At Microsoft Construct, we’re saying that Microsoft Discovery is now usually accessible for all organizations, offering a complete platform for constructing and governing agentic AI workflows.
Breakthroughs in science and engineering hardly ever come from a single perception. They emerge via cycles of speculation, experimentation, refinement, and evaluation throughout groups, instruments, and information.
At present at Microsoft Construct, we’re saying that Microsoft Discovery is now usually accessible for all organizations, offering a complete platform for constructing and governing agentic AI workflows throughout scientific and engineering disciplines. We’re additionally introducing the Microsoft Discovery app in preview, an area desktop expertise that helps researchers, college students, and scientific groups start working with Microsoft Discovery immediately.
Since introducing Microsoft Discovery in personal preview at Microsoft Construct final yr, we have now labored carefully with organizations making use of AI to complicated analysis and growth (R&D) workflows. Their suggestions helped reinforce the place agentic AI must transcend particular person help, like supporting the iterative loops, proof preservation, and power coordination that outline scientific work.
Probably the most difficult issues in R&D require greater than only a immediate interface or a single mannequin response. Scientific workflows require:
- Integration with institutional data and area experience.
- Entry to specialised modeling, simulation, and evaluation instruments.
- Connection to experimental proof and validation information.
- Help for evaluation processes that form analysis choices.
A supplies scientist may have to judge efficiency, security, and value alongside manufacturability and regulatory constraints. A semiconductor group might have to discover a bigger design house with out shedding bodily constancy or traceability. A life sciences researcher may have to attach literature and experimental information with fashions and cohort-level proof earlier than deciding what to validate subsequent.
Microsoft Discovery is designed to work inside these present R&D environments, not substitute them. The platform helps specialists perceive the reasoning path behind outputs and retains human judgment on the middle of scientific and engineering choices. The final availability of Microsoft Discovery marks a major milestone in turning these necessities right into a production-ready platform for R&D environments with governance and transparency inbuilt.
How Microsoft Discovery helps R&D workflows at scale
Microsoft Discovery permits organizations to outline agentic workflows round their very own R&D applications. Groups can create and coordinate specialised brokers, join these brokers to institutional data and exterior scientific data, and orchestrate work throughout modeling, simulation, evaluation, and validation instruments.
On the middle of the platform is the Microsoft Discovery Engine, which helps the core loop of scientific work by serving to groups transfer from proof to hypotheses, via execution and evaluation, and into the subsequent iteration. This loop permits groups to maneuver past remoted evaluation towards repeatable, evidence-driven exploration, the place they’ll examine tradeoffs, query assumptions, and slim a search house in a means that may be reviewed and repeated.
As we continued product growth, we targeted on what it takes to carry agentic AI into manufacturing R&D environments:
- Workflows want to stay reproducible.
- Outputs should be reviewable.
- Proprietary data should be linked and ruled appropriately.
- Agentic methods want to suit into the working mannequin of R&D organizations.
These concerns, together with continued buyer suggestions, helped form the final availability launch and the platform capabilities behind it.
Increasing entry with the Microsoft Discovery app preview
An vital objective for Microsoft is to make superior AI and computing capabilities extra accessible to the folks engaged on a few of immediately’s most troublesome scientific and engineering challenges. Alongside the final availability of Microsoft Discovery, we’re introducing the Microsoft Discovery app accessible in preview immediately.
The Microsoft Discovery app is a localized expertise that offers researchers, college students, tutorial labs, and scientific groups an easier method to start utilizing Microsoft Discovery capabilities with out beginning with a full enterprise deployment. It’s accessible for obtain on the Microsoft Discovery GitHub and customers can get began with a GitHub Copilot account.
This preview extends Microsoft Discovery to earlier phases of exploration, the place analysis concepts start as small-team initiatives, tutorial work, or particular person investigation. The Microsoft Discovery app is designed to decrease the barrier to hands-on exploration, with a sensible entry level for literature exploration, speculation technology, scientific reasoning, and iterative experimentation.
The app lets researchers discover Microsoft Discovery capabilities utilizing their very own working atmosphere. As initiatives mature and complexity will increase, researchers and groups can carry work developed domestically into Microsoft Discovery platform to assist extra superior R&D applications.
Making use of Microsoft Discovery throughout R&D
Throughout preview, organizations helped form the trail to common availability by sharing suggestions on how they had been utilizing Microsoft Discovery to discover superior R&D workflows grounded in domain-specific information, established analysis strategies, and professional evaluation.
Companions are contributing area experience and resolution depth that may assist organizations adapt Microsoft Discovery to the instruments, information, and processes already central to their R&D work. Collectively, this work gives an early view into how Microsoft Discovery is getting used throughout domains and the way a rising ecosystem may help make complicated R&D workflows extra systematic, clear, and repeatable.
