Introduction

Biological discovery is inherently iterative and non-linear. Progress comes through cycles of hypothesis, experimentation, refinement, and review across data, tools, and teams. At Microsoft Build 2026, Microsoft announced general availability of Microsoft Discovery to support exactly this kind of work, as a comprehensive platform for building and governing agentic AI workflows across scientific and engineering disciplines. That vision becomes especially powerful in collaboration with Ginkgo Bioworks. The goal is to enable researchers to scope and plan experiments in Microsoft Discovery and run them directly on Ginkgo Cloud Lab, without requiring in-house automation.

This collaboration brings together complementary strengths. Microsoft Discovery provides the reasoning, orchestration, and compute layer for scientific work: helping researchers turn goals into structured workflows that connect data, models, tools, and evidence. Ginkgo Bioworks contributes autonomous laboratory infrastructure that can execute those workflows and return results for analysis.

Together, this creates a lab-in-the-loop model for biological research. This workflow is traditionally described as a continuous Design–Make–Test–Analyze loop, where scientists generate an experiment plan, hand off validated protocols for lab execution, and then learn from the resulting data to inform the next step. Rather than treating experimentation as a disconnected process, the interplay between Microsoft Discovery and Ginkgo agentic system is designed to create a tighter connection between scientific reasoning and real-world validation.

Integration

One example of this tighter reasoning loop under development is an RNA design-to-data workflow. In this scenario, Microsoft Discovery uses AI agents to help plan and scope the experiment, while Ginkgo’s automated lab synthesizes DNA templates, performs in vitro transcription, purifies, and quantitates yield and purity, and returns the resulting data for downstream analysis. In addition, Ginkgo Cloud Lab provides users with full transparency on the cost of the experiment before any lab experimentation. 

Here is a walk-through of a project aiming to estimate the cost of RNA design, illustrating how customers can use Microsoft Discovery with Ginkgo RNA production offering:

1. Describe the experiment in human language


2. Provide any additional information if required


3. Validate the cost estimate of the Ginkgo Cloud Lab service


4. Place the order


5. Store the results and share

Conclusion

This is a real-world Design–Make–Test–Analyze use case that demonstrates how agentic workflows can adapt based on experimental results and accelerate R&D compared with more manual approaches. This matters because modern life sciences teams need more than isolated predictions. They need workflows that connect long term scientific context, biological data, experimental design, and validation while preserving transparency and keeping experts in control. The collaboration with Ginkgo Bioworks extends that approach into biological experimentation. It also reflects a broader principle behind Microsoft Discovery: extensibility. Microsoft Discovery is a platform that can connect Microsoft innovations with partner tools, models, and datasets. In this case, that means pairing Microsoft Discovery’s agentic orchestration with Ginkgo’s autonomous lab execution to support a more connected model for biological discovery.

“Together, agentic AI and autonomous labs will change every part of the scientific process. Iteration cycles will get faster, experiments will require less manual hands-on time, and computational analyses will become more systematic and exhaustive. By making both easier to use, Microsoft and Ginkgo aim to bring greater speed, scale and reproducibility to pre-clinical research.”

— Jason Kelly, CEO, Ginkgo Bioworks

By connecting agentic AI with autonomous experimentation, Ginkgo Bioworks and Microsoft are working toward a future in which researchers can move faster from hypothesis to insight and do so with greater speed, scale, and reproducibility.

For more information on Microsoft Discovery, click here.

For more information on Ginkgo Bioworks, click here.