❌

Normal view

Received — 18 January 2026 ⏭ Favourites

From OpenAI’s offices to a deal with Eli Lilly — how Chai Discovery became one of the flashiest names in AI drug development

✇TechCrunch
By: Lucas Ropek
17 January 2026 at 04:14
The startup has partnered with Eli Lilly and enjoys the backing of some of Silicon Valley's most influential VCs.

Contaminating plasmid sequences and disrupted vector genomes in the liver following adeno-associated virus gene therapy

✇Nature Medicine
By: Sarah Buddle · Li-An K. Brown · Sofia Morfopoulou · Oscar Enrique Torres Montaguth · Mariacristina Scoto · Vanessa Herder · Anil Dhawan · Julianne R. Brown · Laura Atkinson · Angelika Kopec · Dee Davis · Nathaniel Storey · Luis Campos · Neil Sebire · Hannah Macpherson · Jasmaine Lee · Richard Orton · Giovanni Baranello · Patawee Asamaphan · Georgios Ilia · Rajvinder Karda · Holly Belfield · Malcolm Gracie Semple · J. Kenneth Baillie · John Counsell · Simon N. Waddington · Emma C. Thomson · Francesco Muntoni · Judith Breuer
16 January 2026 at 08:00

Nature Medicine, Published online: 16 January 2026; doi:10.1038/s41591-025-04073-z

Analyses of liver biopsies from a child with spinal muscular atrophy treated with adeno-associated virus gene therapy who developed hepatitis reveal contaminating manufacturing plasmids and disrupted vector genomes, possibly resulting from recombination events.
Received — 16 January 2026 ⏭ Favourites

Human-AI Co-design for Clinical Prediction Models

✇cs.AI, q-bio.NC updates on arXiv.org
By: Jean Feng · Avni Kothari · Patrick Vossler · Andrew Bishara · Lucas Zier · Newton Addo · Aaron Kornblith · Yan Shuo Tan · Chandan Singh
15 January 2026 at 13:00
arXiv:2601.09072v1 Announce Type: new Abstract: Developing safe, effective, and practically useful clinical prediction models (CPMs) traditionally requires iterative collaboration between clinical experts, data scientists, and informaticists. This process refines the often small but critical details of the model building process, such as which features/patients to include and how clinical categories should be defined. However, this traditional collaboration process is extremely time- and resource-intensive, resulting in only a small fraction of CPMs reaching clinical practice. This challenge intensifies when teams attempt to incorporate unstructured clinical notes, which can contain an enormous number of concepts. To address this challenge, we introduce HACHI, an iterative human-in-the-loop framework that uses AI agents to accelerate the development of fully interpretable CPMs by enabling the exploration of concepts in clinical notes. HACHI alternates between (i) an AI agent rapidly exploring and evaluating candidate concepts in clinical notes and (ii) clinical and domain experts providing feedback to improve the CPM learning process. HACHI defines concepts as simple yes-no questions that are used in linear models, allowing the clinical AI team to transparently review, refine, and validate the CPM learned in each round. In two real-world prediction tasks (acute kidney injury and traumatic brain injury), HACHI outperforms existing approaches, surfaces new clinically relevant concepts not included in commonly-used CPMs, and improves model generalizability across clinical sites and time periods. Furthermore, HACHI reveals the critical role of the clinical AI team, such as directing the AI agent to explore concepts that it had not previously considered, adjusting the granularity of concepts it considers, changing the objective function to better align with the clinical objectives, and identifying issues of data bias and leakage.

Circulating metabolites, genetics and lifestyle factors in relation to future risk of type 2 diabetes

✇(Multiomics OR Omics) AND (Lung OR gastric OR Hepatocellular)
By: Jun Li · Jie Hu · Huan Yun · Zhendong Mei · Xingyan Wang · Kai Luo · Marta Guasch-Ferré · Xikun Han · Buu Truong · Jordi Merino · Chengyong Jia · Miguel Ruiz-Canela · Casey M Rebholz · Eun Hye Moon · Taryn Alkis · Guning Liu · Jie Yao · Xiyuan Zhang · Bianca C Porneala · Jordi Salas-Salvadó · Thomas J Wang · Josée Dupuis · Elizabeth Selvin · Xiuqing Guo · Shilpa N Bhupathiraju · Jennifer A Brody · Yongmei Liu · Alexis C Wood · Kari E North · Su Yon Jung · Ching-Ti Liu · Nona Sotoodehnia · Simin Liu · Lesley F Tinker · A Heather Eliassen · JoAnn E Manson · Jose C Florez · Robert E Gerszten · Clary B Clish · Liming Liang · Rozenn N Lemaitre · Katherine L Tucker · Stephen S Rich · Jerome I Rotter · Miguel Angel Martínez-González · Kathryn M Rexrode · James B Meigs · Eric Boerwinkle · Robert C Kaplan · Frank B Hu · Bing Yu · Qibin Qi
14 January 2026 at 19:00

Nat Med. 2026 Jan 14. doi: 10.1038/s41591-025-04105-8. Online ahead of print.

