Orchestra Reseach retweeted
AI-research-skills has finally surpassed 10k stars! It's great to see so many folks find it useful! Let us know what else should be included and suggestions to maximally speed up your research workflows.
Repo: github.com/Orchestra-Researc…
Orchestra Reseach retweeted
1/ For nearly 350 years, science has communicated itself through one object: the paper. A linear narrative, frozen as a PDF, written for a human reader. We've come to treat that format as the medium of science itself.
It doesn't have to be. It's a historical artifact. 🧵
Orchestra Reseach retweeted
Of course I have to do deep dive on this paper and do a NotebookLM Video overview for it, what an existing future.
The Last Human Written Paper: Agent Native Research Artifacts
Agent Native Research Artifact (Ara), a new digital format designed to replace traditional PDF scientific papers. Unlike a standard linear narrative, Ara is a machine executable package that preserves every stage of the scientific process, including failed experiments, code specifications, and raw evidence. This structure aims to eliminate the "storytelling tax" of traditional publishing, making it significantly easier for AI agents to reproduce, verify, and build upon existing research.
My bet: in the near future, 80%⬆️ of CS research will be done by AI in collaboration with humans. However, today's research ecosystem is still built around the human, not the AI scientist.
For example, the 8-page paper PDF is a lossy compression of months of branching exploration into a linear story, optimized for a human reviewer to skim in 30 minutes. It hides two structural taxes:
📖 Storytelling Tax — failures, rejected hypotheses, and dead ends get stripped. On RE-Bench (24,008 runs, 21 frontier models), failed runs = 90.2% of total compute cost, with a 113× median failed-to-success token ratio. Every lab independently rediscovers the same dead ends.
🔧 Engineering Tax — the gap between reviewer-sufficient prose and agent-sufficient spec. Across 8,921 PaperBench requirements (23 ICML'24 papers), only 45.4% are fully specified in the PDF. The rest is tacit lab knowledge. Tolerable when readers were human. Critical now that agents read, reproduce, and extend.
We propose ARA: the Agent-Native Research Artifact — replace the narrative PDF with an agent-executable package, in 4 layers:
🧠 structured scientific logic
⚙️ executable code w/ full specs
🌳 exploration graph (every failure preserved)
📊 evidence grounding every claim
Orchestra Reseach retweeted
My bet: in the near future, 80%⬆️ of CS research will be done by AI in collaboration with humans. However, today's research ecosystem is still built around the human, not the AI scientist.
For example, the 8-page paper PDF is a lossy compression of months of branching exploration into a linear story, optimized for a human reviewer to skim in 30 minutes. It hides two structural taxes:
📖 Storytelling Tax — failures, rejected hypotheses, and dead ends get stripped. On RE-Bench (24,008 runs, 21 frontier models), failed runs = 90.2% of total compute cost, with a 113× median failed-to-success token ratio. Every lab independently rediscovers the same dead ends.
🔧 Engineering Tax — the gap between reviewer-sufficient prose and agent-sufficient spec. Across 8,921 PaperBench requirements (23 ICML'24 papers), only 45.4% are fully specified in the PDF. The rest is tacit lab knowledge. Tolerable when readers were human. Critical now that agents read, reproduce, and extend.
We propose ARA: the Agent-Native Research Artifact — replace the narrative PDF with an agent-executable package, in 4 layers:
🧠 structured scientific logic
⚙️ executable code w/ full specs
🌳 exploration graph (every failure preserved)
📊 evidence grounding every claim
Dream bigger ✨
Orchestra handles the rest.
I’ve been talking with many PhD students lately on brainstorming research ideas
A pattern I keep seeing: when we land on an important research question, their first reaction is to list difficulties:
"Nuh, it's too hard to evaluate ..."
"it's too hard to implement ..."
That instinct made sense before. But with AI agents that execute and brainstorm at lightning speed, finding solutions is no longer the binding constraint.
Taste and ambition are.
The mindset HAS to change. Don't self-censor big ideas into incremental ones just because the easy path is visible.
Ask the important questions first — the tools will help you figure out the rest.
7
602
Orchestra Reseach retweeted
I’ve been talking with many PhD students lately on brainstorming research ideas
A pattern I keep seeing: when we land on an important research question, their first reaction is to list difficulties:
"Nuh, it's too hard to evaluate ..."
"it's too hard to implement ..."
That instinct made sense before. But with AI agents that execute and brainstorm at lightning speed, finding solutions is no longer the binding constraint.
Taste and ambition are.
The mindset HAS to change. Don't self-censor big ideas into incremental ones just because the easy path is visible.
Ask the important questions first — the tools will help you figure out the rest.
Orchestra Reseach retweeted
Last night was a blast. Pitched G @rauchg and others at the @vercel Demo Day, bumped to @dylan522p at @mintlify party, and made so many new friends excited for @orch_research 's vision of building the foundational infrastructure for science.
