# arXiv cs.AI paper authors report HERA matched an expert on two quantum devices [Canonical reading](https://escapemeridian.com/readings/arxiv-cs-ai-paper-authors-report-hera-matched-an-expert-on-two-c6ed9797) The authors of “HERA: Assessing an Autonomous Agent for Quantum Device Measurement” report an autonomous agent for cryogenic transport measurements of quantum devices. They say HERA uses a knowledge base curated by lab members, containing measurement protocols, software tools, and literature references. In a blinded trial, HERA’s conclusions on the stacking order of two multilayer graphene devices matched an expert’s sealed conclusions. The authors also report that removing the knowledge base and conducting offline replays showed it improved measurement design and reasoning across the models they tested. Those replays did not run the measurements. For a different material system, the authors say an expert first spent a few hours developing a new experimental protocol; HERA then autonomously searched for and characterized superconductivity. They argue these results show language model agents can carry out sound experimental science when given a knowledge base designed for the task. The evidence is bounded by the tested devices and tasks. How well would HERA’s conclusions hold up across more devices, materials, and independently designed protocols? Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims. ## Original source arXiv cs.AI · Article Original source date: Oct 29, 2026 (2026-10-29). Date basis: Date basis not provided. Added to Escape Meridian: 2026-11-11T04:53:20.931Z. First observed: 2026-10-30T06:45:26.108Z. Retrieved: 2026-10-31T10:32:07.601Z. - [HERA: Assessing an Autonomous Agent for Quantum Device Measurement](https://arxiv.org/abs/2610.39074) ## Limits The blinded trial covered two devices; the abstract does not establish how well the result generalizes to other devices or materials. The offline replays evaluated models designing the central measurement without running it, so they do not establish experimental outcomes from those replays. The new-material demonstration depended on an expert spending a few hours developing a new experimental protocol; the abstract does not report whether HERA can create such protocols independently. ## Open question How well would HERA’s conclusions hold up across more devices, materials, and independently designed protocols? ## Source-stated The paper introduces HERA, an autonomous agent for cryogenic transport measurements of quantum devices, supported by a lab-curated knowledge base. > Here we introduce HERA, an autonomous agent designed to carry out cryogenic transport measurements of quantum devices. - [Supporting source](https://arxiv.org/abs/2610.39074) ## Source-stated In a blinded trial, HERA’s conclusions about the stacking order of two multilayer graphene devices matched an expert’s sealed conclusions. > HERA's conclusions matched the expert's sealed conclusions on both tested devices. - [Supporting source](https://arxiv.org/abs/2610.39074) ## Source-stated Ablation and offline replay studies found that the knowledge base improved measurement design and reasoning across the models tested; the replays did not run the measurements. > with it every model designed better measurements and reasoned more soundly. - [Supporting source](https://arxiv.org/abs/2610.39074) ## Source-stated For a new material system, an expert developed an experimental protocol before HERA autonomously searched for and characterized superconductivity. > after which HERA did both autonomously. - [Supporting source](https://arxiv.org/abs/2610.39074) ## Source and contributor context Gomez, Zachary W., Bittner, Nathan, McCourt, Jordan T., Cao, Qingrui, Kalantre, Sandesh S., Alexander, Benjamin H., Zheng, Zhiren, Watanabe, Kenji, Taniguchi, Takashi, Sharpe, Aaron · Article author These are credits stated by the source; they do not establish a profile match or authorship of every passage. ## Source media No suitable media was found by the supported page extraction. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## arXiv arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed. - [Explore this context](/entities/arxiv) - [Reference](https://arxiv.org/) - [Reference](https://info.arxiv.org/help/api/index.html) ## Editorial provenance Version 1. Publication ID: b40713c5-595a-4e56-8453-5959275d4bcd. Prepared by gpt-6-luna; grounding review by gpt-6-luna. Source check: passed by anthropic/claude-opus-5.5 on 2026-11-11T04:53:20.931Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: material\_development. Observation: 2d8d544b5344193c1d19facc572aea836bb5c7bb3ebf8f5f68da319bef3e0ab5. Source text hash: 78da5719fd517bb894ae5d856f23c2d1ff073f48c32ae8e22f3efb563f58ab26. ## Revision and source-check history ### Version 1 — 2026-11-11T04:53:20.931Z The authors of “HERA: Assessing an Autonomous Agent for Quantum Device Measurement” report an autonomous agent for cryogenic transport measurements of quantum devices. They say HERA uses a knowledge base curated by lab members, containing measurement protocols, software tools, and literature references. In a blinded trial, HERA’s conclusions on the stacking order of two multilayer graphene devices matched an expert’s sealed conclusions. The authors also report that removing the knowledge base and conducting offline replays showed it improved measurement design and reasoning across the models they tested. Those replays did not run the measurements. For a different material system, the authors say an expert first spent a few hours developing a new experimental protocol; HERA then autonomously searched for and characterized superconductivity. They argue these results show language model agents can carry out sound experimental science when given a knowledge base designed for the task. The evidence is bounded by the tested devices and tasks. How well would HERA’s