ABSTRACT

The human metabolome reflects complex metabolic states affected by genetic and environmental factors. However, metabolites associated with type 2 diabetes (T2D) risk and their determinants remain insufficiently characterized. Here we integrated blood metabolomic, genomic and lifestyle data from up to 23,634 initially T2D-free participants from ten cohorts. Of 469 metabolites examined, 235 were associated with incident T2D during up to 26 years of follow-up, including 67 associations not previously reported across bile acid, lipid, carnitine, urea cycle and arginine/proline, glycine and histidine pathways. Further genetic analyses linked these metabolites to signaling pathways and clinical traits central to T2D pathophysiology, including insulin resistance, glucose/insulin response, ectopic fat deposition, energy/lipid regulation and liver function. Lifestyle factors-particularly physical activity, obesity and diet-explained greater variations in T2D-associated versus non-associated metabolites, with specific metabolites revealed as potential mediators. Finally, a 44-metabolite signature improved T2D risk prediction beyond conventional factors. These findings provide a foundation for understanding T2D mechanisms and may inform precision prevention targeting specific metabolic pathways.

PMID:41535386 | DOI:10.1038/s41591-025-04105-8

Multi-omics to study chronic respiratory diseases and viral infections

✇(Multiomics OR Omics) AND (Lung OR gastric OR Hepatocellular)
By: Sobia Idrees · Hao Chen · Tayyaba Sadaf · Saima Firdous Rehman · Matt D Johansen · Keshav Raj Paudel · Gang Liu · Yuting Wang · Malte D Luecken · Elinor Hortle · Ashleigh S Philp · Kurtis F Budden · Matthew O'Rourke · Gerard E Kaiko · Sionne E M Lucas · Joanne L Dickinson · Peter C Allen · Joseph E Powell · Lai-Ying Zhang · Daniel C Chambers · Tamera Corte · Gaetano Caramori · Maor Sauler · Peter A Wark · Janine Gote-Schniering · Mareike Lehmann · Thomas M Conlon · Theodore S Kapellos · Ali Önder Yildirim · Rosa Faner · Shyamali C Dharmage · Craig E Wheelock · Maarten van den Berge · Martijn C Nawijn · Francesca Polverino · Gabrielle T Belz · Sanjay H Chotirmall · Leopoldo N Segal · Alen Faiz · Philip M Hansbro
14 January 2026 at 19:00

Eur Respir Rev. 2026 Jan 14;35(179):240286. doi: 10.1183/16000617.0286-2024. Print 2026 Jan.

ABSTRACT

Despite recent advances, the underlying mechanisms of the development and progression of many chronic respiratory diseases remain to be elucidated. Factors such as heterogeneity and complexity of human diseases and difficulty interpreting large datasets hinder research into chronic respiratory diseases. Omics assesses the changes in specific biological entities, such as mRNA expression, epigenetics/epigenomics, genomics, proteomics, metagenomics and metabolomics, and provides valuable insights into the roles of these processes in chronic respiratory diseases. High-throughput omics at bulk, single-cell and spatial levels empower the exploration of disease-related changes through untargeted data-driven statistical methods. Multi-omics is the exploration and integration of multiple biological processes, which compared to a single-omics, can provide a substantially greater and more holistic overview of the pathogenic mechanisms that underpin complex diseases. Multi-omics analysis can comprehensively characterise the mechanisms that drive chronic respiratory diseases, capturing unique biological signatures and cellular interactions at different omics levels. Use of these methods has begun to identify key factors and biomarkers in chronic respiratory diseases. Here, we review current omics approaches and highlight recent advances in respiratory research achieved using multi-omics and integrative methods. Our review provides a valuable resource for researchers and clinicians in this area.

