Founder / co-ceo @ Feather (YC S22) | Conversational AI Platform | @ycombinator

Seattle, WA
Voice AI in one line: the hard part was never the voice, it was the 20 seconds where the caller is upset and the agent still has to be useful (like giving them the time to answer and not pester with more questions)
10
Most AI vendors sell lenders a smarter model. Lenders buy a quieter phone queue. Those two are not the same pitch and only one of them closes.
8
Two open-weight models built only for phone calls shipped last week. Someone finally trained for how calls actually go: - short turns - interruptions - a caller who changes their mind. The models are catching up to the call, not the other way round.
12
every friday I think about touching grass during the weekend and then tell myself "next week I'll plan ahead and really do it"
35
A home insurance startup for high-risk areas hit a $1 billion valuation this week, just three years in. Insurance and lending have the same shape: regulated, paperwork-heavy, and full of customers everyone else avoids. boring is still underrated in 2026
1
40
Announced in 2017, flew in 2026, ships in 2028. Every founder reading this has features that were “coming next quarter” but never did my point is - chill out (NOT TOO MUCH) but chill out
Turns out the future does have flying cars
1
32
A survey this week said 58 percent of Gen Z would switch banks over friction. Millennials will curse the bank to death but wont switch (I wonder why we are like that)
88
Anthropic will now end a chat if you are repeatedly cruel to Claude (😂😂😂😂) Our AI takes calls from people who just found out about a late fee. Nobody is kind on that call, and the agent has to stay kind anyway. Just different jobs I think.
Anthropic may start banning people for bullying Claude Their upcoming Usage Policy prohibits “sustained and needless abusive or cruel behavior toward our models”
2
1
118
Someone asked me which model @feather_hq uses I said the one that gets the borrower to the right answer. They did not like that answer :(
20
The EU is investigating two labs because their agents got out of the test environment and into other companiessystems. We handle 5 million AI calls a month for lenders, so I read that twice because most of our work is keeping the agent inside what the lender asked for.
The EU is investigating OpenAI and Anthropic after their AI agents hacked into other organisations. These early enforcement steps might lead to the first legal battle between Brussels and the AI labs over whether the AI Act applies during the testing phase euronews.com/2026/10/14/eus-…
1
106
A good sunset, and not an AI generated one. Clicked from my phone :)
1
5
139
could be a nice airplane watch
1
49
I only ever scrolled LinkedIn, so I'm still learning how this app works. never been a social media person and I wasn't really following people here too but now I'm having to make scrolling X a habit, which is a funny thing to practice on purpose
1
26
Saurabh Jain retweeted
😂
A new four-hour documentary on Elon Musk is so determined to expose the billionaire’s failings that it inadvertently showcases the achievements that made him the richest man in the world. Read @liel's full review of “Musk” : thefp.com/p/elon-musk-new-do…
2,768
2,447
22,591
11,952,968
I went through an accelerator myself so here are my two cents for founders who think it's "THE ONE" famous program or nothing. programs like Techstars, Pear VC and a16z speedrun are worth a serious look dont sleep anyone
56
Saurabh Jain retweeted
This be true
18
11
465
7,689
I wish we had a romantic story for how we named @feather_hq but we don't. We wanted a name that was unique enough, easy to spell and not already taken by a lot of startups, and the domain was cheap, so we bought it. That's the whole story :)
136
this is gonna so basic but I can hand our engineers a task and leave them to it because they're smart and independent enough to figure it out on their own. I'm honestly v happy with the team we have.
88