ben hylak ✓
@benhylak · sf via pennsylvania · joined 30 Jul 2009
aspiring optimist. cto @raindrop_ai, prev @apple design, @spacex avionics
52 354Followers
2 743Following
11 432Posts total
2.6MViews on collected posts
Ultimi post
Cerchiamo nuovi post…
telling an agent "thank you" will cost u a dollar
kinda messed up tbh
4.3K views · 87 likes
· 0 reposts · 14 replies
Open on X →
@benhylak fwiw this was an impressive demo 4.5 months ago
https://t.co/9AQlKyqDIp
2.1K views · 17 likes
· 0 reposts · 0 replies
Open on X →
in march, we wrote an article that called monitoring agents one of humanity's most important problems.
at the time, that seemed like an insane thing to say.
now, it feels increasingly obvious.
🧵blog below https://t.co/9aSIItmTfa
4K views · 44 likes
· 3 reposts · 3 replies
Open on X →
@benhylak yes!
but, 1-2 years tho? :)
nothing takes that long :P
2.9K views · 14 likes
· 0 reposts · 2 replies
Open on X →
this is going to be a really big deal in ~1-2 years
258.2K views · 3.1K likes
· 100 reposts · 32 replies
Open on X →
Claude Fable 5.1 is a beast at agentic CAD!
At Max effort it's the most capable model we've tested so far.
Here, inside Adam's harness in Fusion, it rebuilt the SO-101 gripper around the stock servo and mounted a real Pi Camera Module 3 from the official STEP. https://t.co/J3Hg
511.6K views · 3.2K likes
· 220 reposts · 79 replies
Open on X →
@benhylak self-modifying code died in the 60s because you couldn't debug a program that rewrote itself. the new version works because agents can record why each change was made. the memory matters more than the modification
135 views · 3 likes
· 0 reposts · 0 replies
Open on X →
i've seen the future and it's software that can modify itself
39.6K views · 431 likes
· 23 reposts · 63 replies
Open on X →
People still underestimate what it means that software can write itself. You don’t need to become a programmer.
You need an agent like GPT or Claude, maybe some cheap hardware, and to describe what you want.
15 things you can build now with zero programming knowledge:
•
70.4K views · 33 likes
· 11 reposts · 4 replies
Open on X →
@benhylak @raindrop_ai Damn your motion design is on point! Looks awesome
1.3K views · 7 likes
· 0 reposts · 1 replies
Open on X →
stop paying for judges. starting using signals.
try the agent observability platform trusted by vercel, speak, and the fortune 100.
read more at https://t.co/TuiHnYljnO https://t.co/31BnhQIpqL
2.8K views · 38 likes
· 1 reposts · 1 replies
Open on X →
rd-signal-2 already processes around 20 billion traces a month.
members of the @raindrop_ai team:
- invented fraud transformer models at robinhood
- built the first recommender system at pinterest
- have built agents that have millions of customers
- and more
dm me to join
3K views · 30 likes
· 1 reposts · 1 replies
Open on X →
rd-signal-2 also powers @raindrop_ai's real-time issue detection. many emerging failures are detected in under 5 minutes.
because they share the same foundation, any raindrop issue can be instantly converted to a signal, compared in experiments, and more.
3.1K views · 25 likes
· 1 reposts · 1 replies
Open on X →
you can say "the agent saying it transferred to a human, but didn't call the tool" or "the user complaining that web search results were stale"
@raindrop_ai signals will write the code, assemble the context, train a custom classification head for your task, and self-improve. htt
3.8K views · 32 likes
· 1 reposts · 1 replies
Open on X →
read more at: https://t.co/TuiHnYljnO
rd-signal-2 balances precision/recall at a price that is still 280x lower than GPT-5.6 Luna xhigh.
it powers signals and real-time issue detection in raindrop, allowing our customers to detect and accurately track sparse behaviors. https:
12.7K views · 56 likes
· 2 reposts · 1 replies
Open on X →
most LLM observability providers require customers to sample a very small % of traffic (and pay for the judges!)
but agent failures are often rare. it's common for a behavioral anomaly to be < 1% of traffic.
sampling can completely hide the problem.
5K views · 35 likes
· 1 reposts · 1 replies
Open on X →
introducing rd-signal-2: a frontier classification model that is 1600x cheaper than GPT 5.6 Sol.
free to try in @raindrop_ai, and available via a new API for training/hosting custom classifiers with Zero Data Retention. https://t.co/fYvtAa8mTR
155.3K views · 994 likes
· 72 reposts · 73 replies
Open on X →
Opus 4.7 appears to be SOTA at agentic CAD design https://t.co/m1DeQRC13u
1.5M views · 5.9K likes
· 402 reposts · 133 replies
Open on X →
Rispetto ad account della stessa dimensione
21 post degli ultimi 90 giorni, accanto alla fascia di 10K–100K follower. arriva a molti, ma pochi di loro reagiscono.
Visualizzazioni mediane3 846questo account924mediana per 10K–100K
Copertura, %7.35%questo account3.62%mediana per 10K–100K
Interazione, %0.88%questo account1.52%mediana per 10K–100K
| Metrica | Questo account | Mediana per 10K–100K | Rapporto |
|---|---|---|---|
| Visualizzazioni mediane per post | 3 846 | 924 | 4.16× |
| Copertura (visualizzazioni ÷ follower) | 7.35% | 3.62% | 2.03× |
| Tasso di interazione | 0.88% | 1.52% | 0.58× |
Altri account di questa fascia → Confronta con un altro account → Come sono costruiti questi parametri →
Growth & engagement
How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.
Views per post
1.3K11 Aug
70.4K19 Aug
39.6K30 Aug
135
9331 Aug
5K
511.6K2 Sep
258.2K
2.9K
4K
2.1K
1.2K3 Sep
4.3K
3.2K
Last 14 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.
Engagement rate per post
0.64%11 Aug
0.07%19 Aug
1.34%30 Aug
2.22%
2.15%31 Aug
1.88%
0.70%2 Sep
1.25%
0.56%
1.24%
0.81%
12.21%3 Sep
2.35%
0.41%
Reactions — likes, reposts, replies and quotes — divided by views. Median for 10K–100K accounts is 1.52%.
What the audience does
Likes72.5%14 263 in total
Reposts4.3%840 in total
Replies2.1%418 in total
Quotes1.1%221 in total
Bookmarks20.0%3 937 in total
Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.
The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.
Account simili
Wirex ✓
@wirexapp
104.7K followers
· 36.9K posts
RogersHelps
@RogersHelps
104.7K followers
· 673.9K posts
Luna Promotes 🌙 ✓
@Luna_Promotes
104.7K followers
· 305 posts
Swindon Town Football Club
@Official_STFC
104.7K followers
· 73K posts
【公式】ポケモンカードチャンネル ✓
@PokecaCH
104.7K followers
· 2.4K posts
Devlet Tiyatroları
@devtiyatro
104.7K followers
· 14.6K posts
Muhammad Arif ✓
@ArifRetd
104.7K followers
· 83.6K posts
evelyn ✓
@evelyn_tokyo
104.7K followers
· 124K posts