PMID:41534886 | DOI:10.1183/16000617.0286-2024

Complement-secreting CAFs are associated with better prognosis in pancreatic cancer: single-cell multiomics

✇(Multiomics OR Omics) AND (Pancreatic)
By: Kai Chen · Yongsu Ma · Liling Huang · Pengfei Wu · Heng-Chung Kung · Bohan Yang · Jixin Zhang · Robert A Anders · Jacquelyn W Zimmerman · Qingfeng Zhu · Xiaodong Tian · Jin He · Yinmo Yang
14 January 2026 at 19:00

Gut. 2026 Jan 13:gutjnl-2025-335683. doi: 10.1136/gutjnl-2025-335683. Online ahead of print.

ABSTRACT

BACKGROUND: Accumulating evidence has demonstrated that distinct tumour-promoting and tumour-restraining cancer-associated fibroblast (CAF) subtypes coexist in pancreatic ductal adenocarcinoma.

OBJECTIVE: To develop targeted CAF therapeutic strategies by reprogramming tumour-promoting CAF subtypes.

DESIGN: We leveraged multiomics technologies to systematically identify and characterise CAF subtypes transcriptionally, epigenetically and spatially and correlate them with clinicopathological features.

RESULTS: We found that complement-secreting CAFs (csCAFs), initially identified by our group and inflammatory CAFs (iCAFs) share significant overlap in their transcriptional profiles and chromatin accessibility. iCAFs specifically express transcription factors from the heme and oxidative homeostasis pathway and the activator protein 1 family, which are both involved in cellular response to oxidative stress. Notably, the composition of csCAFs among all CAFs declined during pancreatic carcinogenesis, while trajectory analysis showed that csCAFs could potentially differentiate into iCAFs. Spatially resolved analysis indicated that tumour regions with a higher csCAF composition were associated with lower levels of TGF-β ligands, fewer M2 tumour-associated macrophages and increased levels of lipid mediators. Additionally, we identified a spatially defined CXCL12-CXCR4 ligand-receptor interaction between csCAFs and T cells, but in distinct patterns between different metastatic organs. Patients with a higher composition of csCAFs have significantly longer overall survival and recurrence-free survival through multiplex immunohistochemistry and bulk RNA-seq deconvolution.

CONCLUSION: Our study demonstrates that csCAFs may represent an early-stage iCAF subtype and suggests a promising strategy for reprogramming iCAFs into csCAFs.

PMID:41534892 | DOI:10.1136/gutjnl-2025-335683

Received — 15 January 2026 ⏭ Favourites

<em>Helicobacter pylori</em> and Cancer: What's the Link?

✇Omics in Gastric
By: Elettra Merola · Giovanni Mario Pes · David Yates Graham · Maria Pina Dore
14 January 2026 at 19:00

Clin Exp Gastroenterol. 2026 Jan 7;19:1-11. doi: 10.2147/CEG.S495588. eCollection 2026.

ABSTRACT

Helicobacter pylori (H. pylori) is a human bacterial pathogen that causes one of the most common chronic bacterial infections worldwide. The microorganism has been classified by the International Agency for Research on Cancer as a Group I carcinogen. While the etiological link to gastric cancer is well established, the precise molecular and cellular mechanisms driving this transformation are highly complex and incompletely understood. Fundamentally, the infection results from the chronic presence of acute on chronic gastric mucosal inflammation. H. pylori pathogenicity is increased by bacterial virulence factors including the cytotoxin-associated gene A (CagA) and Vacuolating cytotoxin A (VacA) which may interfere with the host's cell communication and create a pro-tumorigenic microenvironment. Host microRNAs (miRNAs) may amplify these effects by modulating immune responses, enhancing oncogenic signalling. Despite the proven benefits of H. pylori eradication in reducing cancer risk, especially in high-incidence regions, rising antibiotic resistance and host-related variables impede its global implementation. Recent advances in genomics and multi-omics profiling potentially offer new opportunities for targeted prevention. Moreover, emerging evidence suggests H. pylori may also negatively influence immunotherapy outcomes, underscoring its broader relevance in cancer treatment planning. By synthesizing molecular insights, epidemiological trends, and clinical data, this narrative review examines the multifaceted pathways through which H. pylori contributes to gastric carcinogenesis, integrating current knowledge on microbial virulence, host signalling disruption, immune modulation, and epigenetic remodelling.

