So the real question is:
Who is building the reliable, cheap, enterprise-ready agent layer for healthcare, legal, education, logistics, etc.?
Those will be the 2026–2028 unicorns. What do you think, am I too bearish or right?
Drop your take @garrytan
A few days ago, the CEO of Y Combinator, @garrytan, shared this on Twitter about how the next most unicorns can be made by solving problems in places where AI or AI agents are negligible>10%.
Respectfully, I partially agree… but the reality is harsher:
And most importantly, AI is like mathematics.
It’s simple only when you truly understand it.
Otherwise? Nightmare. The space on this chart isn’t “easy money”.
It’s a massive opportunity only if someone solves these real problems.
The bar for breakthrough things fell really hard
2022: GPT-3.5 released (turning point in LLM)
2023: Competitors forced to come out (Google, Anthropic, Mistral)
2024: Agentic workflows and assistants
2025: We remove em-dashes from ChatGPT, now just copy and paste 🤡
@github@GitHubCommunity@GitHubEnt I know copilot is good and works well in vscode, but i guess i only know that, otherwise you guys have wouldnt charge me and still dont give access to betters models
Kindly fix it or refund me please
@github i mean i know copilot exist but can you also remember it and charge me properly for monthly subscription and give full models and better service
I can't use @v0 now for till 4th May, this comes after i cancelled my subscription of v0, @leerob, is it worth it? From a product perspective, this isn't a smart move in the AI landscape.
@v0 i got premium and realised V0 also gets into a loop of prisma and doest show preview of existing website which i was editing (i non technical so bare with me) i even made an admin panel but it isnt letting me load it, please help me fix it