SF is amazing.
Orchestra Reseach retweeted
Evolved Hermes into an autonomous ML/RL Research Lab. 🧪🚀
Integrated 94 Orchestra primitives for zero-shot GRPO & distillation. 🛠️⚡️
Hermes: github.com/NousResearch/herm…
Orchestra: github.com/Orchestra-Researc…
cc @NousResearch @Teknium @JIACHENLIU8 @orch_research
Orchestra Reseach retweeted
It's always great to talk to users around the world and learning that what you built has helped them so much.
A proud founder moment.
Orchestra Reseach retweeted
Recently, I tell everyone I meet: this is the best time to do research 🔬🧑🔬
Firstly, I personally find it so rewarding because I can address my long-held curiosities real quick, and feel the pure joy of creation without infra friction. I'm able to move at the speed of thought, attempting things I never would have dared before.
Moreover, it is not just about better tools, that we can simply do things faster. But the deeper shift is an AI-native mindset 🧠. We need to break the assumptions and mental limits that constrained our imagination for years. There are so many things we couldn't even conceive of attempting before. Now we should actually think bold and go solve them.
And beyond my own experience, I see a future where everyone can be a scientist. With AI solving the execution layer, the only thing that matters is asking the right questions. Every wave of automation ⚙️ has promoted humans up the cognitive stack: blue collar, white collar, and now what I call the platinum collar, researchers.
Not the credentialed kind behind institutional walls, but anyone with a burning question and the obsession ❤️🔥 to chase it.
#AI #Research #agent @orch_research
Join our Slack channel if you want to connect with researchers and scientists from all over the world and speed up your feedback process for Orchestra!
🔗 join.slack.com/t/orchestrare…
Orchestra Reseach retweeted
Excited to learn so much groundbreaking stuff going on in AI + science right now!
Again, this is the absolute best timing for every AI-native researcher to finally do science at full speed.
We are building AI-Native Research IDE @orch_research to massively accelerate the ChatGPT moment for science… x 100🚀
Orchestra Reseach retweeted
I am traveling to SF on the week of April 13th, if you are interested in best tips for vibe research, building (human-AI)^2 collaborations for science, or wanna give us a check, you should reach out to me and @JIACHENLIU8 . We will also be at the Vercel Demo Day!
After 6 weeks of building, Vercel AI Accelerator teams are ready to show their ships.
Join us in SF on April 16 for Demo Day, where founders will pitch to @Vercel, @AWSstartups, @AnthropicAI, @OpenAI, and more.
Save your seat ↓
vercel.com/go/ai-accelerator…
We are honored to show a revolutionized way of doing scientific research at Vercel Demo Day!
After 6 weeks of building, Vercel AI Accelerator teams are ready to show their ships.
Join us in SF on April 16 for Demo Day, where founders will pitch to @Vercel, @AWSstartups, @AnthropicAI, @OpenAI, and more.
Save your seat ↓
vercel.com/go/ai-accelerator…
We addressed most of your feedbacks and Orchestra has become much better because of you.
Happy weekend!
We are processing a large volume of feedbacks. We are reading them all and will address each one of them. Thank you so much for your support of our platform and the 2-person team, please be patient.
Much love from Orchestra ♥️
So there was some Github login redirect issues and we have fixed them. Now you can login with your Github account!
And we will support importing your existing Github repos pretty soon. Stay tuned :)
Orchestra Reseach retweeted
Cursor for researchers just dropped!
We finally have an AI-powered research IDE that is built for the whole open-ended research loop: reading, brainstorming, experimenting, and iterating. Excited to try this out!
Every developer has an IDE. Researchers never had one.
We don’t need another research agent, but a full stack workstation powered by AI.
Today we're launching Orchestra @orch_research, the world's first Research IDE 🔬🧵
Orchestra Reseach retweeted
Awesome work!
In this new era of AI4Research, rapidly advancing productivity is finally unlocking human potential. It’s a great moment for researchers to truly value curiosity and imagination again.
Honestly, I hope the output-as-the-only-reward era is on its way out!
We're living in the BEST era for doing research. 💪
After I graduated from my PhD, the rise of AI-native research gave me a new chance to revisit my research experience.
Lately, doing research feels incredibly rewarding to me. I get to experience the pure joy of curiosity-driven science because I no longer have to worry about the lower-level implementations or getting bogged down by infrastructure 🚀
(I'll be sharing some of my own recent research driven by this very soon!)
But today, let me introduce the New Orchestra 🎻. We wanted to ship a product that absorbs the friction and brings science back to the curiosity.
This issue has been resolved. Agents are back to normal operations. Please ping us if issues still persist. Thank you for your understanding!



