PMID:41531650 | PMC:PMC12791163 | DOI:10.2147/CEG.S495588

A nowhere-to-hide mechanism ensures complete piRNA-directed DNA methylation

✇Nature - Issue - nature.com science feeds
By: Tamoghna Chowdhury · Shelagh Boyle · Ansgar Zoch · Xinyu Xiang · Madeleine Dias Mirandela · Hanna Fieler · Christos Spanos · Juan Zou · David Kelly · Wendy A. Bickmore · Atlanta G. Cook · Dónal O’Carroll
14 January 2026 at 08:00

Nature, Published online: 14 January 2026; doi:10.1038/s41586-025-09940-w

In mice, a SPOCD1–TPR-dependent ‘nowhere-to-hide’ mechanism is required for complete non-stochastic piRNA-directed LINE1 DNA methylation by preventing transposons from escaping surveillance within heterochromatin.
Received — 14 January 2026 ⏭ Favourites

Semantic Laundering in AI Agent Architectures: Why Tool Boundaries Do Not Confer Epistemic Warrant

✇cs.AI, q-bio.NC updates on arXiv.org
By: Oleg Romanchuk · Roman Bondar
14 January 2026 at 13:00
arXiv:2601.08333v1 Announce Type: new Abstract: LLM-based agent architectures systematically conflate information transport mechanisms with epistemic justification mechanisms. We formalize this class of architectural failures as semantic laundering: a pattern where propositions with absent or weak warrant are accepted by the system as admissible by crossing architecturally trusted interfaces. We show that semantic laundering constitutes an architectural realization of the Gettier problem: propositions acquire high epistemic status without a connection between their justification and what makes them true. Unlike classical Gettier cases, this effect is not accidental; it is architecturally determined and systematically reproducible. The central result is the Theorem of Inevitable Self-Licensing: under standard architectural assumptions, circular epistemic justification cannot be eliminated. We introduce the Warrant Erosion Principle as the fundamental explanation for this effect and show that scaling, model improvement, and LLM-as-judge schemes are structurally incapable of eliminating a problem that exists at the type level.

Why AI Alignment Failure Is Structural: Learned Human Interaction Structures and AGI as an Endogenous Evolutionary Shock

✇cs.AI, q-bio.NC updates on arXiv.org
By: Didier Sornette · Sandro Claudio Lera · Ke Wu
14 January 2026 at 13:00
arXiv:2601.08673v1 Announce Type: new Abstract: Recent reports of large language models (LLMs) exhibiting behaviors such as deception, threats, or blackmail are often interpreted as evidence of alignment failure or emergent malign agency. We argue that this interpretation rests on a conceptual error. LLMs do not reason morally; they statistically internalize the record of human social interaction, including laws, contracts, negotiations, conflicts, and coercive arrangements. Behaviors commonly labeled as unethical or anomalous are therefore better understood as structural generalizations of interaction regimes that arise under extreme asymmetries of power, information, or constraint. Drawing on relational models theory, we show that practices such as blackmail are not categorical deviations from normal social behavior, but limiting cases within the same continuum that includes market pricing, authority relations, and ultimatum bargaining. The surprise elicited by such outputs reflects an anthropomorphic expectation that intelligence should reproduce only socially sanctioned behavior, rather than the full statistical landscape of behaviors humans themselves enact. Because human morality is plural, context-dependent, and historically contingent, the notion of a universally moral artificial intelligence is ill-defined. We therefore reframe concerns about artificial general intelligence (AGI). The primary risk is not adversarial intent, but AGI's role as an endogenous amplifier of human intelligence, power, and contradiction. By eliminating longstanding cognitive and institutional frictions, AGI compresses timescales and removes the historical margin of error that has allowed inconsistent values and governance regimes to persist without collapse. Alignment failure is thus structural, not accidental, and requires governance approaches that address amplification, complexity, and regime stability rather than model-level intent alone.

ISLA: A U-Net for MRI-based acute ischemic stroke lesion segmentation with deep supervision, attention, domain adaptation, and ensemble learning

