Hacker Newsnew | past | comments | ask | show | jobs | submit | Simorgh's commentslogin

The only OAuth I'm comfortable with is Github.


As someone who has worked with speech to text, I can tell you from personal experience that audio seems to be broken across the ecosystem.

Bluetooth as an example is very buggy. Siri, as a voice-to-voice intelligence, fails to work most of the time for me.

I think that developers find it hard to ensure a seamless UX for anything that uses audio.


I used to love dark mode, but in the summer months, I came to feel that it's overly depressive, possibly affecting my mood.

In the winter months, it's slightly more usable because of association with lower lighting, darkness and even coziness. However, I think it can affect mood irrespective of the time of year.

Nowadays, I'm often experimenting with the idea of displays which are backlit with colours other than pure white or pure black.


Sign-up / in via the power-button in the top-right!

It's a really simple game.

You can use geneva:interface to play. Hopefully, it is the best chat-interface out there for winning (not just playing).

State-of-the-art, voice-first and very, very fast for computation, compared to chat LLMs.

Currently running Mistral.

Try your best. If you're doing really, really good, especially via geneva:interface or otherwise, feel free to email me via the winter delta dot com homepage, and I will create a leaderboard for the top performers.

Let's see who can do the best!

Hint: Have you tried writing a /POST route? ;-)


Thanks! I use Athens as a personal voice-memo copilot. It’s much harder to use it for sharing ideas with others; e.g. for presentation…

It’s voice-first so you have to express what you’re trying to communicate solely via speech to text. I think that’s part of the challenge. ‘Thinking in voice’, for presenting ideas is quite hard for me.


Is it using a model to generate the ASCII art?

It’s quite clever.


It's gpt-4.1


Got it — it’s interesting how a text model can generate compelling visual representations that are essentially style-matching.

Matching styles using vision models is not that easy, especially if you want to capture the core subject of the prompt.


I know it’s an aside but I’m so impressed with the quality of this (linked) page:

https://capsul.org/

The UX is exceptionally strong. It’s like you’re on a ‘real website’ — it’s quite enveloping in terms of its vibe, palette, ease of comprehension and the colour palette is so cool.

Definitely a tilt in favour of boldness with colour on web.


thx :')


This is a really understated point — especially for western democratic societies.

I would argue that we should not be supportive of regimes (e.g. Israeli Likud) that are accused of violent hostility to war reporting and journalists.

Without open conflict-journalism, not only are war crimes harder to uncover (or easier to get away with) but sensitive voters and their democracies (everywhere) are threatened by a risk of subversion

I would posit that even the suspicion of collaboratively concealed war crimes disperses a planet-scale chilling effect throughout the ‘global social fabric’ of society

I would argue that the risk of extended damage to this fabric was the real motive for the Iraq War.

Maybe we should be challenging Israel much more deeply. We have not applied military pressure upon them, w.r.t. the current conflict…


Hey — working on Athens — it’s a messaging-interface primarily but it does fulfill a lot of these checkpoints — the interface is voice-first and VAD is handled pretty well! Open Banking connection is coming for advanced correlation between messaging and your fin. goals — awesome for the side hustler within!

My email in my profile is there; if you’re interested to call


I use Granola, I’m not affiliated with them. They create detailed meeting notes from whatever I type during the meeting, it uses the transcript and AI to flesh out my notes. It’s a great idea and their tiering is nice. 25 meetings free iirc. I sometimes chat with past meetings and find it useful to deep dive into the narratives.

The bigger issue I have is to do with messaging and clientele-communication itself. I’m building Athens to help me shift more things to voice ai. I think voice computing is the way forward. There are a lot of benefits, it’s essentially more ‘human’ and personal - things get resolved quicker, decisions are better and made faster. I am being asked about meeting integrations for it and that’s on the roadmap.

Here’s a link if you’re interested:

https://athens.winterdelta.com/playlist/536a9778-cd9d-4491-8...


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: