66 hard drives went missing from Zoya Akhtar’s production house recently (easy INR 20L-50L worth of hardware + the cost of the IP leak) which is newsworthy until you realise that 90% of production houses everywhere are at risk of the exact same thing.
Over the past 6 months, I’ve spoken to maybe 100+ agency/production house owners, and almost all of them run on some version of:
hard drives + excel sheet + "someone knows where everything is"
Which is INSANE considering how much these assets are worth (production cost + IP + reputation risk)
There is a better way. Reach out if you’re interested in hearing about it!
🚨 Product Announcement: VideoVault
The Aeos ecosystem produces 100s of videos each month. One shoot involves 100GB+ of cinema grade footage. As the world moves to 4K, file sizes are increasing. Yet most teams are still using SSDs/Cloud for storage & organisation (we were too!)
Proud of our team from @Aeos_Labs - @ycocerious@AjinkyaBodke_@thecmdrunner for winning 2nd place at the India leg of the @elevenlabs Global hackathon in Blore!
They made a super neat AI powered voice agent, that helps people journal their thoughts better 👏👏 well done team!
1,518 builders.
72 hours.
150 made it to the semifinals.
1 mission to win the India's biggest GenAI Buildathon.
See you all in Bengaluru for the Grand Finale!
Google's going to start watermarking everything generated by their AI - well intentioned, but I suspect wont make a difference unless other players (incl in China) and open-source all jump on too
https://t.co/lxLiT4N2n0
I think one of the biggest issues right now is the cost per minute - even assuming 100 hrs of talk time / month, its only marginally cheaper than hiring a human (in India) - sure an argument can be made about the upsides like multilingual, expert knowledge etc, but (un)surprisingly its still a hardsell to Indian businesses
I was speaking at an event in Sri Lanka last week (beautiful country btw) and one of the segments was a brainstorming session where the audience formed small groups and thought of AI/tech ideas together
8/10 teams came up with ideas that were political - "a lie detector for politicians", "a promise/performance monitor for government" etc
Makes sense, given everything Sri Lanka’s been through recently with the economic and political chaos. It really struck how deeply innovation/creativity (even in tech) is tied to someone's lived experiences
Got a chance to visit the very impressive Police Command Center in Bangalore last week - learned a lot about the systems behind every 112 emergency call & response @BlrCityPolice
Yeah I remember reading one of Paul Graham's essays about how "hacking tests" has always gotten in the way of true learning - but I guess it was always a scale issue - how do you mass evaluate a population efficiently - but honestly I think going forward AI will do two things:
1) Change what's actually demanded of the workforce
2) Make more holistic evaluations possible at scale
I called my dad this morning - He was obviously not expecting me to and after the standard hi/hello - he was like "Harsha is this an AI voice? Am I being scammed?" 🤦♂️
It used to be that "training data + training time" would approx determine the competency of an AI model - but now that these new LLMs (o1) deliberately pause to "think", I imagine that the "time spent thinking" during inference will also play a role in their reasoning performance 🤔