Realtime AI News
Thousands of AI agents sift brain-cell data for OpenAI, PerturbAI and Scripps
OpenAI has described work with PerturbAI and Scripps Research in which distributed AI agents searched biological data from millions of brain cells for new research leads. The emphasis is on orchestration at scale rather than a single model query.
OpenAI has published a case study on work with PerturbAI and Scripps Research in which the three organisations used distributed AI agents to look for new research leads in biological data drawn from millions of brain cells.
The hard part with this kind of data is not reading it but narrowing it. Single-cell measurements generate enormous, noisy and heavily correlated records, and what researchers need is not more charts but a shorter list of hypotheses worth spending bench time and reagents on.
OpenAI describes the scale of the effort in thousands of AI agents, and the operative word is distributed: analysis tasks are split across many agents working in parallel, with results folded back into the research process. That is a different proposition from pushing a dataset into a single conversation.
The division of labour is also telling. Scripps Research brings the biological questions and experimental judgement, PerturbAI sits on the data and perturbation side, and OpenAI supplies the models and the orchestration. The combination shows frontier models entering science as participants in the analysis loop rather than as writing assistants.
For life sciences, the practical appeal is saving the scarcest resources: researcher attention and wet-lab capacity. If agents can shrink the candidate space, even by discarding clearly unpromising directions early, experiments can move sooner toward hypotheses with a real chance.
The output is still a set of leads, not findings. Correlations in biological data mislead easily, and any direction a model proposes has to be validated at the bench, which is why the work is framed around promising leads rather than discoveries.
What to watch is whether the approach generalises: other data types, other institutions running similar agent orchestration, and how cost and reliability behave at larger scale. If it does, parallel agents will matter well beyond software engineering and become part of the research infrastructure itself.
Why it matters
Parallel agent orchestration is moving from coding into biological analysis, sketching a new layer of research infrastructure. The test is whether leads survive independent bench validation and replicate at other institutions.
Nearby Updates
All10/21, 13:29
Arkose finds 97.7% of legitimate AI agent sessions carry no verified ID
Security firm Arkose has found that 97.7% of legitimate AI agent sessions carry no verifiable identity, according to PPC Land. That means most working agents look indistinguishable from anonymous scripted traffic to the sites they visit.
10/21, 12:00
e-Conomy SEA report: Southeast Asia's digital economy to reach $368 billion in 2026, $1 trillion by 2035
The 11th e-Conomy SEA report from Google, Temasek and Bain projects the region's digital economy will grow 17% to $368 billion in GMV in 2026 and pass $1 trillion by 2035. It says generative AI is scaling four times faster than mobile internet, with 243 million Southeast Asians already using it.
10/21, 13:45
Yelp sells a $30-a-month AI agent that answers and qualifies leads
Yelp is charging $30 a month for an AI agent that answers inquiries from potential customers and qualifies the resulting leads, according to PPC Land. The low subscription price moves agent software out of pilots and into a line item small local businesses can buy like any other SaaS add-on.
10/21, 15:00
OpenAI-led coalition discloses critical PHP flaws and proposes an AI-era response playbook
OpenAI said its Patch the Planet initiative spent two months coordinating maintainers, researchers and security vendors to fix critical vulnerabilities in PHP, one of which was found with its frontier models while Wiz independently reported both. It also published a five-stage coordinated disclosure framework plus a PHP version checker and an upcoming Codex skill for operators.