-
Tweets28
-
Followers7K
-
Following1
-
Likes194
There's a world in which the most useful model for a lawyer is not the generalist model that knows how to solve math theorems and write Rust pipelines, but one that's designed to be useful just for them. We're starting to see what that could look like. Incredibly cool to see Harvey blowing the frontier models out of the water with their first post-train on very hard economically valuable tasks (2x'ing performance on Legal Agent Bench). We're proud to be working alongside them towards this vision.
A big thank you to our friends at @harvey for an awesome collaborative effort: @ItsJulioPereyra @nikogrupen @gabepereyra @spencerpoff ❤️🚀
The full post is available on our website: engram.com/blog/legal-age…
Today we're publishing our first research blog, Understanding a Law Firm through Study. We're sharing a glimpse of a future where agents are trained with native memory:
@ItsJulioPereyra Calderwood & Harkness is nothing without its people ❤️
In collaboration with @harvey, we’re excited to build a new kind of agent environment to reflect realistic knowledge work: an entire synthetic law firm, Calderwood & Harkness, with over 100M (!) tokens of documents and 250 client cases. 💼 Today’s AI models know a lot about the law, but don’t understand how the law is practiced, because this information remains proprietary within firms. Unlike how agents are benchmarked today – starting each task from scratch with a new set of context — lawyers accumulate knowledge over time, building on years of experience. Calderwood & Harkness makes it possible for agents to do the same. Legal agents do many tasks in the same environment, making it possible to leverage memory and experience to do better work over time. We’ve had a great time co-developing this benchmark with Harvey’s research team, partnering @ItsJulioPereyra @nikogrupen @gabepereyra. We’ll share more results on this soon.
Our founder @jxmnop recently issued some unfounded claims that got community noted. We deeply apologize for the confusion caused by his original post, the follow-up post, and the follow-up to the follow-up post. Nevertheless, we stand by his conviction in his own takes---and in strong open-source models like Inkling. Congrats to our friends at @thinkymachines !
people are underestimating what a big deal this is this is the ONLY open-weight model that's trained without distilling from OpenAI or Anthropic • Kimi distills • GLM distills • Qwen distills • Nemotron distills (Kimi & DeepSeek, which counts) basically a fully different
Our CEO @dan_biderman might be the best in the world in the niche intersection of leading an AI lab and cooking delicious food. We're rolling out our m̶e̶a̶t̶b̶a̶l̶l̶s̶ models to some early customers. Let us know if you want a taste!
In this episode, @EngramLab co-founder and CEO @dan_biderman joins @allenpark to cook Mediterranean meatballs with yellow rice and talk about building AI that actually learns from you: why long context, RAG, and compaction eventually break down, how Engram compresses knowledge
We are thrilled to announce Engram's pivot to dairy farming. Our primary activity will be fostering 🐐s like @jxmnop -- who recently won an ICML outstanding paper award for his work on language model memorization and capacity! We're incredibly proud to work with Jack every day, and you can too. Come join the farm 🌾
my paper won an award at icml 😁 some thoughts: • this work was rejected from NeurIPS. i cleaned it up a small amount and it got great reviews from ICML! don't give up • ICML received 24k submissions and only gives out 7 awards, which is crazy. feeling grateful • i
John and Jordi from @tbpn are in the pretraining data, but most people aren't. Frontier models train on trillions of tokens and still have no idea who you are. We're fixing that. Thanks for hosting us – and banging the gong for Engram. This meant more than you'll ever know.
Engram cofounder @jxmnop just raised $98M to build a new type of AI. He says models don't need to get smarter over time. Instead, they just need to know you better and better over time. Jack describes what he's building: "Our product is a new type of AI. We have a pretty
Models know the internet. @EngramLab believes they should know you. Congrats to @dan_biderman and the team. few have worked on continual learning longer.
Thank you to @Nasdaq for supporting Engram on our launch day yesterday! Some have commented that this photo looks AI-generated. It's not. This really happened. Feel free to send this picture to your moms. We're certainly going to.
By scaling compute on user context, we reduce token spend. But it's about more than lowering cost. To develop expertise is to reduce the energy it takes to solve a problem, freeing capacity to solve harder problems yet Was great chatting about this with the amazing @LM_Braswell!
The models we use every day are brilliant strangers. They forget your organization the moment a chat ends, then relearn it on the next query. @EngramLab fixes that. It learns your world once and reuses that memory, matching frontier systems on 1-10% of the tokens. @Microsoft,
Dan and Sabri sat down with @LM_Braswell to talk about our technical direction. Thanks for having us!
The models we use every day are brilliant strangers. They forget your organization the moment a chat ends, then relearn it on the next query. @EngramLab fixes that. It learns your world once and reuses that memory, matching frontier systems on 1-10% of the tokens. @Microsoft,
The models we use every day are brilliant strangers. They forget your organization the moment a chat ends, then relearn it on the next query. @EngramLab fixes that. It learns your world once and reuses that memory, matching frontier systems on 1-10% of the tokens. @Microsoft, @NotionHQ, and @Harvey are already testing it within their organizations. Congratulations to the team, and hear directly from @dan_biderman (CEO and co-founder) and Sabri Eyuboglu (CTO and co-founder) with @LM_Braswell ⬇️
Discussing @EngramLab's technical bets with @LM_Braswell at @kleinerperkins's Amazon room.
