Bennett Brownlow

213 posts

Bennett Brownlow banner
Bennett Brownlow

Bennett Brownlow

@bennett2b

enterprise @cursor_ai

sf Bergabung Şubat 2025
226 Mengikuti587 Pengikut
Bennett Brownlow
Bennett Brownlow@bennett2b·
Tag Cursor in slack and you too can ship magically fast
Robin Ebers · AI for Small Business@robinebers

@leerob this is honestly one of my most magical moments in the Cursor Slack whenever I find a smaller bug, a random Cursor employee just pops in, tags Cursor, and it's in the next version absolutely insane timeline we live in

English
0
3
28
1.6K
Bennett Brownlow me-retweet
Lee Robinson
Lee Robinson@leerob·
The Cursor Slack has bots solving customer issues, followed by other bots reproducing and confirming fixes. All built on our SDK!
Lee Robinson tweet media
English
73
32
1.2K
116.9K
Bennett Brownlow me-retweet
Ryo Lu
Ryo Lu@ryolu_·
here's my talk at Cursor Compile some thoughts on how we build in the age of AI and what doesn't change
English
86
117
1.8K
112.5K
Bennett Brownlow me-retweet
Cursor
Cursor@cursor_ai·
Three announcements from our keynote at Compile, including how we're training a new model with SpaceX.
English
240
760
6.7K
1.1M
Bennett Brownlow
Bennett Brownlow@bennett2b·
This is a great way to think about model selection and eval interpretation. At enterprises, users will often set and forget the most expensive model and blow through CFO’s budgets in days. Composer 2.5 has been a massive help for my customers dealing with these issues, but there there is still a place for the most expensive, most intelligent models like Fable. It’s increasingly important to use the right model for a given task.
Jediah Katz@jediahkatz

Right size your requests. Don't think of it as it "writes 5% better code," but more like it can handle the next 5% of previously unsolvable tasks. If Composer 2.5 is already doing well for your work, you don't need to use a more expensive model! But when you have a problem that's too hard you can go up

English
1
3
22
3.3K
Bennett Brownlow me-retweet
Cursor
Cursor@cursor_ai·
Claude Fable 5 is now available in Cursor. It sets a new state of the art on CursorBench at 72.9%, 8 points above the previous best.
Cursor tweet media
English
266
454
6.1K
1.2M
Bennett Brownlow me-retweet
Dwarkesh Patel
Dwarkesh Patel@dwarkesh_sp·
Recently met @srush_nlp and he started giving me an impromptu lecture on how targeted on-policy self-distillation works. I asked him if I could record it on my iPhone. The basic idea is this: if the model made a mistake at some point in the rollout (for example, calling a tool that doesn't exist), we want to discourage this specific error, but we don't want to just learn from the final reward, because it's a very noisy signal spread out over the whole trajectory. So we have another model read this trajectory and figure where the error was made. It simply inserts some hint tokens to the part of the trajectory right above where the mistake was made. Now with these injected hint tokens, have the model run a forward pass. You're not having to regenerate a new rollout - aka no new decode required. The hint causes the model to assign lower probabilities to the error tokens. You then trains the original model to match these new probabilities, teaching it to downweight that specific mistake.
English
42
174
2.5K
419.6K
BenIt Pro
BenIt Pro@BennettBuhner·
@bennett2b Could this be used sort of in a "/goal" kinda way? Haha
English
1
0
1
289
Bennett Brownlow
Bennett Brownlow@bennett2b·
Agents are getting better at figuring things out for themselves. The /loop skill uses the Cursor harness to encourage the agent to continue iterating until the task is complete. Try "/loop until this PR merges"
Jediah Katz@jediahkatz

Did you know Cursor can watch output from terminals and take action? It's very extensible. I used it to make a /loop skill, which wakes the agent up on a schedule. Try "/loop until this PR merges" or "/loop 1h check #​infra-logs for anything critical". Should I do /goal next?

English
3
0
34
2.6K
Jediah Katz
Jediah Katz@jediahkatz·
Did you know Cursor can watch output from terminals and take action? It's very extensible. I used it to make a /loop skill, which wakes the agent up on a schedule. Try "/loop until this PR merges" or "/loop 1h check #​infra-logs for anything critical". Should I do /goal next?
English
28
17
293
30.4K
Bennett Brownlow me-retweet
eric zakariasson
eric zakariasson@ericzakariasson·
cursor in slack can now read documents attached in the thread, including .txt, .log, .json, .zip, .pdf, or .docx files!
English
13
17
267
19.1K
Bennett Brownlow me-retweet
John Bai
John Bai@johnbai·
Here's a step by step: 1. Open the in-app browser to select, scribble, and describe changes to any component in design mode. 2. Type /multitask or queue changes to run them in parallel — great for touching several unrelated components at once. 3. Get your steps in, grab a coffee, text mom to just say hi.
John Bai@johnbai

It's kinda crazy that you can just use multitask design mode to make a bunch of frontend UI changes, then ask for a shared link canvas with a recap of the changes in like 5 min

English
7
14
214
25.3K
Bennett Brownlow me-retweet
xAI
xAI@xai·
Composer 2.5 is now available inside Grok Build. Composer 2.5 is a fast, highly intelligent model that excels on long-running tasks and following complex instructions.
English
604
843
7.5K
32.1M

Jelajahi

@srush_nlp @BennettBuhner @jediahkatz @elonmusk @BarackObama @taylorswift13 @cristiano @BillGates