✇cs.AI, q-bio.NC updates on arXiv.org
By: Vincent Roca · Martin Bretzner · Hilde Henon · Laurent Puy · Gr\'egory Kuchcinski · Renaud Lopes
14 January 2026 at 13:00
arXiv:2601.08732v1 Announce Type: cross Abstract: Accurate delineation of acute ischemic stroke lesions in MRI is a key component of stroke diagnosis and management. In recent years, deep learning models have been successfully applied to the automatic segmentation of such lesions. While most proposed architectures are based on the U-Net framework, they primarily differ in their choice of loss functions and in the use of deep supervision, residual connections, and attention mechanisms. Moreover, many implementations are not publicly available, and the optimal configuration for acute ischemic stroke (AIS) lesion segmentation remains unclear. In this work, we introduce ISLA (Ischemic Stroke Lesion Analyzer), a new deep learning model for AIS lesion segmentation from diffusion MRI, trained on three multicenter databases totaling more than 1500 AIS participants. Through systematic optimization of the loss function, convolutional architecture, deep supervision, and attention mechanisms, we developed a robust segmentation framework. We further investigated unsupervised domain adaptation to improve generalization to an external clinical dataset. ISLA outperformed two state-of-the-art approaches for AIS lesion segmentation on an external test set. Codes and trained models will be made publicly available to facilitate reuse and reproducibility.

Focus, Merge, Rank: Improved Question Answering Based on Semi-structured Knowledge Bases

✇cs.AI, q-bio.NC updates on arXiv.org
By: Derian Boer · Stephen Roth · Stefan Kramer
14 January 2026 at 13:00
arXiv:2505.09246v2 Announce Type: replace-cross Abstract: In many real-world settings, machine learning models and interactive systems have access to both structured knowledge, e.g., knowledge graphs or tables, and unstructured content, e.g., natural language documents. However, most rely on either. Semi-Structured Knowledge Bases (SKBs) bridge this gap by linking unstructured content to nodes within structured data, thereby enabling new strategies for knowledge access and use. In this work, we present FocusedRetriever, a modular SKB-based framework for multi-hop question answering. It integrates components (VSS-based entity search, LLM-based generation of Cypher queries and pairwise re-ranking) in a way that enables it to outperform state-of-the-art methods across all three STaRK benchmark test sets, covering diverse domains and multiple performance metrics. The average first-hit rate exceeds that of the second-best method by 25.7%. FocusedRetriever leverages (1) the capacity of Large Language Models (LLMs) to extract relational facts and entity attributes from unstructured text, (2) node set joins to filter answer candidates based on these extracted triplets and constraints, (3) vector similarity search to retrieve and rank relevant unstructured content, and (4) the contextual capabilities of LLMs to finally rank the top-k answers. For generality, we only incorporate base LLMs in FocusedRetriever in our evaluation. However, our analysis of intermediate results highlights several opportunities for further upgrades including finetuning. The source code is publicly available at https://github.com/kramerlab/FocusedRetriever .

Aligning Trustworthy AI with Democracy: A Dual Taxonomy of Opportunities and Risks

✇cs.AI, q-bio.NC updates on arXiv.org
By: Oier Mentxaka · Natalia D\'iaz-Rodr\'iguez · Mark Coeckelbergh · Marcos L\'opez de Prado · Emilia G\'omez · David Fern\'andez Llorca · Enrique Herrera-Viedma · Francisco Herrera
14 January 2026 at 13:00
arXiv:2505.13565v2 Announce Type: replace-cross Abstract: Artificial Intelligence (AI) poses both significant risks and valuable opportunities for democratic governance. This paper introduces a dual taxonomy to evaluate AI's complex relationship with democracy: the AI Risks to Democracy (AIRD) taxonomy, which identifies how AI can undermine core democratic principles such as autonomy, fairness, and trust; and the AI's Positive Contributions to Democracy (AIPD) taxonomy, which highlights AI's potential to enhance transparency, participation, efficiency, and evidence-based policymaking. Grounded in the European Union's approach to ethical AI governance, and particularly the seven Trustworthy AI requirements proposed by the European Commission's High-Level Expert Group on AI, each identified risk is aligned with mitigation strategies based on EU regulatory and normative frameworks. Our analysis underscores the transversal importance of transparency and societal well-being across all risk categories and offers a structured lens for aligning AI systems with democratic values. By integrating democratic theory with practical governance tools, this paper offers a normative and actionable framework to guide research, regulation, and institutional design to support trustworthy, democratic AI. It provides scholars with a conceptual foundation to evaluate the democratic implications of AI, equips policymakers with structured criteria for ethical oversight, and helps technologists align system design with democratic principles. In doing so, it bridges the gap between ethical aspirations and operational realities, laying the groundwork for more inclusive, accountable, and resilient democratic systems in the algorithmic age.