The models we use every day are brilliant strangers. They forget your organization the moment a chat ends, then relearn it on the next query. @EngramLab fixes that. It learns your world once and reuses that memory, matching frontier systems on 1-10% of the tokens. @Microsoft,
Today we announced our Initial Public Offering, a humble article on X. Thanks to the @NYSE for supporting us so early in our journey! And thanks to all of our lovely supporters here on X dot com for following along. More soon 🔜
Kaustubh Kukarni @Kaustubh_Kulk
0 Followers 50 Following
deb @dtdebb
0 Followers 82 Following
Rishi @Rishi_S1
0 Followers 478 Following
Neel Raja @neelrr1
14 Followers 87 Following
Mark Ally @markaally
9 Followers 68 Following
Eli Huang @EliHuangicl
0 Followers 57 Following
Kevin Hu @OldGunix
435 Followers 1K Following Am I a human dreaming to be an LLM? Or am I an LLM dreaming to be a human?
chltjsqls @lovebotcx
23 Followers 1K Following
Tom Sherry @TomSherryhp7l
0 Followers 1 Following
DarcyDonavan.x @darcydonavan
232K Followers 98K Following 🎬 Film & Television Actress 🎵 Billboard Recording Artist 🚀 Producer • 💎Owner: https://t.co/5vgiG5kVfz ⭐ Owner: https://t.co/2W80uzTbFu 🌎 Co-Owner: https://t.co/yYR1HDfsM9
NG @N_G_64
2 Followers 135 Following
Adam Marblestone @AdamMarblestone
12K Followers 8K Following Technologist, Scientist Co-founder and CEO @Convergent_FROs Tweets do not represent employers
Rajas Bansal @rajasbansal
57 Followers 224 Following
stechert @stechert
867 Followers 7K Following Wrote search @Splunk, planner/schedulers/vision/datamining @NASAJPL, robotics @UCLA. Opinions mine.
Zyyeric @zyyeric
1 Followers 325 Following
ravinder syal @ravisyal
629 Followers 7K Following Independence OS : A machine for human flourishing: making easy things easy and hard things possible, with limitless human agency as both the means and the end.
Hans Ibarra @hansibarra13
37 Followers 101 Following Building Dialogus (YC S26) Ex Youtube, Gmail, Antigravity @ Google CS @ Georgia Tech
Jude @ObservusZ
1 Followers 28 Following
Allen Z @allen_zz0
7 Followers 1K Following
Denis Efremov @DEfremov
455 Followers 2K Following Tech Investor, Principal @r136ventures, ex-@Seedstars, ex-@McKinsey, Author at https://t.co/xTIjAvQQhE, 1x husband, 2x dad ❤️ b2b & fintech
Billy Vythikowski @vythikowski
4 Followers 530 Following
Haozhe Zhang @HaozheZhang666
77 Followers 430 Following 3D by coding Building BenchCAD, professional 3D modeling benchmark, adopted by @AnthropicAI and @OpenAI
Bards @ViswapriyaM
74 Followers 2K Following “The best life is the one in which the creative impulses play the largest part and the possessive impulses the smallest.” — Bertrand Russell
Ayush Chandekar @0xbl4ze
216 Followers 383 Following Inference | GSoC'25@Git | Dev @sdslabs | IITR'27 | Hobbyist
Daniel P Jeong @danielpjeong
164 Followers 467 Following PhD student @mldcmu @SCSatCMU + intern @AbridgeHQ | ML for Healthcare | Previously @MSFTResearch, @columbia, @nasajpl
\nabla @_gradient_flow_
8 Followers 516 Following
Mohammadreza Mohseni @mrm_196
98 Followers 650 Following Enabling Large-Scale AI Training @ Google Vertex AI
Ajay Dheeraj @ajay_dheeraj
9 Followers 748 Following
Alvaro Alliende @Aalliende
12K Followers 9K Following Director de Empresas y Consultor. Finanzas, Management, Int. Artificial,FinTech, Medios de Pago et al. MBA UCLA, Ing. Matemática.
papuna veltauri @Shampunai
1 Followers 31 Following
Adolfo Gutierrez @donnetox
55 Followers 1K Following
Catherine @soporifictree
0 Followers 38 Following
Timmy Ghiurau @itzik009
2K Followers 2K Following AI × XR × Culture | Co-founder @MidBrainAi and @thepointlabs | ex-@volvocars innovation lead |Teaching machines to make sense. Nerd by day rock star by night 🎸
turja🕺 @turjchowder
1K Followers 938 Following building village labs, @ycombinator and tinkering on side quests
eshock_hkh @Eshock_hkh
5 Followers 325 Following
Evan Engram @eazyengram
77K Followers 494 FollowingTrends for United States
You might like
