What we fund
Two tracks, one year, community-set priorities.
There are no fixed categories. The documents below map some of the work the community has prioritised so far, but these are designed to be illustrative rather than limiting.
Where the ideas come from
This fund does not run themed calls. Instead it takes its cues from work the community has already prioritised for itself. The Edinburgh workshop of March 2026 produced forty-six candidate activities across its working groups — tradeoffs frameworks, incentives for publishing and crediting code, verification and validation, the evolving RSE role, training, playbooks for managers and maintainers, access to AI tools, and collaboration between people rather than only with machines.
Treat all three as illustrative and sources of inspiration.
The two tracks
- Award size
- up to $50,000
- Total available
- to be confirmed
- Duration
- 3–6 months
- Applications
- close late October 2026
- Award size
- up to $200,000
- Total available
- to be confirmed
- Duration
- 9–12 months
- Applications
- late 2026 / early 2027
Exploratory grants — examples
We are especially interested in tools and workflows that take on the routine parts of maintaining research software, so that more human time goes to design, judgment, and collaboration.
| Work | Budget |
|---|---|
| Agentic verification tools and workflows that scaffold testing and review | ~$50k |
| Agent workflows that take on routine maintenance — issue triage, dependency updates, release chores | ~$30k |
| Workshops and hack hours at community conferences to prototype agent skills | ~$25k |
| A public, citable website for a responsible-AI risk register and tradeoffs framework | ~$20k |
| A shared template and published collection of AI-assisted workflow case studies | ~$25k |
| A writing sprint to produce v1 of an AI costs-and-benefits framework | ~$20k |
| Launching a community of practice for educators teaching computational skills with AI | ~$25k |
| An initial playbook for research team leaders navigating AI adoption | ~$15k |
Grants may include travel to bring distributed teams together.
Impact awards — larger work, wider reach
These awards are for work too big for an exploratory grant: building and evaluating tools the whole community can run — verification infrastructure, benchmark suites, evaluation harnesses — delivering training at scale, or sustained investigation of the harder questions.
On the research side, questions we would like to see progress on include:
- How do we validate LLM-written tests? What does correct mean when code and tests share an author?
- How is scientific code review actually changing? Ethnographic work on shifting verification norms.
- Does generative AI make translating theory into code easier, or harder?
- Do teams with research software expertise get better results from AI than researchers working alone?
These awards are not expected to settle the questions. We hope they will help shape the agenda the field works on over the next few years.
What we are unlikely to fund
- Ongoing salary or maintenance costs that continue after the award ends.
- Compute, licences, or hardware as the substance of the proposal rather than a small supporting cost.
- Building a new research software package for one project, with no wider practice outcome.
- Work with no public output — everything funded here should leave something others can use.
- Outputs that cannot be released openly. Software must carry an OSI-approved license, and everything else — reports, curricula, playbooks, data — must be CC BY or CC0. Outputs should also be publicly hosted: code on a platform such as GitHub or GitLab, and documents, data, and other files archived on a service such as Zenodo.
Unsure whether an idea fits? Ask before you write a proposal.