@MrJefago 3) Built POC automating Android + GCP setup when integrating RC SDK for the 1st time: https://t.co/nFUFubmYMU (1.5x)
4) Truly resonate with mission of RC - helping small businesses take off!
More about me here: https://t.co/Qq44J6TXEq
@HarveenChadha always wondered - why is there no openrouter for India? was paying for GPT subscription until recently, but sometimes wanted to use sonnet etc
like recharge for 100 and get access to 10 models (like openrouter)
I've been getting my hands dirty with voice experiences (I learnt that they're called VUIs, voice UIs) - and decided to publish some latency metrics I was seeing!
These are just indicative metrics, I didn't do the tests with scientific rigour so take with grain of salt
1/5
For conversations to feel natural, user perceived latency should be within 500ms is the industry benchmark (where did this benchmark arise? anyone have the source literature?)
And both these experiences fall short of that target - by a mile!
4/5
@deepakshenoy Everyone - from mum to cook been saying for the past month prices of veggies and fruits are through the roof (Mumbai) - how do I reconcile these 2? Is it the low base rate effect you spoke about in the previous tweets?
@PoojaIsNagpal "Anthropic ships a sub 200ms latency voice model with native tool calling" - gpt-realtime/gpt-realtime-mini already do this and seem to work kinda well - so v curious - what am I missing?
(sorry if it comes across as a n00b Q - I'm not an expert in this space)
I didn't find any great literature on UX considerations when building voice forms - so I put some lessons I learnt when I recently built one.
Context: I recently built a free Splitwise clone called halfsies - helps you track shared expenses - between housemates, friends etc
1/7
Full post here - let me if y'all have tried to build something like this and have different approaches to some of the new challenges.
https://t.co/cX17DkaD05
6/7