PapersSelected73°9月26日 21:33

MIT CSAIL's JAZ agent framework runs on a single invoke primitive

elvis@omarsar0

Post

Why we picked it

Memory and self-improvement are usually bolted on as separate subsystems; JAZ shows both can just be code the agent writes inside its own loop, with hooks enforcing constraints and monitoring. Builders designing agent harnesses can use this to cut complexity and cost instead of adding memory layers. The paper is arXiv:2609.26891, with an explore guide at academy.dair.ai.

MIT CSAIL researchers present JAZ, an agent framework built around one primitive: invoke.

Body

The LLM writes code, can call invoke recursively, and treats all of its inputs and history as variables in the code environment.

With prompting only and no memory system, JAZ beats Letta (MemGPT) by 8% at half the cost on the recall-heavy part of StuLife.

On AppWorld, it beats ACE, a self-improvement method, by 4% at lower cost.

Original · elvis

Keep your agent harness minimal, folks. This paper from MIT CSAIL shows why. (bookmark it) While being very minimal it looks effective and promising for building self-improving agents. They present JAZ, an agent framework with one primitive, invoke. The LLM writes code, can call invoke recursively, and sees all of its inputs and history as variables in the code environment. With prompting only and no memory system, it beats Letta (MemGPT) by 8% at half the cost on the recall-heavy part of StuLife. On AppWorld it beats ACE, a self-improvement method, by 4% at lower cost. Memory and self-improvement are usually built as separate subsystems. Here both are code the agent writes inside the loop, and hooks handle the constraints and monitoring you want to enforce. Paper: arxiv.org/abs/2609.26891 Explore Paper: academy.dair.ai/papers/harness…📊 19 ⚡

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Papers" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include JAZ, MIT CSAIL, agents, Letta.

Source attribution: curated from elvis (@omarsar0).

#JAZ #MIT CSAIL #agents #Letta #open-source models

Keep reading

Recent stories with the same tag or section

Models BuildersSelected19:56

MIT open-sources JAZ framework built on a single invoke primitive

A minimal prompt-and-loop design outperforming heavyweight memory infrastructure suggests agent scaffolding is largely redundant overhead, which changes how teams should budget for agent architectures. Developers building long-horizon agents can study the JAZ repo and test whether dynamic scoping replaces their vector-store and meta-optimisation layers. Benchmarks come with explicit dollar costs, making direct comparison with Letta (MemGPT) and ACE easy to replicate.

#JAZ #MIT #agents #Letta

Gorden Sun@Gorden_SunSource

Papers BuildersSelected05:54

Arena.ai studies Interaction-as-a-Judge evaluation for computer-use agents

Evaluation methodology is quietly converging with agent capabilities: the better a model operates a GUI, the better it can grade outputs, creating a feedback loop between computer-use progress and benchmark quality. Teams building coding agents, game-generation tools, or computer-use products can download the two open trace datasets to train or validate their own judges. Compared with static human-vote arenas, interaction-based judging offers a cheaper way to scale evaluation as generated apps grow more complex.

#Arena.ai #computer-use agents #agents #AI evaluation

lmarena.ai@lmarena_aiSource

Papers BuildersSelected05:53

Arena.ai Open-Sources Datasets for Interaction-as-a-Judge Evaluation

Evaluation shifts from comparing static code screenshots to judging how agents actually behave inside an app, which changes how generation quality can be measured for web development agents. Builders of coding agents and eval pipelines can use the two open-sourced trace datasets to train reward models or benchmark correctness without collecting their own user complaints data. Compared with human-preference-only arenas, this adds an automated signal that scales with computer-use capability.

#Arena.ai #computer-use agents #agents #evaluation

lmarena.ai@lmarena_aiSource

Papers BuildersSelected05:53

Wayne Chi Explains Interaction-as-a-Judge for Agent Evaluation

The core takeaway is that evaluation signal can be harvested from real user interactions instead of hand-written test suites, shifting agent benchmarking toward live platform data. Researchers and teams building computer-use agents can apply this pattern to score traces with LLM judges or reward models trained on user complaints. Code Arena and Agent Arena already supply the battle and complaint data, giving practitioners a concrete starting point for evaluation pipelines.

#Arena #computer-use agents #evaluation #reward models

lmarena.ai@lmarena_aiSource

Papers BuildersSelected04:26

Stanford Researchers Introduce Speculative Self-Distillation for LLM Fine-Tuning

The non-obvious insight is that you do not have to choose between on-policy accuracy and training cost — selective teacher intervention gets both, plus measurably less forgetting. Developers and teams doing domain adaptation or continued fine-tuning can use SSD-style distillation to add new knowledge without degrading existing capabilities at a fraction of the token budget. Anyone benchmarking against standard on-policy distillation pipelines should compare training time and retention metrics directly.

#Speculative Self-Distillation #NeurIPS 2026 #fine-tuning #open-source models