Paul Copplestone and Ant Wilson on building the backend for AI agents

Behind every self-driving lab sits a human who taught a robot arm exactly where the vial rack lives, and who gets to teach it all over again the moment the experiment changes...

Professor Ali Jadbabaie describes how AI, digital transformation, climate change, and complex geopolitics are forcing engineers and policymakers to rethink how infrastructure is designed, built, and managed.

AI agents are becoming accessible to everyone, raising the bar for developers building the next generation of agent products. When capable, general-purpose agents are available out of the box, differentiation increasingly comes down to the architectural decisions and specialized experiences that make an agent uniquely valuable. In this episode, Dan Jeffries, creator of Nautilo, joins us to discuss what he's learned building a platform for personal and collaborative AI agents. We explore his approach to persistent memory, coordinating work across people and agents, managing private context, and designing deterministic tools that complement LLM capabilities. Dan also shares his views on multi-agent architectures, computer use, and the tradeoffs between local and cloud models. The conversation offers a practitioner's perspective on where the opportunities lie as AI agents become more capable, more accessible, and more difficult to differentiate. 🗒️ Full show notes: https://twimlai.com/go/782.

Since coding is solved, why code for the job?

Researchers are adapting olmOCR 2 to make historical photo libraries searchable locally and improve transcription of handwritten archaeological records.

arXiv:2610.29680v3 Announce Type: replace-cross Abstract: Using an LLM to generate training data for another is among the most general ways to train one, and it remains common practice. Some frontier teams have closed that route for their own models: what comes back is an answer, or a short summary, without the detailed reasoning trace behind it. That is a little ironic: they were themselves built on distilling human knowledge, and distilling an open-source model is still entirely routine; the restriction applies to their own outputs alone. Recent work recovers the hidden reasoning by replaying encrypted blocks across models. It somehow works, but it breaks the providers' terms of service, so it is not something a product can be built on long-term. We built the alternative. We take the context and the teacher's answer as input, and generate the reasoning trace ourselves, in the style of the student's own model family. Training is two-stage, supervised and then adversarial: a generator learns to regenerate the family's own reasoning traces from their context and answer, and a critic learns how real a reasoning trace looks. Applied to a teacher's answer, the generator writes one in the family's style. An independent judge rates how likely traces are to be genuine on a scale from 0 to 1: ours average 0.68, real traces 0.66, and traces selected by the critic from four candidates 0.82. These are authenticity scores, not correctness rates. We test the idea on five benchmarks with everything else held fixed. On general tasks the five training corpora are about the same; on coding, the generated reasoning beats Fable 5's own thinking content. With the same 5,000 questions and the same teacher answers, HumanEval+ improves by 3.7 points and MBPP+ by 7.1 at the final checkpoint, and the MBPP+ lead holds at every checkpoint. We suspect that reasoning written in the student's family style is easier for it to learn from.

arXiv:2610.28747v2 Announce Type: replace-cross Abstract: Gold's theorem shows that no learner can identify an arbitrary language in a superfinite class from positive presentations alone. We prove a positive result of a different shape. Fix an innate \emph{recursive contrast operator} $R$ mapping any string to a finite set of structurally bounded perturbations of it, and a discriminative scorer of bounded capacity trained by InfoNCE-style contrastive estimation \citep{vandenoord2018,macollins2018} on self-generated perturbations. Under a faithfulness condition relating operator, scorer, and target, the trained scorer achieves \emph{PAC conditional identification} from positive data alone: with high probability it ranks all but an $\varepsilon$-fraction of positives above every out-of-language perturbation in their own orbit, with sample complexity polynomial in the capacity, the orbit size, the score bound, and $1/\varepsilon$. The negatives are constructed by the learner, not supplied by an oracle, and no global normalizing constant over strings is ever computed. The misranking bound follows from the geometry of the contrastive loss---per-example loss floors plus a definite loss increment under misranking---yielding an $\varepsilon^{-2}$ rate, together with an explicit approximation gap that quantifies the cost of in-language (``colliding'') perturbations and vanishes for orbit-pure operators. The ranking guarantee lifts to a global membership classifier under a margin condition, and the same operator composed with the learned scorer is a normalization-free Metropolis--Hastings generator. We instantiate the hypotheses for substitutable context-free languages \citep{clarkeyraud2007} and the subregular classes \citep{heinz2010,hayashikamigaito2025}, and state an open conjecture locating the construction in the mildly context-sensitive hierarchy.

arXiv:2610.29196v2 Announce Type: replace-cross Abstract: Recent advancements in Vision-Language Models (VLMs) enable end-to-end visual grounding via language prompts and generative decoding, demonstrating largely improved zero-shot and open-vocabulary detection capabilities. As a result, VLMs are rapidly being deployed at the edge in safety-critical real-world systems. However, the physical adversarial robustness of edge VLMs has not been systematically studied, and the fundamental differences in output with modern object detectors create methodological challenges for fair evaluation. To compare the two families directly, we evaluate every selected representative model on one outcome: whether an adversarial patch worn on a person evades detection of the model. We report this by introducing a unified metric, conditional attack success rate (cASR), counted only on images the model still localizes under a matched gray occluder, and partition every failure into omission (no box), substitution (a box on the patch), and displacement (a box on neither). We benchmark the physical adversarial vulnerability of compact grounding VLMs against a range of object detectors, and find that the VLMs can be as vulnerable as the detectors. We disclose that failure mode is largely associated with the output format of a model rather than its architecture family. Of the three, only omission is visible to a downstream system; displacement and substitution return confident, well-formed boxes, making them potentially more dangerous failures. Further, we find that, under distribution shift, VLMs that primarily failed by omission instead shift to failing more by displacement/substitution. Thus, a grounding VLM can inherit the detector's vulnerability while showing the possibility to adopt more dangerous failure modes. Our code and data are available at https://github.com/ericliang1/physical-adversarial-robustness.