Anna Yuan retweeted
Nvidia didn't win on chips alone. It won because an army of engineers made CUDA so good that every competitor lost on software.
That advantage is now something AI can generate.
Part 2 of my conversation on what @BusinessInsider called AI rewriting Nvidia's moat. businessinsider.com/nvidia-cโฆ
๐
Anna Yuan retweeted
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
Iโve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
โข 20-200x faster
โข 40-400x cheaper (w/ output tokens free)
โข Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
Replying to @scaleagentic
@scaleagentic is officially incorporated!
A small line item on paper,
but a big moment in reality.
1
9
269
We believe the next generation of companies wonโt be run by stacks of tools.
Theyโll be run by autonomous systems that think, act, and adapt.
1
5
97
๐ ๐ง๐ผ๐ฑ๐ฎ๐โ๐ ๐๐ต๐ฒ ๐ฑ๐ฎ๐! Weโre going live at ๐ญ๐ฃ๐ ๐.๐ฃ. ๐ ๐ผ๐ฟ๐ด๐ฎ๐ป ๐๐ต๐ฎ๐๐ฒ ๐ง๐ผ๐๐ฒ๐ฟ for @Techweek_ #SFTechWeek hosting ๐๐ผ๐ป๐๐ฒ๐
๐๐๐ฎ๐น ๐๐ด๐ฒ๐ป๐๐ถ๐ฐ ๐๐ + ๐๐ง๐ ๐๐ฎ๐ฏโก
A packed afternoon of hands-on convos and live workshops with some of the brightest minds building the future of Agentic AI from @ycombinator and Inception Studio.
๐ค Can't wait to jam with::
@iamfaranna & Victor Kovalev ๐บ๐ฆ โ @scaleagentic
@conor_ai & Manu Ebert โ @hyperspell
@AlexLee611 โ @truewindai
@dariusemrani โ @scorecardai
@SDinodiya62090 โ @simplaiofficial
@n_sri_laasya โ @SoFi
@RoeyBenChaim โ @zenitysec
๐ง๐ต๐ฒ ๐ณ๐๐๐๐ฟ๐ฒ ๐ผ๐ณ ๐๐ง๐ ๐ถ๐ ๐๐ด๐ฒ๐ป๐๐ถ๐ฐ. Agents that think, remember, and act. Letโs build the #OPERATORS, not just tools. โก
1
219
We have 400+ on the waitlist, but if you can DM me why this topic will shift the course of your business, we will make room for you!
partiful.com/e/rFdbHkJS6jI5pโฆ
AI agents are everywhere. Most are toys.
Weโre building the operators โก
๐๐๐ผ๐ป๐๐ฒ๐
๐๐๐ฎ๐น ๐๐ด๐ฒ๐ป๐๐ถ๐ฐ ๐๐ + ๐๐ง๐ ๐๐ฎ๐ฏ
Oct 6, 1โ4pm @ J.P. Morgan Chase Tower
Hands-on workshop + founder-led convos:
โข GTM at scale @iamfaranna & Victor @scaleagentic
โข Memory & Context @conor_ai & Manu @ @hyperspell
โข Real-world workflows @AlexLee611 @truewindai
โข Agent Eval @dariusemrani @scorecardai
High-signal only.
๐ RSVP: partiful.com/e/rFdbHkJS6jI5pโฆ
AI agents are everywhere. most are tools.
@ScaleAgentic + @hyperspell, we built the ๐ผ๐ฝ๐ฒ๐ฟ๐ฎ๐๐ผ๐ฟ๐โก and the ๐บ๐ฎ๐ด๐ถ๐ฐ ๐๐ฎ๐ป๐ฑ. ๐ช
We are teaming up again to host one of the most exclusive sessions of @Techweek_
Join @conor_ai, Manu, and Victor for top notch crowd, intentional conversations, agents that actually work.
๐ Grab your spot: partiful.com/e/rFdbHkJS6jI5pโฆ
๐ @ycombinator and inception studio founder? DM us if you want to lead a discussion.
๐ ๐๐ผ๐ป๐๐ฒ๐
๐๐๐ฎ๐น ๐๐ด๐ฒ๐ป๐๐ถ๐ฐ ๐๐ + ๐๐ง๐ ๐๐ฎ๐ฏ
๐ ๐ข๐ฐ๐ ๐ฒ, ๐ญโ๐ฐ๐ฝ๐บ
๐ ๐.๐ฃ. ๐ ๐ผ๐ฟ๐ด๐ฎ๐ป ๐๐ต๐ฎ๐๐ฒ ๐ง๐ผ๐๐ฒ๐ฟ
not a panel. a choose-your-own-advanture lab:
โก Hands-on workshop with @scaleagentic + @hyperspell
@ycombinator & Inception Studio founders leading intimate convos on:
โข Memory & Context in Agentic AI
โข Integration into Real Workflows
โข Scaling GTM with AI agents
โข Market Intelligence & Research Agents
โข Security & Enterprise Readiness
Grab you spot: partiful.com/e/rFdbHkJS6jI5pโฆ
95
When your โpartner in crimeโ hops on a plane to make sure your data pipes donโt explode at 2amโฆ thank you, @conor_ai ๐
@hyperspell gave us structured, queryable access to our data sources overnight โ so we could stay focused on what @ScaleAgentic does best: agent reasoning and hyper-personalized GTM actions.
Couldnโt have asked for a better kind of chaos โ would do it again in a heartbeat!
Built in 48 hours
Paid pilots on day 2
Powered by @hyperspell
insane execution by @iamfaranna and team, proud to support customers like this
If foundation models are the ๐๐ฒ๐ฟ๐ฟ๐ฎ๐ฟ๐ถ๐ and ๐๐ฎ๐บ๐ฏ๐ผ๐ฟ๐ด๐ต๐ถ๐ป๐ถ๐ of AI, then @ScaleAgentic is building the F1 drivers โ trained to operate at the edge with precision and intent.
At @HumanXCo, we put that to the test.
In just 48 hours, we built, validated, and sold ๐๐ด๐ฒ๐ป๐ ๐ฆ๐ฐ๐ผ๐๐ โ our flagship ๐๐๐ฝ๐ฒ๐ฟโ๐ฃ๐ฒ๐ฟ๐๐ผ๐ป๐ฎ๐น๐ถ๐๐ฎ๐๐ถ๐ผ๐ป ๐๐ for GTM โ landing paid pilots, securing enterprise partnerships, and attracting 200+ founders and GTM teams to join the waitlist.
Read the full story linkedin.com/pulse/from-ideaโฆ
Anna Yuan retweeted
Here are 20+ examples of why Claude Code is the New Secret Weapon for developers and non-technical users. ๐งต
(save this + share with your team)

ALT Claude Code + AgentOps is the fastest way to pinpoint and fix agent failures. Claude Code lives in your terminal/IDE, maps your repo, and can diff and apply fixes; with Cmd+Esc you summon it and route a question straight to the AgentOps MCP server. Because both speak the open Model Context Protocol, Claude can call tools like get_trace or get_complete_trace to pull session replays, spans, and metrics from AgentOps in-lineโno log spelunking or context copy/paste. The result: a tight loopโobserve, ask, patch, verifyโwithout leaving the editor, and with data staying local when you use local MCP servers. AgentOps already instruments 400+ frameworks, so your debugging scales as your agent stack grows.










