The past 24 hours were defined by two threads: AI moving deeper into drug development, and PyTorch loosening its ties to a single hardware vendor. OpenAI bought a drug-development coordination platform, Anthropic published lab results for Claude-designed protein binders, and Google open-sourced TorchTPU at PyTorch Conference. Meanwhile, Otter.ai landed $320 million and Gemini's personalization features arrived in the car.

1. OpenAI acquires drug-development platform Orchestra Bio

On October 20, OpenAI announced it has acquired Orchestra Bio, a platform that helps biopharma teams coordinate the drug-development process. On Orchestra, people and AI agents work together on drug programs: scientists and program leaders can delegate the monitoring and tracking of studies and campaigns to agents. OpenAI will combine Orchestra's team and technology with Rosalind, its life-sciences line that includes GPT-Rosalind and Rosalind Workbench, with the aim of building agents that connect scientific findings to the studies, decisions and next steps needed to advance a medicine. As an example, OpenAI says a failed manufacturing batch can delay the start of a clinical trial, and Orchestra's agents can identify affected activities and timelines and compare responses. Financial terms were not disclosed.

Why it matters: OpenAI's life-sciences push now extends beyond models and tools into R&D operations, complementing its partnerships with Novo Nordisk and Moderna.

Source: OpenAI

2. Google open-sources TorchTPU, bringing native PyTorch to TPUs

At PyTorch Conference 2026, Google announced that TorchTPU, its native PyTorch stack for TPUs, is now open source on GitHub. Co-developed with Meta's PyTorch team, TorchTPU lets existing PyTorch workloads run on TPUs with minimal code changes, with eager mode for debugging and torch.compile for production performance. Google also announced ecosystem integrations: in Hugging Face Transformers v5, deploying to TPUs takes device_map="tpu"; vLLM and SGLang gain TorchTPU backends; llm-d 1.0 ships with Red Hat, IBM and the community; and TorchTitan reached up to 52% MFU on Qwen 3 8B with near-linear scaling on up to 8,192 Ironwood TPUs. LiteRT, Google's on-device inference runtime, also joins the PyTorch ecosystem.

Why it matters: A day after AWS's TorchNeuron, Google is plugging its own accelerators into native PyTorch, widening developers' hardware choices beyond GPUs.

Sources: Google Cloud Blog; GitHub: google-pytorch/torch_tpu

3. Anthropic: Claude designs proteins that bind scorpion neurotoxins

Anthropic published research on October 20 in which an internal research model, operating within Claude Science and working with Citrini Research, designed minibinders against five scorpion neurotoxins from Mexico, Brazil, North Africa and the Middle East. Over 64 hours of largely autonomous work, Claude used 12 design tools to generate 318,484 designs and selected 180; the team added 31 more. Adaptyv Bio and Twist Bioscience lab-tested all 211, and 17 met the binder criteria. The two strongest, targeting the Cn2 toxin, showed single-digit nanomolar affinity. Anthropic stresses it has not tested neutralization in cells or animals, so it does not claim to have produced an antivenom, and it is releasing all 211 designs and their data.

Why it matters: It is another case of AI agents producing results validated in the wet lab, and it points to cheaper discovery for neglected diseases with little commercial investment.

Source: Anthropic Research

4. Otter.ai secures $320 million in non-dilutive capital from General Catalyst

According to TechCrunch, meeting note-taking tool Otter.ai announced on October 20 that it has raised $320 million from General Catalyst's Customer Value Fund (CVF), which provides capital without diluting equity and earns returns on the results of marketing spend. Otter says it will use the money for enterprise growth, go-to-market, international expansion and product innovation. The company says 40 million people have used its product to date, up from 25 million last year, and it crossed $100 million in ARR last year.

Why it matters: The AI meeting-assistant market is crowded (Fireflies.ai, Granola, Read AI and others), and Otter is moving from transcription toward workflow automation built on meeting data.

Source: TechCrunch

5. Gemini's Personal Intelligence comes to Android Auto

The Verge reports that Google is bringing Personal Intelligence to Gemini in the car. Once enabled, Gemini in Android Auto can draw on information from Gmail, Calendar, Maps, Photos, YouTube and Search, for example pulling flight details from a Gmail confirmation to check the flight's status, or responding to "take me to my doctor's appointment." The feature is opt-in and requires the phone to be connected to the car. It is rolling out to Android Auto now, with cars with Google built-in to follow in the coming months.

Source: The Verge

---

In one line: OpenAI and Anthropic both doubled down on AI for drug discovery, Google put PyTorch on TPUs with TorchTPU, and capital and products kept pushing into enterprise meetings and the car.