Yale Engineering
A collaboration throughout Professor David Kwabi’s group at Yale Engineering and researchers from Microsoft used the Discovery Engine to advance the frontier of agentic small molecule design for grid-scale aqueous natural redox move batteries (ORFBs).
ORFBs are promising, main candidates for sustainable, environmentally pleasant, long-duration power storage, however difficult to optimize. Electrolytes should steadiness complicated molecular properties like redox potential, aqueous solubility, artificial tractability, and electrochemical reversibility. The Discovery Engine, constructing on our cognitive loop by way of in-situ optimization analysis, permits long-horizon scientific reasoning whereas guaranteeing belief in your entire course of.
With these capabilities, the group used the agentic loop to drive in-silico exploration and convergence of candidates, interpret experimental outcomes, and suggest diagnostic experiments. Consultants at Yale Engineering led all experimental characterization, verified outcomes interpretation, and evaluated the sensible applicability of the designs. The analysis is obtainable right here.
This work introduces a robust new framework for advancing battery science with AI. By endowing an agent with the power to purpose from and adapt to experiments, we mix the strengths of human-led experimentation with AI’s capability to discover huge chemical design areas – and we’re solely starting to see what it will probably do.
—David Kwabi, Affiliate Professor, Yale
Georgia Institute of Know-how
Georgia Tech is exploring how an agentic AI system can re-evaluate the prebiotic plausibility of histidine, a biochemically vital amino acid whose emergence underneath believable prebiotic circumstances stays unclear regardless of its ubiquity in biology. Classical machine studying and AI approaches have struggled on this area as a result of lack of standardized datasets and the inherently multimodal nature of the info.
The proposed situation requires a multi-agent AI system composed of specialised AI ‘scientists’ for distinct information modalities, together with mass spectrometry evaluation, literature extraction, planetary mission information retrieval, and chemical response pathway modeling.
These brokers will collaborate via a central reasoning coordinator to combine various and heterogeneous datasets, aiming to maneuver from “absence-of-evidence” to a sturdy, evidence-based evaluation of histidine’s prebiotic viability. The framework developed can be repurposed to research different contested biosignatures, constructing a scalable pipeline for origins-of-life inquiry.
Our collaboration with the Microsoft Discovery group via the Georgia Tech AI for Analysis program has been extremely priceless, each scientifically and operationally. Working collectively on agentic AI methods to probe questions concerning the origins of life has given us early publicity to the cutting-edge embodied within the Discovery platform, whereas additionally enabling genuinely shut technical collaboration. This hands-on partnership has enabled significant bidirectional studying.
—Dr. Amirali Aghazadeh, Assistant Professor, Faculty of Electrical and Pc Engineering, Georgia Tech
Pacific Northwest Nationwide Laboratory
Microsoft and Pacific Northwest Nationwide Laboratory (PNNL) are rewriting the principles of scientific discovery, unleashing AI that doesn’t simply help researchers however orchestrates your entire discovery journey from new hypotheses to real-world experiments.
Powered by Microsoft Discovery, cutting-edge robotics and AI brokers work like a digital analysis group: imagining experiments, reasoning throughout mountains of scientific information, designing brand-new molecules, and studying on the fly from dwell laboratory outcomes at PNNL.
In power storage, this collaboration is fast-tracking the hunt for next-generation natural redox move battery supplies—breakthroughs that might slash our reliance on essential minerals like vanadium whereas offering cheaper, extra scalable power storage applied sciences that make our energy grid harder than ever.
In biosystems engineering, Microsoft Discovery is plugging straight into PNNL’s laboratory automation infrastructure to launch self-driving scientific workflows that autonomously design, run, and fine-tune organic experiments in actual time.
Collectively, Microsoft and PNNL are pioneering a brand new mannequin for science, the place robotics and autonomous laboratories fuse with AI and cloud infrastructure into one clever, closed-loop discovery engine that dramatically reduces the timeline from concepts to breakthroughs and opens the door to a brand new period of innovation in power, biology, and materials synthesis.
—Robert Runkle, Physicist and Lead for Autonomous Discovery Technique, Pacific Northwest Nationwide Laboratory
Ginkgo Bioworks
Ginkgo Bioworks and Microsoft are collaborating to carry agentic AI into organic discovery. Specialised brokers can analyze organic datasets, generate hypotheses, and design experiments to execute on an autonomous lab. Quickly, researchers will be capable of scope and plan experiments in Microsoft Discovery and run them straight on Ginkgo Cloud Lab—no in-house automation required.
