Rahulraj

founder | Reliability engine for institutions that can’t afford a bad answer

[Rahulraj @rahulwithraj]

Agent Evaluations are so broken with techbros checking text output and calling it a day

While pitching a potential customer [top 10 Indian auto] they mentioned how the engineering vendor had not hooked up their agent to their CRM completely and it was just spewing random numbers for the last month when asked about services left.

[Dima @dimakhanarin]

Current situation in the "services-as-software" market.

[Rahulraj @rahulwithraj]

@AdamSmasher424 I have learnt about a new game today

I will make this game my new daily addition

[Rahulraj @rahulwithraj]

Customers who want you to automate their workflows without allowing python libraries in their environments really need some life coaching

[Rahulraj @rahulwithraj]

@sambharia @beatsinbrief Good take

[Rahulraj @rahulwithraj]

Istg it’s so taxing to try and write thoughtful personalised messages to each potential customer who’s connected with you and AI seems to never get it quite right

Really considering becoming a meat proxy

[Rahulraj @rahulwithraj]

@bot @elonmusk @mattyp @mattyp can you pretty please look into this. I really really do not want to move to dots 😭

[Rahulraj @rahulwithraj]

Can't upgrade my Grok Bot tier.

@bot @elonmusk @mattyp

[Rahulraj @rahulwithraj]

@nayibbukele GTA 6 trailer

[Rahulraj @rahulwithraj]

@tryingtobeprat Bradbury project AI safety edition

[Rahulraj @rahulwithraj]

@GergelyOrosz I think the current range interview methods will assimilate to give birth to a new standard format - much like what DSA was.

Everyone is figuring what that new format will be , it’ll be fine.

[Rahulraj @rahulwithraj]

Am i retarded or is everyone else larping

[vas @vasuman]

It’s not just NVIDIA. Sources confirm Google's Gemini team was also leaking confidential information to China.

China is reportedly asking them to stop.

[Rahulraj @rahulwithraj]

@utk7arsh @dkrishan8

[Rahulraj @rahulwithraj]

me when i have to act like it's the first time i'm hearing about agentic improvement loops that my customer's head of engineering "invented" yesterday

[Rahulraj @rahulwithraj]

@AryanSeth07 @ezyang ELI5

[Rahulraj @rahulwithraj]

The office aesthetic i aspire to recreate

Functional Chaos

[Rahulraj @rahulwithraj]

while evaluating contact centre agents, an important figure to track is % of uncovered intents in production.

a bunch of companies only look at the journeys they wish to fulfil and eval those, ignoring what real customers ask, simply marking them as out of scope.

best example i saw: ~20% of queries on a company's chatbot were about jobs available at their org. made perfect sense to put a careers page link as a reply

[Rahulraj @rahulwithraj]

@ns123abc netiher do they let me upgrade 😢

[Rahulraj @rahulwithraj]

@AryanSeth07 @bot @elonmusk @mattyp print("The hottest new programming language is English")

[Rahulraj @rahulwithraj]

@AryanSeth07 @bot @elonmusk @mattyp Imagine writing code yourself in 2026, NGMI