🚀 Exciting insights from the recent World Policy Conference (WPC)! 🌍
Discover how AI agents with enhanced reasoning and planning capabilities are set to revolutionize the way we work.
https://t.co/MT82qOzu2n
#EnterpriseAI#CollaborativeAI#AIWorkforce
Wand is the operating system for the hybrid workforce, where humans and AI agents work side by side to get real work done.
Work is changing. Wand is how it happens. Here’s what we build — and why it matters 👇
https://t.co/MO3ZMcOfNc
$NBIS just dropped another successful customer story. 👌🏻
Wand AI, an innovative player in LLM optimization, is leveraging $NBIS high-performance infrastructure to reinvent how large language models reason and scale.
The goal was to reduce output length without sacrificing accuracy — all while cutting compute costs and latency.
The results:
• Shorter, more concise outputs
• Maintained or improved accuracy across STEM benchmarks
• Lower latency and more efficient scaling — even under tight inference constraints
Thanks to $NBIS, Wand AI is now delivering smarter, leaner LLMs — faster, cheaper, and with zero performance trade-offs.
We are looking for the Top 0.01% to build the future with us.
Pass our bar, and you’ll get $10,000 in cash — even if you don’t join.
We want builders building. Period.
Click for details: https://t.co/DxCBIeFBCi
@WandAI_
Here are the top AI Papers of the Week (April 7 - 13):
- NoProp
- The AI Scientist V2
- Concise Reasoning via RL
- Rethinking Reflection in Pre-Training
- Efficient KG Reasoning for Small LLMs
- Agentic Knowledgeable Self-awareness
(bookmark for later)
Read on for more:
Huge thanks to @rasbt for recognizing our latest research! We’re lucky to have some of the most brilliant minds in AI helping us push the boundaries of what's possible.
Curious how we’re making reasoning more concise?
Dive into the blog here: https://t.co/IeR3qb8PJO
As we all know by now, reasoning models often generate longer responses, which raises compute costs. Now, this new paper (https://t.co/UbBv4rzM09) shows that this behavior comes from the RL training process, not from an actual need for long answers for better accuracy. The RL loss tends to favor longer responses when the model gets negative rewards, which I think explains the "aha" moments and longer chains of thought that arise from pure RL training.
I.e., if the model gets a negative reward (i.e., the answer is wrong), the math behind PPO causes the average per-token loss becomes smaller when the response is longer. So, the model is indirectly encouraged to make its responses longer. This is true even if those extra tokens don't actually help solve the problem.
What does the response length have to do with the loss? When the reward is negative, longer responses can dilute the penalty per individual token, which results in lower (i.e., better) loss values (even though the model is still getting the answer wrong).
So the model "learns" that longer responses reduce the punishment, even though they are not helping correctness.
In addition, the researchers show that a second round of RL (using just a few problems that are sometimes solvable) can shorten responses while preserving or even improving accuracy. This has big implications for deployment efficiency.
Don’t miss INSEAD Tech Talk X as our CEO, Rotem Alaluf, discusses the future of agentic AI and autonomous organizations, alongside Professor Peter Zemsky. How will the #AIworkforce reshape enterprises and decision-making?
https://t.co/bpPs6RBPsY
#EnterpriseAI#CollaborativeAI
If you're attending the Web Summit Qatar, don’t miss the chance to connect with Jean-Paul Sacy, our Head of Middle East & Africa. With 15,000+ founders, investors, journalists & visionaries gathering in Doha, it’s the perfect place to exchange ideas and explore new opportunities.
How are humans and the AI workforce interacting? Rotem Alaluf (CEO at Wand AI), Ramesh Raskar (Associate Professor at MIT Media Lab), and Steve Nouri (CEO at GenAI) discusses critical topics that are shaping the future of human-AI collaboration. https://t.co/NSkctYQWHX
Headed to WAICF – World AI Cannes Festival? Don’t miss Philippe Chambadal as he takes the stage to share insights on the future of the AI workforce.
https://t.co/w5NzNoHhCn
#AIworkforce#AIworker#EnterpriseAI#CollaborativeAI
We’ve all heard the skepticism—“AI doesn’t really work.” But when done right, #enterpriseAI can transform how teams operate. One of our customers built an #AIworkforce within days using our platform, driving sales & increasing sales team capacity.
Curious? https://t.co/cCNo7nqEzg
🎉 Unwrapping the Joy of #AIAgents: 12 AI Use Cases to Kick Start 2025! 🎁
5️⃣ AI Agents for Insurance Risk Mitigation – Unlocking Smarter Decisions
We empower risk managers to move beyond reactive processes and into predictive, data-driven intelligence. https://t.co/bSVLapmRDz
🎉 Unwrapping the Joy of #AIAgents: 12 AI Use Cases to Kick Start 2025! 🎁
4️⃣ AI Agents for Sales Teams: Sleigh Your Sales Goals
Gift your Sales Team their own #A workforce to start boosting sales immediately. Request a demo today: https://t.co/1wT3w2h1sP
🎉 Unwrapping the Joy of #AIAgents: 12 #AI Use Cases to Kick Start 2025!
🎄 1: Finance – Deck the Books with AI Agents
Give your finance team the gift of time. AI eliminates repetitive tasks, reduces errors, and lets your humans focus on strategic tasks.
https://t.co/N0PmvVxtMx
Discover how #AI can be trained to reason critically. Our Senior AI Researcher, Allen Roush, will present his latest paper: OpenDebateEvidence: A Massive-Scale Argument Mining and Summarization Dataset—a groundbreaking resource that helps #AIagents think like champion debaters.
Our team is at #NeurIPS2024, one of the most anticipated AI and ML conferences of the year! We're excited to connect with some of the brightest minds and dive into groundbreaking research shaping the future of AI.
https://t.co/E6wIR1bHEw
What happens when AI agents learns to self-critique?
Join us at #NeurIPS2024 as Allen Roush presents OpenDebateEvidence, a groundbreaking dataset teaching AI to think like champion debaters.
https://t.co/g3LAyYxM1F
#AI#NeurIPS2024#Research#AIAgents#CollaborativeAI