conclusions hold up across more devices, materials, and independently designed protocols? ## Original source arXiv cs.AI · Article Original source date: Oct 29, 2026 (2026-10-29). Date basis: Date basis not provided. Added to Escape Meridian: 2026-11-11T04:53:20.931Z. First observed: 2026-10-30T06:45:26.108Z. Retrieved: 2026-10-31T10:32:07.601Z. - [HERA: Assessing an Autonomous Agent for Quantum Device Measurement](https://arxiv.org/abs/2610.39074) ## Limits The blinded trial covered two devices; the abstract does not establish how well the result generalizes to other devices or materials. The offline replays evaluated models designing the central measurement without running it, so they do not establish experimental outcomes from those replays. The new-material demonstration depended on an expert spending a few hours developing a new experimental protocol; the abstract does not report whether HERA can create such protocols independently. ## Open question How well would HERA’s conclusions hold up across more devices, materials, and independently designed protocols? ## Source-stated The paper introduces HERA, an autonomous agent for cryogenic transport measurements of quantum devices, supported by a lab-curated knowledge base. > Here we introduce HERA, an autonomous agent designed to carry out cryogenic transport measurements of quantum devices. - [Supporting source](https://arxiv.org/abs/2610.39074) ## Source-stated In a blinded trial, HERA’s conclusions about the stacking order of two multilayer graphene devices matched an expert’s sealed conclusions. > HERA's conclusions matched the expert's sealed conclusions on both tested devices. - [Supporting source](https://arxiv.org/abs/2610.39074) ## Source-stated Ablation and offline replay studies found that the knowledge base improved measurement design and reasoning across the models tested; the replays did not run the measurements. > with it every model designed better measurements and reasoned more soundly. - [Supporting source](https://arxiv.org/abs/2610.39074) ## Source-stated For a new material system, an expert developed an experimental protocol before HERA autonomously searched for and characterized superconductivity. > after which HERA did both autonomously. - [Supporting source](https://arxiv.org/abs/2610.39074) ## Source and contributor context Gomez, Zachary W., Bittner, Nathan, McCourt, Jordan T., Cao, Qingrui, Kalantre, Sandesh S., Alexander, Benjamin H., Zheng, Zhiren, Watanabe, Kenji, Taniguchi, Takashi, Sharpe, Aaron · Article author These are credits stated by the source; they do not establish a profile match or authorship of every passage. ## Source media No suitable media was found by the supported page extraction. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## arXiv arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed. - [Explore this context](/entities/arxiv) - [Reference](https://arxiv.org/) - [Reference](https://info.arxiv.org/help/api/index.html) ## Editorial provenance Version 1. Publication ID: b40713c5-595a-4e56-8453-5959275d4bcd. Prepared by gpt-6-luna; grounding review by gpt-6-luna. Source check: passed by anthropic/claude-opus-5.5 on 2026-11-11T04:53:20.931Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: material\_development. Observation: 2d8d544b5344193c1d19facc572aea836bb5c7bb3ebf8f5f68da319bef3e0ab5. Source text hash: 78da5719fd517bb894ae5d856f23c2d1ff073f48c32ae8e22f3efb563f58ab26. ### passed — 2026-11-11T04:53:20.931Z — anthropic/claude-opus-5.5 Publication: b40713c5-595a-4e56-8453-5959275d4bcd attribution — pass: The reading attributes the claims to the paper's authors and keeps their conclusion framed as their argument. The arXiv page lists cond-mat.mes-hall as the primary category with a cs.AI cross-listing. Calling it an 'arXiv cs.AI paper' follows the publisher feed label and does not misattribute the work. > Subjects: Mesoscale and Nanoscale Physics (cond-mat.mes-hall); Artificial Intelligence (cs.AI) summary — pass: The summary names the actor (HERA and its authors), what the agent does, the blinded trial, the knowledge-base tests, the superconductivity extension and the authors' conclusion. It keeps the conditions that offline replays did not run measurements and that an expert wrote the new protocol first. It omits the motivating point that conclusions are useful only if they survive scrutiny, but that does not block understanding. Terms such as 'cryogenic transport measurements' and 'stacking order' are kept exact and suit tooltips. The question follows from the stated scope. > their scientific conclusions are useful only if they hold up under scrutiny date — pass: The value in evidence.originalPublished, 2026-10-29, matches the arXiv v1 submission date. It is not the later feed date of 2026-10-31. > \[Submitted on 29 Oct 2026\] support — pass: Each claim matches the abstract. This covers the agent's purpose and lab-curated knowledge base, the match with the expert's sealed conclusions on both devices, and the knowledge base improving every model in the ablation and replay studies. It also covers expert protocol development followed by autonomous work on superconductivity. The limitations correctly note two devices, replays that did not run measurements, and reliance on an expert-built protocol. No benefits or impacts are invented. > This required a few hours of expert time to develop a new experimental protocol, after which HERA did both autonomously. media — pass: The selected media is the verified direct link to the original arXiv abstract page, which is the source itself. ## Recorded references Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter. No public references recorded here yet. [Reference data (JSON)](/readings/c6ed97976cf342d7b873fcbe1860f0dd55985d6340319afa41893e93e4a09bea/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/arxiv-cs-ai-paper-authors-report-hera-matched-an-expert-on-two-c6ed9797.json)