context windows went from 4k to 200k tokens in like 2 years. models got smarter. agents can now run for hours without stopping. none of those things alone are crazy. all three at once is actually insane and i don't think people are taking it seriously yet

is there a version of this where two 19 year olds close a B2B deal or are we genuinely delusional 502 emails. 11 opens. 1 "not interested, thank you." asking for a friend

cs kids are doing class projects. finance kids are practicing DCF models. i'm on a call trying to explain AI voice agents to an HVAC guy who just wants his phone answered. none of us really know what we're doing tbh

sat in econ lecture today thinking about who actually makes money when GPT-5 and Claude and Gemini all do the same thing probably not the companies spending billions to build them

our competitor could copy everything we built in a weekend. same n8n flows, same claude api, same voice receptionist setup. i think about this more than i probably should.

HVAC company with 10,000 past customers in a spreadsheet has better AI training data than most startups. they just don't know it yet.

my co-founder is a finance major. i'm a finance major. we built a voice receptionist that answers calls and books jobs for plumbing companies. no CS. no team. just figuring it out. every industry has a version of this waiting to happen.

everyone around me expects JP Morgan, startups-as-a-phase, or a "real job." that's not the hard part. the hard part is staying sure about something nobody around you believes in yet.

finance classes taught me DCF models, ratio analysis, how to value a company Claude can do all of it in 30 seconds nobody in the curriculum is asking what the analyst job looks like after that

claude dropped an update last week and our voice receptionist just... got better. we didn't touch anything. that's never happened to me before in anything i've ever built.

built a voice receptionist for a trades business using n8n, claude, and some api keys i figured out as i went. zero CS classes. no team. 19 years old. that would have cost someone millions to build in 2015. i'm not exaggerating.

pivoted our pitch 4 times in 6 months not because we kept failing. because every time we talked to someone real, we found out we were solving the wrong problem still not sure we have it right

classes kill your momentum right when you're in flow. but then you have a random tuesday at 2pm with nothing and you ship something real. college is the worst and best schedule for this simultaneously.

built a voice receptionist that answers calls, qualifies leads, and books jobs. never took a CS class. just me, claude, and a lot of trial and error. that's what it means practically.

if every HVAC company buys the same AI receptionist from the same vendor, the AI stops being an advantage. the business that wins is still the one with better reviews, faster callbacks, and a guy who actually shows up on time.

living in a dorm convinced AI is going to change everything is a weird headspace to be in. like the stakes feel enormous and then your roommate asks if you want chipotle and you say yes obviously

our AI voice receptionist answers calls, qualifies leads, and books jobs a human receptionist costs $40k/year we're two 19 year olds who built this in a dorm with no CS background that second part still doesn't feel real

read something about how factory workers in 1850 didn't see it coming either. sitting here at 19 building AI tools for plumbers and wondering if we're early or just wrong

best prompt engineers i've seen aren't coders. they're people who can describe exactly what they want without leaving anything ambiguous. that's just writing.

my finance professor spent 45 minutes today on DCF models. i spent last night debugging a voice bot that answers calls for plumbing companies. i don't know which one matters more and that's genuinely unsettling.