Cover image

Keep the Record, Rebuild the Rule

TL;DR for operators An AI agent can retain a complete record of what it did without permanently storing every rule it infers from that record. This distinction matters when an instruction derived from one successful task is subsequently reused in circumstances where it no longer applies. A new agent-memory architecture called Anti-Persistence keeps execution trajectories and outcomes but generates task-specific guidance only when needed, discarding that guidance afterward. Across nine benchmark–model combinations, it achieves 59.9% mean held-out success, compared with 53.5% for trajectory-only memory, 49.0% for abstraction-only memory, and 49.3% for joint memory. ...

November 7, 2026 · 8 min · Zelina
Cover image

The Facts Didn't Make It to the Decision

TL;DR for operators An AI team can possess every fact needed for a decision and still reach the wrong conclusion. TeamGym, a benchmark for distributed-information decision-making, reports an average 22.9-percentage-point loss in decision success when necessary facts are distributed among agents instead of given to one agent. Requiring those agents to retrieve the facts from source documents introduces a further 19.9-point loss.1 ...

November 7, 2026 · 10 min · Zelina
Cover image

When the Evidence Changes, Does the Agent?

TL;DR for operators An AI agent can finish a workflow correctly, explain its decisions, and recover after retries without demonstrating that it can respond appropriately when information changes. Sejin Kim and Sundong Kim propose evaluating adaptive reasoning through two complementary capabilities: transferring applicable rules across related tasks and revising decisions when new evidence invalidates an earlier assumption.1 ...

November 7, 2026 · 8 min · Zelina
Cover image

Better Commands, Missing Capabilities

TL;DR for operators An agent’s ability to complete a desktop task depends on more than the model operating it. The commands exposed by the application, the effort required to discover them, and the ability to inspect saved results can materially change performance. ANIS, an Agent-Native Interface Synthesizer, generates agent-oriented commands from existing software capabilities and validates their effects on persistent artifacts.1 On 368 desktop tasks, DeepSeek V4 Flash improves from 68.5% success with native SDK access to 74.5% with ANIS, while average online token consumption falls from 84.4K to 21.2K per task. ...

November 6, 2026 · 9 min · Zelina
Cover image

The Cost of Remembering Every Possible Answer

TL;DR for operators An agent must often retain information before knowing which question it will need to answer later. Compressing that information reduces immediate storage, but some answers may then require access to an archived record. The relevant costs include retained state, archive storage, retrieval width, decoding latency, and the questions the representation can still answer. ...

November 6, 2026 · 7 min · Zelina
Cover image

The Answer Depends on the Next Search

TL;DR for operators An internal research assistant may need to discover several technical relationships before it can answer a question. Selecting documents or graph records only once, before reasoning begins, can exclude relationships that become relevant later. Prospector1 demonstrates that this search behavior can be learned by a compact 4B-parameter language model. On 195 semiconductor questions, it achieves 83.9% single-pass accuracy, compared with 87.7% for the strongest evaluated frontier model using the same graph tools. Selecting among 32 sampled episodes with a trained verifier raises Prospector’s score to 91.3%, but adds substantial inference work. ...

November 5, 2026 · 8 min · Zelina
Cover image

Tomorrow's Answers, Yesterday's Evidence

TL;DR for operators An intelligence agent that forecasts tomorrow’s events can learn from today’s completed predictions, but only if it preserves what was knowable when each forecast was made. Otherwise, its apparent improvement may reflect access to answers revealed after the deadline. EvoForesight investigates three ways to improve such an agent: updating model parameters, retaining reusable forecasting procedures, and generating new questions around observed weaknesses. In controlled experiments, accumulated skills reduced tool calls by approximately 44% on recurring structured tasks while maintaining numeric accuracy. On heterogeneous tasks, they increased searches but improved forecasts. ...

November 5, 2026 · 9 min · Zelina
Cover image

A Successful Run Is Not a Reproduced Result

TL;DR for operators An agent can finish a scientific simulation, generate plausible figures, and still leave the published claim unverified. PaperBenchX1 tests this distinction across 93 reproduction tasks. Its leading configuration earns 59.08% average rubric credit but fully reproduces only 13 tasks (13.98%) under the benchmark’s strict scoring rule. Validation is the weakest assessed stage. For R&D leaders, the decision is not simply which agent can operate scientific software; it is which agent can regenerate the specified evidence, demonstrate agreement with the scientific target, and report what remains unresolved. These results describe scoped computational tasks, not the probability of success in scientific research generally. ...

November 3, 2026 · 6 min · Zelina
Cover image

Predicting More Can Teach an Agent Less

TL;DR for operators An assistant searching documents or navigating a shopping website encounters far more information than it needs for its next action. Training it to predict everything it observes may therefore impose unnecessary learning demands. In one reported comparison, adding full-observation prediction to reinforcement learning reduced average multi-hop question-answering accuracy from 18.88% to 17.71%. A method that selected observation content instead raised accuracy to 22.06%. ...

November 3, 2026 · 8 min · Zelina
Cover image

Saved, Still Rebuilt: The Search Cost of Reusing AI Workflows

TL;DR for operators An AI system can save a successful workflow and still spend substantial effort reconstructing how that workflow was originally assembled. Efficient reuse depends on whether the completed procedure is directly available to subsequent planning as a callable operation. FormCraft proposes a capability architecture in which primitive and composed operations share typed interfaces, retain their formation provenance, and can participate in further compatible composition. Verification determines which operations enter a persistent library. ...

November 2, 2026 · 8 min · Zelina