Slingshots // Four is live!
Another batch, another fourteen projects shipping open source artifacts that I’m excited about. See below for the full list. A few themes from the batch:
1. Harness time: This quarter harnesses shot up in visibility, with increased appreciation for the parts of an agent that aren’t the model and what they can give you. This batch includes a harness supporting persistent agency (Headlong); a harness supporting self-improvement via context management (ACE); a harness supporting self-improvement of the agentic harness itself (Exo); and a harness supporting multi-agent setups (dotagents).
2. Up and down the stack: There’s innovation to be had at every layer of the stack. We saw projects worth backing at the level of kernels (Thunderkittens); inference (M*); training (Marin); data (Atlantic-CC); evals (Decision Assistants); optimization (SkyDiscover); applications (Wagglehaus), etc.
3. Talk of the town: Slingshots continue to tackle timely problems at the frontier. Concurrent with Jev taking over X, an even faster type 1 decision model (CLM) came out of academia. With RSI on the brain of every big lab researcher, we now have a public literature retrieval benchmark grounded in authors’ firsthand accounts of their research process (ScholarCatalyst). And a new RL training method (Pedagogical RL) addresses head-on the question vexing RL practitioners of how privileged information can be used to discover generalizable behaviors, not just verify.
Congrats to all the researchers in this batch and the previous three who continue to ship their research out into the world with open artifacts that can be adopted and built upon. Open science leads to better outcomes for everyone.
And the good times will continue to roll – batch five is looking just as good. If you’re a researcher with a project that’s gaining momentum and ready for an adrenaline shot, shoot me a DM or apply at https://t.co/HRUiLPRZzT.
👏 @a1zhang, @AlexKrentsel, @ankrgyl, @Azaliamirh, @bariskasikci, @bfspector, @chelseabfinn, @chipro, @Diyi_Yang, @dlwh, @drmapavone, @HazyResearch, @jackyk02, @jonasgeiping, @JonSaadFalcon, @jure, @kothasuhas, @lateinteraction, @martin_casado, @mertcemri, @michaelryan207, @MikeCafarella, @NickAJalbert, @NoahZiems, @nsagan271, @ohmyksh, @percyliang, @qizhengz_alex, @RussoMatthew, @samrmadden, @shagarw21, @ShashwatGoel7, @shulynnliu, @simran_s_arora, @souradipchakr18, @stuart_sul, @tianyu_li_, @tim_kraska, @WilliamBarrHeld, @yoonholeee, @YejinChoinka, @Stanford, @MIT, @LaudeInstitute, @CarnegieMellon, @UW, @UCBerkeley, @MPI_IS, @UWMadison
I've been thinking about the impact of AI on education and spent some time trying out the latest AI detection tools this weekend.
https://t.co/KhOQb4GmjS
Recent agentic systems (Claude Code, Codex, RLM, etc.) push context out of the prompt and into the environment (e.g., as files). This helps them maintain long-term knowledge about their goals and functionality.
🚨 While this is a good idea, we show a surprising result: systems that use external environments like this perform much better when given a small, fixed-size, in-context, agent-managed cache that "𝘱𝘦𝘦𝘬𝘴 𝘪𝘯𝘵𝘰" these environments.
🚀 Our paper, 𝗣𝗘𝗘𝗞: 𝙖 𝙨𝙮𝙨𝙩𝙚𝙢 𝙛𝙤𝙧 𝙗𝙪𝙞𝙡𝙙𝙞𝙣𝙜 𝙖𝙣𝙙 𝙢𝙖𝙞𝙣𝙩𝙖𝙞𝙣𝙞𝙣𝙜 𝗮𝗻 𝗼𝗿𝗶𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻 𝗰𝗮𝗰𝗵𝗲 𝙛𝙤𝙧 𝙇𝙇𝙈 𝙖𝙜𝙚𝙣𝙩𝙨, introduces this idea.
Compared with strong baselines, including RAG, Compaction Agents, and SOTA prompt-learning frameworks, PEEK dominates the cost–quality Pareto frontier: achieving +6.3–34.0% in quality, with fewer iterations and lower cost.
Paper: https://t.co/67pm4Dqbw5
GitHub: https://t.co/JNMehuzN9M
More in the thread below! (1/N)
📢 CfP: Conference on Innovative Data Systems Research #CIDR2027 (Amsterdam, The Netherlands). Papers due: August 4, 2026. @ViktorLeis, @samrmadden, and I are looking forward to your paper & demo submissions! https://t.co/Vf4TDqWqJO
📣 We're spreading the word about #SemBench -- a brand new benchmark for semantic query processing over multimodal workloads including text, image, audio, and tabular data!
📜Paper: https://t.co/Vai6Z3aRQK
💻Website: https://t.co/yAPaQwhagF
💾Code: https://t.co/nSBE9vEXbC
📢 CfP: Conference on Innovative Data Systems Research #CIDR2026 (Chaminade, USA). Papers due: August 5, 2025. Gustavo Alonso, @samrmadden, and I are looking forward to your paper & demo submissions! https://t.co/oOB4kPp3R3
@sirrice@sigmod@VLDBconf Just go build stuff you think is cool, and publish about it in the venue you think is the best fit. Our community will be fine; we will find fascinating ways to build new AI powered data systems!
@sirrice@sigmod@VLDBconf Thanks for this post, @sirrice . As I’m sure I’ve said to you, I’m not a huge fan of “navel gazing”. We had these same conversations in 2000, 2010, etc — and we found ways to contribute to the Internet, Big Data, Cloud, …
@joe_hellerstein@overleaf Did I miss some announcement? It looks like UCB pays for a site license? For those asking, I find for multi-author papers supporting concurrent editing is significantly better than git-based versioning. I do agree that it is worrying that papers are now archived on Overleaf.
It took three years to finish, but our follow-up to the 2006 "What Goes Around Comes Around" is finally out! Stonebraker and I examine the last 20 years in databases and discuss why relational databases + SQL will continue to remain on top.
📄PDF: https://t.co/ZwTWSxXLWb
🚨 NEW EPISODE: High Impact in Databases with... Joe Hellerstein (@joe_hellerstein) is available now!
🎧 Listen on Apple ➡️ https://t.co/02zfDQ6O5Y
🎧 Listen on Spotify ➡️ https://t.co/PgTrDblzJx
☕️ Support the podcast ➡️ https://t.co/tVHEIk5EgN