abhinav

everyone's building better voices for AI, but who's teaching AI to actually listen? @hume_ai lets developers send audio, ask custom questions in plain language and get structured probabilities back. no transcript, no separate classifier for every task really interesting directi

someone really made The Sims, but every character is an AI agent with its own computer, memory and actual work to do @getskydive is making managing an AI team feel like running a virtual office, while agents handle real workflows in the background this is a pretty fun glimpse

i’m starting to think AI agents are only as useful as the context they have about you @heyrobynai builds agents that know your tools and preferences, then actually gets the work done, from research and reports to finished decks this is closer to a personal AI team than another

the hard part of scaling AI agents isn’t just getting them to work, it’s controlling what they can access and what they spend @mintive brings MCP access controls, per team LLM budgets, real time tool call security and agent monitoring into one platform this is the infrastructur

the impressive part about @snaptrude is that it goes beyond generating building designs it builds detailed 3D models with coordinated structure, plumbing, electrical and construction drawings, while keeping everything editable less time drafting, more time exploring better desi

what stands out about @axiommathai formalizing CFSG is that the process uncovered actual errors in the original proof and led to new mathematics 15,000 pages of work, generations of mathematicians, now machine checked in Lean this is what AI should be doing, pushing the boundar

the creative work is the fun part, everything around it is what eats up your week proposals, project updates, approvals, invoices, it never really ends MonoDesk brings the whole client workflow into one place so you can spend less time managing work and more time actually makin

ai labs have already trained on so much of the public internet, interesting data now is what’s sitting inside companies, from private codebases to real world RL environments @datavendor_ai makes that data easier to discover, evaluate and buy without the usual weeks of back and