Blaine Korte
14.4K posts


@scottastevenson And then what happens if *every* contract's obligations are reviewed in realtime... The downstream effects get insane quite quickly.
English

The velocity of agreements is increasing rapidly. Do you feel it?
Our view at Spellbook is that we will need new global AI infrastructure for massive contract flow before the decade is out.
The speed of business is about to 10x.
And then 100x.
Zinnie Zhang@zinniezhang
@scottastevenson @rabois Agreed. My rate of signing DocuSigns has increased.
English

@rlj_law @zackbshapiro Thank you for articulating my reaction so well. The article is great, regardless of how it was typed.
Slop is such a perfect word for the useless crap that a model *can* produce, I’ll be sad if it turns into the word we label everything generated with.
English

If @zackbshapiro does even 80% of what he says he does and "wrote" it just by having Claude reflect on what it's done for him, it's still a valuable contribution
We do a disservice to ourselves by dismissing the possibility that "slop" may still contain new, valuable truth
Rob Freund@RobertFreundLaw
The Claudemaxxing article was slop, not surprising I guess.
English

@michaelkoper Great. Will give it a shot and see if it fits for us.
English

@bkorte Yes, Kamal with EC2 and CrunchyData for the managed database. I deleted the Redis dependency before end to make it easier.
I'm more of a Heroku type of person but I was surprised how easy it is right now with Kamal (with a little help of Claude Code)
English

Ok, the latest update, I promise!
I tested a real client facing proposal on production. This is the most important page of the app. It also fires a lot of queries. Here you can see the clear difference between dedicated CPUs (t3.medium) and shared CPU (Heroku 2x 1GB dyno's)

Michael Koper@michaelkoper
Ok, the latest update here. I changed availability zone to be the exact same as CrunchyData. And now the latency is just 0.3ms, which is awesome!
English

@michaelkoper Same - I migrated our db to CrunchyData.
I'm prioritizing db connection latency with alternatives. Render is the winner so far, I can locate both in US-East-1.
English
Blaine Korte retweetledi

On a 2-on-1, the defenseman covers the pass.
Polling Canada@CanadianPolling
What would you consider to be "Canadian Values"?
English

@TomaOfficer Absolutely. They can go off the rails much faster than you’d expect. I find managing the timing of compaction helps quite a bit.
Starting new threads after planning is a big help too.
But as the models and harnesses keep improving, it matters less and less.
English


@emo_eth @charlieholtz @Shpigford Wow - this worked. That linebreak was the issue.
Still have the auth loop without the env var, but I'm up and running! Thanks @emo_eth
English

@bkorte @charlieholtz @Shpigford i also ran into this. adding the CLAUDE_CODE_OAUTH_TOKEN= env var worked for me
copying the oauth token from the terminal that conductor pops up included a sneaky linebreak that i had to remove
English

@charlieholtz i'm stuck. any ideas? clicking on the claude setup-token runs me through the typical token setup and shows me the token in a terminal window, but it's not getting picked up by Conductor.
not sure what happened...can't recall ever seeing this message.

English

@charlieholtz @Shpigford Also should note that for me, setting the env var does not work.
Latest macOS, latest version of Conductor. Tried using the packaged Claude to log in, and it worked but that authorization wasnt used inside Conductor. Reinstalling didn’t work.
Any ideas?
English

@Shpigford Shoot sorry about this, thanks. Got at least one other report of this so it’s not just you! Looking into it
English

Damn. Stuck in a @conductor_build auth loop. Reinstalling does nothing.
English

@zlwaterfield @flybayer @claudeai @linear @tembo I don’t think these solutions are feasible (for me) without it using my claude account.
Plus, not having more control over how the agent runs and what commands it uses, what the branch names look like, etc, is making me rethink Linear as the agent runner right now.
English

@_swanson @markahesketh Same. I really want to use it incrementally, but need something smaller than a full page as a primitive.
English

@markahesketh yeah im really struggling to know if we should stay steady or do a claude-led migration : /
English

@scottastevenson Ever theme one of these? Man, we have it so good these days.
English


@michaelperry @NWischoff This. The days are long but the years are short.
English

@NWischoff Goes by so insanely fast. Enjoy every second of it.
-Dad of a 4 and 6 year old.
English

@ryancarson Absolutely. In the moments of downtime in life, I want to be checking in on agents doing some work.
Not doing that feels like I’m wasting time, token credit, and missing an opportunity.
It’s getting even harder to disconnect and turn off work mode.
English

I’m starting to develop bad FOMO if my agents aren’t running 24x7.
It feels like the opportunity cost is extreme.
It’s like you have a team of 100 engineers just sitting there because you haven’t given them work to do.
Not a great feeling actually.
And I’m feeling this even after shipping a feature on a Saturday.
Anyone else experiencing this?
English

@IngrahamAngle Ah yes, that maximum of 50,000 EVs priced less than $25,000 USD are so destructive.
English

Canada is trying to destroy the North American auto industry.
nytimes.com/2026/01/16/wor…
English




