@Runtypical

building @runtypelabs • 🤠→🌉 • 🧠 APIs, evals, carcinization

 San Francisco
Joined January 2026
  • Tweets168
  • Following66
  • Followers44
18Media
Include

Only show posts containing:

Exclude

Hide posts containing:

Time range
-
Minimum likes
When all better options for integration between systems fail just ask claude to copy the data to your clipboard for you
13
I thought we were going to have to do our own finetune for this tbh
Runtype Labs
@RuntypeLabs18h
The worst agent failures don't always throw errors. We're building Runtype Trace Signals with @Cloudflare's Clairic + Clef Flash to find frustration and unanswered questions. The goal: analyze every trace, at scale, without breaking the bank. runtype.com/changelog/trace-…

Trace Signals: your agent finished. Did it actually help?

Find frustration, repeated requests, and unanswered questions in your agent traces. Cloudflare's specialized models make broad trace analysis affordable.

runtype.com
10
Replying to @RuntypeLabs @Cloudflare
Thanks @Cloudflare @michellechen
1
23
Using decision models to help AI agents in shared spaces pass the vibe check
Runtype Labs
@RuntypeLabsOct 19
Putting an AI agent behind an @mention in Slack doesn't solve the noise problem. It just moves the routing work onto your team. Someone still has to notice a message matters, remember which bot handles it, and tag it in. Here's how we're using Jev and small decision models to move beyond the @mention: runtype.com/blog/jev-in-the-…

Beyond the @mention

Decision models let channel agents listen continuously, pull in the right scrollback, and step in without waiting for an @mention.

runtype.com
1
24
Replying to @eastdakota @threepointone
I’m in
1
1,482
Finally a real benchmark zachlandes.com/trolley-probl…

Jev Meets the Trolley Problem · Zach Landes

I asked Jev and 13 other decision models the Trolley Problem.

zachlandes.com
32
Hope to do more events like this! Moving to the bay area for me was mostly about feeling like I was closer to the cutting edge when everything's changing, and sharing what I've learned feels like giving back.
Runtype Labs
@RuntypeLabsOct 6
Last night our CPO @Runtypical joined @Cloudflare and @stripe at Cloudflare's SF HQ to talk about selling to agents. The room was full of tech leaders, startup founders, and retailers. 🧵
24
Replying to @Runtypical @joelhooks @sean_j_roberts
Ha fair enough. Funnily enough thats when you fall into cold start territory… Goldilocks problem
1
1
34
Replying to @joelhooks @sean_j_roberts
D1 is for read-heavy use cases that benefit from edge replication; not for anything write heavy
2
2
35
Replying to @o_kwasniewski
@poteto did it better imo
57
Replying to @threepointone
Do you think I haven’t already automated this
1
404
Replying to @michellechen @threepointone @Cloudflare @huggingface
I was really struggling to find a decision model provider I could use in a HIPAA environment so this is awesome because I already have BAA with Cloudflare!
4
233
Replying to @michellechen
👀
60
Replying to @CompleteSkeptic
@CompleteSkeptic I want to use Jev for HIPAA/medical use cases. I can get ZDR via some of the gateways, but I need BAA. Do you plan to offer this?
46
Nathan Booker retweeted
lauren
@potetoSep 26
using opus 5.5 max fast to fix a typo
Interesting things
@awkwardgoogleSep 24
Long-range hornet nest removal by a kid
48
27
2
1,107
139,003
Replying to @irvinebroque
Team cooked with this one
1
57
Replying to @RhysSullivan
Your code mode composition makes sense from your position as a gateway, but I think there are other contexts where it’s almost essential when you have a large number of tools. Especially if you’re trying to get into plugin marketplaces that punish large numbers of tools (like >20)
46
No one is talking about Jev in Pacifica
1
1
67
Replying to @michellechen
I just figured they started running quants after the initial PR win from the new model + when they need to claw the GPUs back for training the next one
262
Replying to @rottedfuck @antirez
That’s not necessarily always a priority?
43
Load more