittle even today.
Some of my colleagues @nilenso.com managed to get better code by just saying "write it like Rich Hickey would" which seemed to have activated the right neurons
Some of my colleagues @nilenso.com managed to get better code by just saying "write it like Rich Hickey would" which seemed to have activated the right neurons
June 11, 2025 at 8:30 AM
indeed, prompting is still extremely sensitive and brittle even today.
Some of my colleagues @nilenso.com managed to get better code by just saying "write it like Rich Hickey would" which seemed to have activated the right neurons
Some of my colleagues @nilenso.com managed to get better code by just saying "write it like Rich Hickey would" which seemed to have activated the right neurons
in my case, I did not get into the weeds of the workflow, because my own workflows are constantly changing as these tools and models evolve
I tried to focus on what consistently helped throughout all the changes, and the best rule of thumb I could come up with is "what helps the human helps the ai"
I tried to focus on what consistently helped throughout all the changes, and the best rule of thumb I could come up with is "what helps the human helps the ai"
June 11, 2025 at 8:26 AM
in my case, I did not get into the weeds of the workflow, because my own workflows are constantly changing as these tools and models evolve
I tried to focus on what consistently helped throughout all the changes, and the best rule of thumb I could come up with is "what helps the human helps the ai"
I tried to focus on what consistently helped throughout all the changes, and the best rule of thumb I could come up with is "what helps the human helps the ai"
glad you found it useful!
June 11, 2025 at 8:23 AM
glad you found it useful!
a trap I see with a lot of software engineers is missing the big picture. the code and the tech is not the engineer's most valuable artifact. it actually is the systems that they are enabling and building.
especially in hard complex problem spaces like public health!
that's why I took this angle.
especially in hard complex problem spaces like public health!
that's why I took this angle.
January 6, 2025 at 8:28 AM
a trap I see with a lot of software engineers is missing the big picture. the code and the tech is not the engineer's most valuable artifact. it actually is the systems that they are enabling and building.
especially in hard complex problem spaces like public health!
that's why I took this angle.
especially in hard complex problem spaces like public health!
that's why I took this angle.
reading meadow's thinking in systems gave a lot of structure and vocabulary to the type of reasoning I found enjoyable when trying to make sense of the world.
with the work I did in public health, these patterns were plain as day to me. I had to write about it.
with the work I did in public health, these patterns were plain as day to me. I had to write about it.
January 6, 2025 at 8:28 AM
reading meadow's thinking in systems gave a lot of structure and vocabulary to the type of reasoning I found enjoyable when trying to make sense of the world.
with the work I did in public health, these patterns were plain as day to me. I had to write about it.
with the work I did in public health, these patterns were plain as day to me. I had to write about it.