Software Daily Summary
Software Daily Summary: What Happened on November 16, 2026
OpenScientist analyzes five biomedical cases in minutes
Apple and OpenEvidence launched a health AI study and app. AutoDNA demonstrations showed AI agents coordinating molecular-biology experiments, with people still handling key lab tasks.
Today in brief
OpenScientist completed analyses in minutes across five biomedical case studies, compared with months of human effort and expertise.1 Apple and OpenEvidence launched a health AI study and app, and Shanghai Jiao Tong University scientists demonstrated AutoDNA, a system that coordinates molecular-biology experiments.23
OpenScientist tackled five biomedical research case studies
Experts assessed OpenScientist across five biomedical case studies. Its analyses took minutes, compared with months of human effort and expertise.1 In Alzheimer's research, it quantified blood biomarker performance and built a plasma-proteomics survival model that predicted outcomes in a hidden evaluation cohort.1
OpenScientist proposed a repeatable mechanism linking tau pathology with changes in lysosomal acidification. Its multiple myeloma hypotheses had support in an external cohort, with weaker signals in randomized controls.1 In a blinded, prespecified proteomic comparison, its results agreed with an independent human analysis.1
OpenScientist can run in a fully air-gapped setting with open-source models, supporting work with sensitive clinical datasets.1 The authors say the system could assist human experts with biomedical data analysis.1
Apple and OpenEvidence study trust in conversational health AI
Apple and OpenEvidence launched a study and app focused on conversational health AI. The study will examine what drives consumer trust and the quality of AI responses.2 Apple could use the findings to guide product development.2
AutoDNA coordinates molecular-biology experiments, but still needs people
Scientists at Shanghai Jiao Tong University built AutoDNA, a laboratory system whose AI agents turn natural-language requests into molecular-biology experiments and coordinate instruments.3 Demonstrations included optimizing DNA synthesis and storing, then retrieving, a digital message in DNA.3
People still configure the lab, prepare reagents, replenish supplies, handle some equipment failures and interpret some results.3 The demonstrations used nucleic acids, not cells or animal models. They did not establish that inexperienced users could perform dangerous biological work.3
AutoDNA simulates instrument code before execution, which may catch operational errors but does not assess whether an experiment could be misused.3 Julius Lucks questioned whether the system has enough human oversight. Chen said safeguards should include explicit human authorization and the ability to monitor or halt experiments.3
Also published
- How Diamond Age Supported a Budget-Conscious Biotech with a Backlog of Data Diamond Age Data Science
- Toward Connected Laboratory Ecosystems GEN
- Commercial Pharma Has a Data Debt Problem. Why Minimum Payments Won't Fix It. Pharmaceutical Commerce
- Keeping Human Expertise at the Center of AI Pharmaceutical Executive
- Build, Buy, or Rent: Moving AI Beyond the Pilot Stage Pharmaceutical Executive
- How AI-Enabled Trial Matching Is Changing Who Gets Identified and Screened Applied Clinical Trials
- The Hidden Risk in Pharma’s AI Rush Pharmaceutical Executive
- Agentic AI Models and Rethinking Pharma Field Teams Pharmaceutical Executive
- The infrastructure decisions that will enable tomorrow’s biggest breakthroughs BioPharma Dive
- Your AI pilots aren’t stalling because of the AI – they’re stalling because of your environment BioPharma Dive
- Catching the signals sooner: AI’s role in proactive pharmacovigilance pharmaphorum
- Simulating tie knots What do you mean heterogeneity?