Sherlock

running | official @framer expert |
designing at @theresidency india, @googledevgroups and friends of @figma hyd.

[Sherlock @sherlock_ux]

short intro post before i make an actual one:
- i'm nik, 21 y/o
- always loved computers and psychology
- attended a ui/ux workshop and realised design is perfect mix of psy. and cs.
- did TONS of free design work, while gaining more and more visibility.
- started getting freelance gigs, with $5 for 7 screens.
- kept up with lots of volunteer work, earned a great word of mouth among the best engineers in the city.
- fast forward, i'm now the go-to designer in my city for community initiatives by google dev groups, figma and more.
- worked with 8+ startups to design products 0-1, designing some of the coolest products i could.
- now, kinda running a design studio, on the way to becoming design engineer and handle cooler projects.

[Sherlock @sherlock_ux]

what if..

[Sherlock @sherlock_ux]

tweet

[Sherlock @sherlock_ux]

@Praha37v @originkit originkit

[Sherlock @sherlock_ux]

@figma holy moly, WHAT

[BentoLabs AI (YC P26) @BentoLabsAI]

We ran our recursive learning layer on Terminal-Bench 2.0. Same agent. Same model. Same harness. Same budget.

The result: Claude Sonnet went from 42.2% → 52.4%. A +10.2 percentage-point lift, significant at p < 0.05, with a 13:3 task-level win/loss ratio (internal).

The only variable was a learning layer.

We wrote a full technical breakdown on what changed, why it worked, and what this means for production AI agents.

Read it here 👇
https://t.co/VhVCgURuXv

[Sherlock @sherlock_ux]

:)

[Y Combinator @ycombinator]

.@BentoLabsAI is the monitoring and learning layer for long-running agents. Their learning layer gives agents model-jump gains: Sonnet 4.5 went 42.2%→52.4% on TB2 (Internal).

Congrats on the launch, @Abhinavv_soni &amp; @kacppian!

https://t.co/LTy5sslfni

[BentoLabs AI (YC P26) @BentoLabsAI]

We are live on the YC Directory.

https://t.co/6nEr0It6Do

[Sherlock @sherlock_ux]

The Internet makes AI agents sound very calm. Production does not. Working on changing that story.

@BentoLabsAI

[Sherlock @sherlock_ux]

@thefoundingco_ @theresidency we got residency in hyderabad?!!! lesgooo

[Sherlock @sherlock_ux]

@an5rag @deoeven dope stuff

[Sherlock @sherlock_ux]

@itsss_chidu wdym a month of dark mode, you've been using light mode your whole life???

[Sherlock @sherlock_ux]

@JaswanthBeere thanks :)

[Sherlock @sherlock_ux]

career update.
joined https://t.co/3y1uLlo74b (YC P26) as a product designer.

we're building recursive learning for ai agents to minimize failures in production.

;)

[aman @amannhq]

this is exactly why AI agents fail in production.

no regression detection. no drift alerts. just an agent confidently hallucinating "they must be cooking" while the restaurant is closed.
every failure like this is a silent regression that reached a real user.

that's the problem @BentoLabsAI solves.
cc @deepigoyal @zomato 👀

[aman @amannhq]

this is exactly why AI agents fail in production.

no regression detection. no drift alerts. just an agent confidently hallucinating "they must be cooking" while the restaurant is closed.
every failure like this is a silent regression that reached a real user.

that's the problem @BentoLabsAI solves.
cc @deepigoyal @zomato 👀

[Sherlock @sherlock_ux]

what are the best podcasts you guys would recommend?

[Sherlock @sherlock_ux]

@star_yutish @crackedhousehq would love to visit next time im in bangalore

[Sherlock @sherlock_ux]

scrollytelling landing page, built with @framer live at https://t.co/XDG4ThdydB

[Sherlock @sherlock_ux]

scrollytelling landing page, built with @framer live at https://t.co/XDG4ThdydB

[Sherlock @sherlock_ux]

@Praha37v count me in

[Sherlock @sherlock_ux]

Landing Page designed + built on @framer, animated with @jittervideo

-&gt; live at https://t.co/TATWw2c96q