Catalyst Grant for Agentic Research Workflows

Deadline: 05-Oct-2026

The Catalyst Grant 2026 is an international initiative by Digital Science supporting early-stage startups, innovators, research teams, students, and individuals developing trustworthy artificial intelligence tools for research workflows. The 2026 call focuses on agentic workflows, where AI-powered systems do more than assist researchers. These tools can autonomously perform multiple steps within research processes while incorporating trust, provenance, transparency, and governance. The most promising ideas can receive up to £25,000, or approximately $31,800 USD, to advance solutions that can contribute to scientific progress and societal impact.

About the Catalyst Grant

The Catalyst Grant is designed to accelerate innovation across research by supporting the development of new software tools and technologies. The 2026 programme places particular emphasis on AI systems capable of operating across multiple stages of the research lifecycle. Rather than focusing only on isolated AI functions, the programme seeks solutions that can support or autonomously perform connected research tasks. A key requirement is that these agentic workflows should be built around trustworthy AI, with appropriate attention to provenance and governance. What Are Agentic Research Workflows? Agentic research workflows use AI systems that can take action across multiple steps of a research process rather than simply responding to individual prompts. Potential applications can span areas such as: - Research writing - Research discovery - Evidence synthesis - Peer review - Funding decisions - Research integrity - Manuscript screening - Drug discovery - Reviewer matching The Catalyst Grant is particularly interested in solutions where autonomous or semi-autonomous AI activity is accompanied by transparency, traceability, and appropriate governance. Examples of Solutions the Grant May Support The programme highlights several types of potential applications.

Examples include:

- AI co-writers that maintain transparent contribution logs. - Regulatory-grade evidence synthesis agents. - Explainable tools for drug discovery. - Automated systems for reviewer matching. - Auditable manuscript screening systems. These examples illustrate the broader goal of developing AI that can operate within research workflows while maintaining accountability and trust. How Much Funding Is Available? The Catalyst Grant offers an award of up to £25,000, approximately $31,800 USD. The funding is intended for the most promising early-stage ideas that can advance science and generate societal impact through trustworthy, agentic AI workflows. Applicants do not need to have a fully developed commercial product to be considered. Who Can Apply? The Catalyst Grant has broad international eligibility. It is open to anyone with an early-stage idea, including: - Startup founders - Research teams - Students - Individual innovators - Early-stage research projects Applicants can be based anywhere in the world. The programme is therefore suitable for people at different stages of developing an AI-based research solution, provided the proposed idea fits the focus of the 2026 call. Do Applicants Need a Finished Product? No. Applicants do not need to have a completed product or established business.

The programme can consider:

- Well-formed concepts - Early prototypes - Working products - Early-stage research projects Applicants also do not need to have revenue. This makes the Catalyst Grant particularly relevant to innovators who have identified an important research workflow problem and are developing an initial solution. What Does “Trustworthy AI” Mean for This Grant? Trust is a central part of the 2026 Catalyst Grant. The programme focuses on agentic systems that incorporate elements such as: - Provenance - Governance - Transparency - Accountability - Traceability Applicants should therefore explain not only what their AI system can do, but also how users can understand, evaluate, and trust the system’s actions and outputs within a research environment. What Problems Can Applicants Address? Applicants should identify a meaningful problem within research workflows and explain why their proposed solution is needed. Potential areas include challenges associated with: - Producing and managing research content. - Finding relevant research and evidence. - Synthesizing large bodies of evidence. - Supporting peer-review processes. - Making research funding decisions. - Protecting research integrity. - Screening manuscripts. - Matching reviewers with research. - Supporting scientific discovery. The strongest concepts should show how agentic AI can address a genuine problem rather than simply applying AI technology without a clear research use case. Application Requirements Applicants must provide a clear explanation of their proposed idea and its development stage.

The application should describe:

- The problem being addressed. - The proposed approach. - How the solution fits an agentic workflow. - How trustworthiness is incorporated into the workflow. - The current stage of development. Applicants should make the relationship between the problem, proposed AI solution, and research workflow clear. How Applications Are Evaluated Applications are evaluated across seven areas: - Team - Problem - Solution - Competitors - Market - Progress to date - Fit with Digital Science Applicants should therefore present their idea from both technical and practical perspectives. A compelling technology alone may not be enough; applicants should also explain the problem, competitive environment, development progress, market opportunity, team capability, and connection to Digital Science. Team Applicants should clearly communicate who is developing the proposed solution and what relevant skills or experience the team brings. For individual applicants, this means demonstrating their own relevant capabilities and ability to advance the idea. For startups and research