Bobby Bayer

CTO & Co-Founder @ TrueClaim (YC W24)

[Bobby Bayer @rwbayer]

hiring for more of this rn

[zack @zack_overflow]

You just need that one ADHD person on your team who gets frustrated by every source of slowness and optimizes everything to be faster

[Bobby Bayer @rwbayer]

👀

[Claude @claudeai]

One more thing: we’re halving the price of cache reads on Claude Sonnet 5.5, to $0.10 per million tokens. That makes Sonnet 5.5 around 20% cheaper to run on most long-running work.

[Bobby Bayer @rwbayer]

The reminder we all need. It’s crazy how quickly “best practices” are internalized and then outdated in the new era.

[Nate Berkopec @nateberkopec]

Unless you have an eval showing otherwise, prefer less.

Less in your prompt, less in your skills, less documentation, less LOC, less everything. Less, less, less.

[Bobby Bayer @rwbayer]

@garrytan I had this realization recently. We have a task that models have not been able to reliably complete but has felt "close" for awhile. The new models have not missed. Still takes ~30 min per task and is too expensive at scale, but it's insane to see what's possible now

[Bobby Bayer @rwbayer]

@noahrshinn Is there a standard we can agree on for this type of interface? This would allow reach beyond the platforms you’re starting with. Seems inevitable

[Rohan Pavuluri @RPavuluri]

Maybe a hot take: if someone asks you for "a blurb" or "a forwardable note" to make an intro, you can assume it's going to be a Tier 2 Intro.

The Tier 1 Intros are when the person making the connection both (1) believes in you so much, and (2) has so much credibility with the person receiving the intro that all they need to do is say something to the effect of "you both need talk" end of story.

Better yet if they can make the intro/pitch for you of who you are and why you should talk now. If someone else is pitching you in their own words without any help from you (and at their own accord), you know that's a good sign.

The vast majority of intros are not Tier 1 Intros but Tier 2 Intros since intros often include loose ties and it takes time, often longer than a 30 min intro call, to build trust in any relationship on both sides (with people requesting an intro and with the people receiving). That's totally okay and how the world works and often better than the alternative.

But I think it's good for people to know signs of whether a better intro might be out there if getting to someone in the right way is especially important to you.

[Adit @aditabrm]

Your life as a founder will be much better once you internalize that (almost) everything that isn’t going well is ultimately your fault

[Bobby Bayer @rwbayer]

I would pay 10s of dollars for a tool that automatically threads slack replies when people forget to reply in thread. Has anyone solved this problem?

[Bobby Bayer @rwbayer]

The first version of this feature ran on a spreadsheet.

One of the first prescription drug analyses we built at TrueClaim in 2023 compared every Rx fill against the @costplusdrugs rate list. We ingested those rates manually out of a weekly email.

Now, we're integrated with their API. Every day, new pharmacy claims get compared against Cost Plus rates. If a member would save more than $5 a fill, they get a notification — with what they paid, what it would cost, and the three steps to switch.

If you run a self-funded health plan and you want to see what our tools would turn up in your own data, reach out! #selfinsured #healthplan #digitalhealth

[Bobby Bayer @rwbayer]

Fun fact: was my first YC event ever after being accepted to the 24 batch ~1 week prior

[Bobby Bayer @rwbayer]

Woah! Fun stuff - I was there for this.
Maybe in the minority, but believed it both then and now. The healthcare AI arms race is not the answer. Both employers and patients are losing right now, but this doesn’t mean it’s a long term win for providers either.

[TJ Parker⚡️ @tjparker]

I wish I had the video, but I went on a rant at @ycombinator’s Health & Bio Summit in 2023 about how AI would increase healthcare costs in the near term, specifically because providers would use it to optimize billing.

Everyone thought I was cuckoo. And yet, here we are.

[Alex Hormozi @AlexHormozi]

You can make speed the "one value to rule them all" to win.

If you solve for speed you automatically solve for:

Talent: dumb people slow
Truth: lies take u wrong way
Quality: bad shit slows brand
Work ethic: get more done faster
Focus: fewer is faster
Ownership: long term

[float @notesbyfloat]

opus 5.5 on medium effort just works man, im HOOKED

[ayman nadeem @aymannadeem]

one good diff is worth a thousand agents

[Bobby Bayer @rwbayer]

Can confirm. It is a race for legacy industries to document institutional memory and companies that already had operational excellence around documentation will get a headstart.

[signüll @signulll]

if you give opus 5.5 perfect access to a company’s email, slack, docs, browser, databases, calendar, internal tools, permissions, institutional memory, etc, plus enough reliable computer use & verification loops, & nearly all of white collar labor could already be done without any human.

it is that good. it is true agi.

[Dylan O'Sullivan @DylanoA4]

A reminder that nobody has done the reading, and if you do that alone you’ll soon find yourself absurdly ahead of the curve

[Tyler Maran @TylerMaran]

tried using jev to flag browser agents this weekend.

it runs through the activity logs live every 3 seconds and builds an average score over the session

sure you could do this without jev, but also it took a few hours and cost less than $0.01 per session

[Bobby Bayer @rwbayer]

time to short every consumer services business that doesn’t have stripe link enabled on checkout

[DHH @dhh]

Momentum opens all doors.

[dax @thdxr]

people do not understand how big the ai market is

everyone can win

well not you. but everyone else can