Collectively, agentic AI and autonomous labs will change each a part of the scientific course of. Iteration cycles will get quicker, experiments would require much less handbook hands-on time, and computational analyses will turn out to be extra systematic and exhaustive. By making each simpler to make use of, Microsoft and Ginkgo purpose to carry higher velocity, scale and reproducibility to pre-clinical analysis.
—Jason Kelly, CEO, Ginkgo Bioworks, Inc.
Causaly
Causaly offers agentic options that compound the world’s biomedical proof with a company’s proprietary data to ship assured, traceable, cited choices at each stage, from discovery via launch.
Drug discovery doesn’t endure from a scarcity of knowledge. It suffers from a scarcity of reliable interpretation. Microsoft Discovery brings scientific computation over enterprise information, and Causaly brings the prior data, mechanistic reasoning, and provenance wanted to show these alerts into choices. Collectively, we may help researchers transfer from uncooked information to evidence-backed judgment a lot quicker and with higher confidence.
—Yiannis Kiachopoulos, Co-Founder and CEO, Causaly
Cambridge Consultants
With Microsoft Discovery, Cambridge Consultants helps reveal how AI brokers, simulation, and bodily lab methods can work collectively in a closed-loop discovery course of.
These autonomous, AI-powered cycles can flip months of experimental work into days or hours. The result’s a extra linked mannequin for R&D, one designed to speed up candidate technology, experimental planning, and real-world validation.
Microsoft Discovery has the potential to assist researchers transfer quicker from promising concepts to real-world outcomes. We see this as an vital step towards extra scalable, built-in, and clever R&D.
—Joe Corrigan, Chief Know-how Officer, Cambridge Consultants
Wiley
At each stage of the analysis and growth course of, life sciences and pharmaceutical groups want quick entry to essentially the most present, credible proof accessible. Wiley Analysis Agent: Life Sciences delivers a constantly up to date index of multiple million authoritative, high-quality, and trusted articles with hybrid search capabilities to assist superior scientific reasoning.
The agent searches, retrieves, and synthesizes related findings right into a coherent, evidence-based response to queries. It may possibly function as a stand-alone analysis service, or in orchestration with different Microsoft Discovery brokers, becoming naturally into the broader scientific reasoning workflows that Discovery permits. The Wiley Life Sciences Analysis Agent would be the first of a number of Wiley brokers supplied commercially on the Microsoft Discovery platform over time.
Scientific discovery depends upon connecting trusted proof with more and more highly effective AI methods. By bringing Wiley’s authoritative life sciences analysis into Microsoft Discovery, we may help life sciences and pharmaceutical groups speed up speculation technology, experimentation, and outcomes interpretation throughout a steady scientific reasoning loop.
—Josh Jarrett, Senior Vice President and Normal Supervisor of Utilized Analysis Intelligence at Wiley
BHP
BHP, the biggest mining firm on the earth, is utilizing Microsoft Discovery to speed up discovery of superior copper leaching options—in a matter of months as an alternative of years.
As copper demand grows and new deposits turn out to be more durable to search out and dearer to develop, enhancing restoration from present ores is a essential lever to assist meet future provide wants. This partnership has given our technical specialists the instruments they should slim an nearly infinite discipline of prospects right down to a small variety of choices that might someday be deployed in our international copper operations. We’re testing in opposition to the realities of our ore our bodies and working constraints, so we’re fixing for what can really work in observe. This reveals how expertise and human experience may be utilized collectively to resolve complicated, real-world challenges.
—Jessica Farrell, Vice President Innovation, BHP
Syensqo
Syensqo is a worldwide science firm creating groundbreaking options that improve the best way we dwell, work, journey, and play. The corporate is at present leveraging Microsoft Discovery to scale agentic AI that accelerates discovery, improves decision-making, and unlocks measurable enterprise influence, notably within the growth of next-generation warmth switch fluids for semiconductor manufacturing.
We at the moment are getting into a brand new section of our partnership with Microsoft, targeted on scaling AI brokers throughout analysis, gross sales and advertising and marketing to drive near-term development. By connecting buyer demand to scientific growth and back-to-market execution, agentic AI is enabling quicker cycles, sharper prioritization, and tangible influence on income development and enterprise efficiency.
—Mike Radossich, CEO of Syensqo
GSK
GSK, the worldwide biopharma, is working to speed up the invention, growth, and supply of medicines and vaccines to sufferers.
Working with companions like Microsoft Discovery, we see the chance to quickly iterate on candidate molecules, doubtlessly accelerating decision-making by way of fast information technology and evaluation.
—Christopher Austin, Senior Vice President, R&D Applied sciences, GSK
Microsoft Discovery is usually accessible. The Microsoft Discovery app is obtainable in preview. Preview options and capabilities are topic to alter.