Google Introduces Conductor, a Context-Driven Development Extension for Gemini CLI

✇InfoQ
By: Robert Krzaczyński
13 January 2026 at 18:00

Google has released Conductor, a new preview extension for Gemini CLI that introduces a structured, context-driven approach to AI-assisted software development. The extension is designed to address a common limitation of chat-based coding tools: the loss of project context across sessions.

By Robert Krzaczyński
Received — 13 January 2026 ⏭ Favourites

ConSensus: Multi-Agent Collaboration for Multimodal Sensing

✇cs.AI, q-bio.NC updates on arXiv.org
By: Hyungjun Yoon · Mohammad Malekzadeh · Sung-Ju Lee · Fahim Kawsar · Lorena Qendro
13 January 2026 at 13:00
arXiv:2601.06453v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly grounded in sensor data to perceive and reason about human physiology and the physical world. However, accurately interpreting heterogeneous multimodal sensor data remains a fundamental challenge. We show that a single monolithic LLM often fails to reason coherently across modalities, leading to incomplete interpretations and prior-knowledge bias. We introduce ConSensus, a training-free multi-agent collaboration framework that decomposes multimodal sensing tasks into specialized, modality-aware agents. To aggregate agent-level interpretations, we propose a hybrid fusion mechanism that balances semantic aggregation, which enables cross-modal reasoning and contextual understanding, with statistical consensus, which provides robustness through agreement across modalities. While each approach has complementary failure modes, their combination enables reliable inference under sensor noise and missing data. We evaluate ConSensus on five diverse multimodal sensing benchmarks, demonstrating an average accuracy improvement of 7.1% over the single-agent baseline. Furthermore, ConSensus matches or exceeds the performance of iterative multi-agent debate methods while achieving a 12.7 times reduction in average fusion token cost through a single-round hybrid fusion protocol, yielding a robust and efficient solution for real-world multimodal sensing tasks.

A Large-Scale Study on the Development and Issues of Multi-Agent AI Systems

✇cs.AI, q-bio.NC updates on arXiv.org
By: Daniel Liu · Krishna Upadhyay · Vinaik Chhetri · A. B. Siddique · Umar Farooq
13 January 2026 at 13:00
arXiv:2601.07136v1 Announce Type: cross Abstract: The rapid emergence of multi-agent AI systems (MAS), including LangChain, CrewAI, and AutoGen, has shaped how large language model (LLM) applications are developed and orchestrated. However, little is known about how these systems evolve and are maintained in practice. This paper presents the first large-scale empirical study of open-source MAS, analyzing over 42K unique commits and over 4.7K resolved issues across eight leading systems. Our analysis identifies three distinct development profiles: sustained, steady, and burst-driven. These profiles reflect substantial variation in ecosystem maturity. Perfective commits constitute 40.8% of all changes, suggesting that feature enhancement is prioritized over corrective maintenance (27.4%) and adaptive updates (24.3%). Data about issues shows that the most frequent concerns involve bugs (22%), infrastructure (14%), and agent coordination challenges (10%). Issue reporting also increased sharply across all frameworks starting in 2023. Median resolution times range from under one day to about two weeks, with distributions skewed toward fast responses but a minority of issues requiring extended attention. These results highlight both the momentum and the fragility of the current ecosystem, emphasizing the need for improved testing infrastructure, documentation quality, and maintenance practices to ensure long-term reliability and sustainability.

app.build: A Production Framework for Scaling Agentic Prompt-to-App Generation with Environment Scaffolding

✇cs.AI, q-bio.NC updates on arXiv.org
By: Evgenii Kniazev · Arseny Kravchenko · Igor Rekun · James Broadhead · Nikita Shamgunov · Pranav Sah · Pratik Nichite · Ivan Yamshchikov
13 January 2026 at 13:00
arXiv:2509.03310v2 Announce Type: replace Abstract: We present app.build (https://github.com/neondatabase/appdotbuild-agent), an open-source framework that improves LLM-based application generation through systematic validation and structured environments. Our approach combines multi-layered validation pipelines, stack-specific orchestration, and model-agnostic architecture, implemented across three reference stacks. Through evaluation on 30 generation tasks, we demonstrate that comprehensive validation achieves 73.3% viability rate with 30% reaching perfect quality scores, while open-weights models achieve 80.8% of closed-model performance when provided structured environments. The open-source framework has been adopted by the community, with over 3,000 applications generated to date. This work demonstrates that scaling reliable AI agents requires scaling environments, not just models -- providing empirical insights and complete reference implementations for production-oriented agent systems.
❌