Show HN: Littlebird – Screenreading is the missing link in AI • Hacker News Board

Re: Show HN: Littlebird – Screenreading is the missing link in AI

#31

Post by r3gal08 » Wed, Mar 25, 2026, 1:48 AM UTC

anonhn58 wrote: post #8

Hi folks, Tushar here from the engineering team at Littlebird. AMA!

How are you handling the data extraction? Is it a multimodal VLM (OCR+LLM) or a standard OCR engine feeding a separate LLM? I’ve been hitting a wall trying to understand how this viable. The compute overhead for real-time analysis at scale seems massive without a serious backend. How are you managing the frequency?

Re: Show HN: Littlebird – Screenreading is the missing link in AI

#32

Post by anonhn58 » Thu, Mar 26, 2026, 11:11 AM UTC

r3gal08 wrote: post #31

anonhn58 wrote: post #8

Hi folks, Tushar here from the engineering team at Littlebird. AMA!

How are you handling the data extraction? Is it a multimodal VLM (OCR+LLM) or a standard OCR engine feeding a separate LLM? I’ve been hitting a wall trying to understand how this viable. The compute overhead for real-time analysis at scale seems massive without a serious backend. How are you managing the frequency?

hi, while vision is going to be a part, it's hard to scale both on server and the client(it's resource intensive and the battery will drain faster on client). we hook deeper into the OS layer with accessibility, Apple Script and other ways to get raw text. This also lets us create a privacy friendly app with granular data controls for the user.

> compute overhead for real-time analysis at scale seems massive without a serious backend. you're still right about this part though, and we do have a serious backend.

Re: Show HN: Littlebird – Screenreading is the missing link in AI

#33

Post by beatrice111 » Tue, Mar 31, 2026, 10:28 AM UTC

timwis wrote: post #19

reasonmethis wrote: post #17

Earlier quoted context omitted.

Hey, I'm Dmitriy from Littlebird. For your use case, would you want the underlying LLM to be local as well, so that your data doesn't get sent to the "big dog" LLM providers? It's an important consideration because it wouldn't be nearly as smart then - though I totally understand if that's the only way it could work for you. We are SOC2, GDPR, and CCPA compliant btw.

I'm not the OP but I came here to voice the same concern. I would love to use something like this. I also signed up for rewind.ai and Limitless and pre-ordered the pendant. But ultimately I cancelled it out of privacy concerns. I wonder if it could be local storage and you could provide your own Open Router endpoint? That way it could be a local model or your own deployment of GPT/Claude in Azure/Bedrock/Vertex etc w…

Maybe you could try AirJelly, a context-aware proactive AI companion we are building recently. Data are all stored locally, on your MAC. Local LLM models would be difficult, as we also use Claude/Gemini to serve as the model provider. But as the software builder, we don't collect any user information, anything for improving our product would be sent by our users voluntarily (just send feedback in the app). If you would like to try, you could visit our website airjelly.ai.