Prasad

Full-stack + AI engineer. Building Examlly. Context Engine and Nexus UI in the open. Ex founding engineer at VectorShift (YC). Open to work.

[Prasad @prasad_patewar]

With AI 1 engineer can do the work of 10 lets say

That means companies shouldn't downsize by /10 but should even increase the engineering workforce

Thus increasing the R&D output over the competitions
But still
Whats stopping companies from doing more R&D and experiments ?

[Prasad @prasad_patewar]

Interesting read

[Xiaoyin Qu @quxiaoyin]

Prediction: the value of all verifiable tasks in the long run is 0. There is no profit there.

Anything where the success is defined in a finite and non-ambiguous way, it’s over. Value is 0.

The profits are all in taste/judgement related tasks where some like it and some hate it. Vibes only. No rationality.

Labs are hillclimbing verifiable tasks now and it might sound impressive that OpenAI found those math proofs but soon others will follow and players like DeepSeek will drive its cost to electricity.

Ambiguity is the life line and where the money is.

- Anything that’s irrational, vibe driven and controversial with no consensus definition of good is worth something.
- If the society all agree on what good looks like for a given task with no ambiguity, it’s worth 0.
- caveat: coming up with verifiable task to do is worth a lot because the ability to come up is not a verifiable task by itself

[Prasad @prasad_patewar]

@get_artcraft proved the point i was making few days earlier.
There should be more such great open source project that world needs.

We are also creating our own in open source.
Check out - https://t.co/FP7LdsZmhD

[Prasad @prasad_patewar]

With AI everybody is building something and trying to sell it from the start.

I think there should be more open source projects than ever.

Imagine if VLC, Linux, Git, VsCode, etc paywalled them from the start. Great devs always built for the world

Lets build on that LEGACY

[Prasad @prasad_patewar]

Hot Take
If you are the expert in your field, the AI output related to your field seems very average
but for other fields the output seems pretty good

Its not AI, its human bias towards things that we know and dont know

[Prasad @prasad_patewar]

@NehaTanti335133 @GnaniAi nice niche use case

[Prasad @prasad_patewar]

The rate at which rupee is sliding, lot of indian AI SaaS companies profit margins are shifting as they need to pay the bills in dollars

India IT desperately needs its own Cloud and AI infra at indian costs

There should be some report on this

[Prasad @prasad_patewar]

Software development with AI is more addictive than manual for product mindset engineers

[Prasad @prasad_patewar]

Future of Interfaces is chat (as of now)

[Luke Wroblewski @LukeW]

every app is being CHAT-ified.

[Prasad @prasad_patewar]

Where are SLMs (Small Language Models) ?
Specially trained for specific industry

If someone is building product for healthcare/defence/software etc
The usecase of that industry should be covered through SLMs

Less cost and Faster response with more accuracy

WDYT all ?

[Prasad @prasad_patewar]

@dark_coderz Yes and its pretty clear

[Prasad @prasad_patewar]

@0xPrajwal_

[Prasad @prasad_patewar]

Heads up !!

Claude code and Opus 5.5 seems fast and doing great work but sometimes its speed is the result of it reading saved memory files and not the actual codebase

ask claude to update its memory files after significant change in codebase
Dont rely on it answers blindly

[Prasad @prasad_patewar]

Is anybody here building AI harness for specific use case ?

Please share your experience here

[Prasad @prasad_patewar]

I think its first time in history that humans developed machines which can think on behalf of humans

As humans we mostly developed machines which solved the physical labour and optimized it and people paid for that

AI is totally on different spectrum

[Prasad @prasad_patewar]

@aslambeg84 Getting a job and even working on it is just minor part of total work

Being startup founder you own a lot of things and departments of work so its not even a comparison

[Prasad @prasad_patewar]

@tlakomy its not dead but its not fun anymore like before

[Prasad @prasad_patewar]

well this is ai generated image - but it has the crux of what i want to express

[Prasad @prasad_patewar]

I miss the days when teams thought which feature is worth building and how it should gets built.
Teams got serotonin after successful feature launch not dopamine hits.

now a days its just push first with AI and see later.
arent we all tired of this ?

somewhere it feels wrong

[Prasad @prasad_patewar]

llms are good at implementing but they are best at conducting research and pocs

this just feels like wonder -
research through docs, libraries try various approaches and test around scenarios where things will break, present the report at last

one week of dev work in minutes