@rohanmanchanda@rohanmanchanda - fully agreed !
@cursor_ai works best when you are good in programming and aware of how to build software architecture, it will definitely increase your coding speed by 10x.
@gokulr We are exactly building this at @ThalamusHQ and executing compound AI firsthand!
A foundational platform (with 20+ AI Agents) targeting a niche Proposal industry and can expand in other domains as well.
The future is not one agent—but an ecosystem of agents working together!
@paraschopra On the consumer side, the 1-2% developers who are using APIs will hardly shift here unless they are actually building for Indian specific consumers. Those buildings for the global will still stick with western APIs.
@paraschopra TBH building a Foundational model from India doesn't make sense (other than pride) from a monetization point of view unless you are targeting governments or indian enterprises(B2B).
On the consumer side, it's very hard to shift people from existing models like ChatGPT.
@kimberlywtan@a16z Completely agree , this is in the line in which we are building @ThalamusHQ , starting with a niche use case of RFP workflow, with further expanding agents into other dimensions!
@gauravmunjal Personal story: After my last startup acquisition, I thought of building LLM from scratch called Rabbit (~personally spent $1500 on GPU for training), and it was progressing really well. Pitched to some of the investors but no luck.
Here is the code: https://t.co/N5YNFZWMC4
@gauravmunjal Investors not being bold and probably their lack of knowledge in deep tech. They are still in the world of looking for the early traction and belong to the "wait and watch" category.
@ChiragCX@lateinteraction Something in the workflow automation space for Enterprises using multi-agents. Felt like #DSPy helps in easily building smaller modules (i.e. blocks ) for creating a bigger one.