---
title: 'Software Daily Summary: What Happened on November 9, 2026'
headline: AI protein designs bind targets, while neuroscience agents miss most benchmark
  tasks
description: Today’s software news also included a Duke–Network Bio partnership to
  build disease models from biological and clinical data.
url: https://thebench.bio/news/software/2026-11-09
topic: software
date: '2026-11-09'
covers:
- '2026-11-09'
date_published: '2026-11-09T12:03:10Z'
date_modified: '2026-11-10T00:02:40Z'
tags:
- Software
sources: 3
publisher: TheBench.Bio
---

# Software Daily Summary: What Happened on November 9, 2026

**AI protein designs bind targets, while neuroscience agents miss most benchmark tasks**

Today’s software news also included a Duke–Network Bio partnership to build disease models from biological and clinical data.

## Today in brief

Anthropic’s AI-designed minibinders bound targets in lab tests, but only 354 of 1,320 tested designs worked. [1] Separately, Latch Bio’s neuroscience benchmark found that the best-performing AI agent configuration passed 32.2% of its evaluations. [2]

## Also worth knowing

- Network Bio and Duke University will build AI disease models using molecular profiles from biobank samples combined with anonymized patient records and treatment outcomes. [3]

## The day in numbers

- **354 of 1,320**: Claude-designed minibinder candidates that bound their targets in testing, a reported overall hit rate of 27%. [1]
- **32.2%**: The highest pass rate in NeuroscienceBench, achieved by Claude Opus 5.5 using Pi. [2]

## Also published

- [AI Can Draft the CSR Faster, But Does That Make Drug Development Faster?](<https://www.appliedclinicaltrialsonline.com/view/ai-draft-csr-faster-drug-development>), Applied Clinical Trials
- [ChemDoodle 3D updated](<https://macinchem.org/2026/11/09/chemdoodle-3d-updated/>), Mac in Chem
- [Single Partner Platforms: Enhancing the Prescription Journey for Manufacturers](<https://www.pharmexec.com/view/single-partner-platforms-enhancing-prescription-journey-manufacturers>), Pharmaceutical Executive
- [More on web browser extensions](<https://macinchem.org/2026/11/09/more-on-web-browser-extensions/>), Mac in Chem
- [The growing use of synthetic data in modern pharma](<https://pharmaphorum.com/digital/growing-use-synthetic-data-modern-pharma>), pharmaphorum
- [This Week In Cheminformatics: Issue #046](<https://thisweekincheminformatics.substack.com/p/forty-six>), This Week in Cheminformatics
- [AI beats scientists at finding errors in clinical research papers](<https://www.science.org/content/article/ai-beats-scientists-finding-errors-clinical-research-papers>), Science News

## Sources

1. [New AI Models Are Turbo-Charging Protein Design](<https://www.genengnews.com/topics/artificial-intelligence/new-ai-models-are-turbo-charging-protein-design/>), GEN, 2026-11-09
2. [NeuroscienceBench: understanding AI agents’ ability to make the right decisions from neuroscience data](<https://blog.latch.bio/p/neurosciencebench-understanding-ai>), Latch Bio, 2026-11-09
3. [Network Bio partners with Duke University on AI disease models](<https://www.pharmaceutical-technology.com/news/network-bio-duke-university-ai-disease-models/>), Pharmaceutical Technology, 2026-11-09

---

TheBench.Bio wrote this from the reporting cited above, with a language model drafting and editing.

[1]: <https://www.genengnews.com/topics/artificial-intelligence/new-ai-models-are-turbo-charging-protein-design/>
[2]: <https://blog.latch.bio/p/neurosciencebench-understanding-ai>
[3]: <https://www.pharmaceutical-technology.com/news/network-bio-duke-university-ai-disease-models/>