Applications are open for the Digital Science Catalyst Grant 2026, an international initiative supporting early-stage ideas that use technology to advance research and create meaningful impact. The 2026 theme focuses on trusted agentic workflows — AI systems that can plan, execute and review multiple steps of a research task rather than simply responding to a single prompt.
Digital Science is looking for early-stage startups, research teams, students, founders and individual innovators developing AI tools that can operate within real research workflows while maintaining provenance, governance and accountability. Successful applicants can receive up to £25,000, with the total award potentially shared among multiple recipients.
What Is the Digital Science Catalyst Grant?
The Digital Science Catalyst Grant supports innovation in software tools and technologies designed to advance research and increase its impact on society. The program is intended to identify and support promising ideas at an early stage — applicants do not necessarily need a completed product or established revenue.
For 2026, the focus is on AI systems that can actively support research processes while giving researchers and institutions confidence in how those systems operate.
2026 Theme: AI That Acts on Research
Artificial intelligence is increasingly moving beyond generating answers and summaries toward systems capable of carrying out complex, multi-step tasks. The 2026 Catalyst Grant is looking for technologies that make these agentic research workflows practical and trustworthy.
Ideal solutions should combine automation with safeguards such as:
- Audit trails
- Governance controls
- Research provenance
- Accountability
- Transparent workflows
- Appropriate human oversight
Who Can Apply?
The Catalyst Grant welcomes early-stage ideas from a wide range of applicants, including:
- Startups and founders
- Research teams
- Students
- Individual innovators
- Researchers developing new technology
- Early-stage research projects
Applicants may submit a prototype, working product or well-developed concept.
What Types of Projects Are Being Sought?
Applicants should be developing agentic tools that address genuine challenges within the research ecosystem. A strong proposal should demonstrate that the technology can:
- Work across multiple steps rather than simply answer one prompt
- Fit into workflows already used by researchers, institutions or publishers
- Maintain appropriate audit trails, governance and provenance
- Address a real research-related problem
- Demonstrate a meaningful outcome that can be measured
Highlighted areas of the research lifecycle
1. Trusted Authorship and Writing: AI co-writing systems with transparent contribution records, autonomous citation and source-verification agents, cryptographic systems for verifying authorship, and AI tools for high-stakes academic writing.
2. Enterprise Research Workflows: Evidence-synthesis agents for regulated environments, knowledge-graph workflows with research provenance, explainable drug discovery systems, automated target-identification tools.
3. Funding and Institutional Decisions: Researcher identity and disambiguation agents, automated reviewer matching, conflict-of-interest screening, grant portfolio intelligence for program officers.
4. Publication Integrity: Auditable manuscript screening, citation verification, figure provenance, multi-step editorial workflows, and comprehensive audit trails for publication processes.
What Makes an Idea Agentic?
A conventional AI tool may respond to a single instruction. An agentic workflow, by contrast, can involve planning, carrying out actions, checking results and responding appropriately to what happens along the way. If the entire task can be completed effectively through a single well-designed prompt, the project may not fit the 2026 theme.
How to Test Your Idea Before Applying
- What research decision does your agent affect? Identify the decision, who makes it, when it is made and how frequently it occurs.
- Where does it fit in the research lifecycle? Writing, discovery, evidence synthesis, peer review, funding decisions or research integrity.
- Where will the technology be used? Identify the software, platform or system in which your agent needs to operate and who already uses that system.
- Does the workflow genuinely require multiple steps? Determine whether your system needs to plan, execute and review actions.
- What happens when the system encounters a problem? Test situations where the agent should refuse, flag an issue or escalate to a human.
- How will you measure success? Define a clear outcome such as researcher hours saved, faster funding decisions or improved research integrity.
How Applications Are Evaluated
Applications are assessed across seven areas: team, problem, solution, competitors, market, progress to date, and fit with Digital Science. Applicants who progress further may be invited to a short interview involving a live demonstration where applicable, followed by questions about their project.
Catalyst Grant 2026 Key Dates
StageDateApplications openTuesday, 1 September 2026Applications closeMonday, 5 October 2026 at 17:00 BSTJudging period12 October – 20 November 2026Winners announcedWeek commencing 30 November 2026Maximum awardUp to £25,000
The available award may be split among more than one recipient.
How to Apply
Applicants should be prepared to explain the problem they are addressing, their proposed solution, why it represents an agentic workflow, how it can operate within a trustworthy research environment, the current stage of development, and any prototype or demonstration available.
Apply on the Digital Science website
Source: AfriCareers (africareers.net)





















