Research Ireland is Ireland’s new science funding agency. It’s an amalgamation of Science Foundation Ireland (SFI) and the Irish Research Council (IRC).

Government institutions don’t reorient themselves very often. The establishment of Research Ireland (RI), then, is an opportunity. It’s a chance to take stock, set priorities, and set the right trajectory for the next decade.

RI is seeking input on its first corporate strategy. A public consultation is open until 22 April. Here’s my take on what RI should – and shouldn’t – be doing.

Keep. Stop. Start.

First, context is important. In 2020, SFI listed five backdrops. One, the economic fallout from the pandemic. Two, the ongoing COVID threat. Three, the effects of climate change. Four, the geopolitics of Brexit. Five, rapid technological advancement, particularly in machine learning.

Much is different in 2025.

The scientific landscape has changed utterly. AI has facilitated unprecedented breakthroughs. Research funders are all-in on metascience. ARPAs are pioneering better ways to do research. Team science is the new norm. Living literature reviews are replacing textbooks.

The global context is similarly alien. Mercantilism has riotously returned. Major powers are geared for war. The climate crisis has delivered wildfires and storms. Energy costs are exorbitant. Supply disruptions are driving scarcity.

With that in mind, here are some thoughts:

What should Research Ireland keep doing?

It’s important to note that SFI and IRC were broadly successful against their mandates. SFI attracted significant research and industry investment. IRC supported excellent individual early-stage researchers. Both dramatically increased the foundational capacity of Irish science.

The first thing SFI did well was Frontiers of the Future. This was SFI’s most important grant for principal investigators. It provided substantial funding in two streams. Projects funded high-risk, high-reward research. It was designed to facilitate highly innovative and novel approaches to research. Awards provided larger-scale funding. It was more collaborative and impact-focused. Frontiers supported good work but was undermined by the associated admin. Regular small changes to the Frontiers application created confusion. It also increased compliance costs. This confusion was multiplied across both streams. Going forward, Research Ireland should increase the grant size and number of grantees. Resolving Frontiers into two separate grants would improve ease of use. Frontiers should also adopt features of exemplar grants. These include NIH R35 and R01 grants. R01 is a bread-and-butter grant that minimises year-on-year changes. It is extremely dependable. R35 (aka MIRA) provides even greater stability and flexibility, thereby enhancing productivity. Ring-fenced streams should be made available to early-career researchers.

The second thing RI does well is work with other funders. In March 2025, RI began a five-year partnership with the NSF. Making the most of this partnership should be a strategic goal. RI should work closely with the TIPs directorate. Staff exchanges and joint facilities development could be game-changers. The partnership should be expanded to access U.S. research tools. This includes the National AI Research Resource Pilot. RI should also partner with other funders. Good options include Schmidt Futures, Open Philanthropy, and UKRI.

The third thing SFI did well was Innovate for Ireland (I4I). Over 10 years, I4I will fund up to 1,200 PhD scholarships. Each “iScholar” will get a €28,000 yearly stipend and extra training. The higher stipend is a good step, but it still falls short compared to Irish apprentices and cadets. I$I’s current themes and focus on sustainability might be too broad and also too prescriptive. Focusing on key sectors, like precision fermentation, could have a bigger impact. I4I should also offer strong support to help iScholars follow bold ideas after they finish their PhDs.

The fourth thing SFI did well was manage research infrastructures. SFI improved equipment utilisation through access charge plans (ACPs). The latest guidelines for creating ACPs were published in May 2021. These should now be updated to require that ACPs are posted online and made public. Connecting ACPs to an updated LIRE—the research infrastructures database—would create a central tool for both public and private researchers. They could find equipment, check costs, and book access in one place.

The fifth thing SFI did well was establish Research Centres. SFI’s centres-led approach successfully brought in private funding. Centres that have shown strong results should continue to receive support. As journal impact factors matter less, other ways to measure success—like patents and spin-outs—should be used. Some centres managed by SFI and funded by industry worked well, but others like INFANT and APC, which were not directly managed or refunded by SFI, may have performed better. These kinds of centres could be given more freedom after review. Some should explore research management models based on well-known research groups like BBN (ARPAnet), CMU (driverless cars), or Lockheed Skunk Works (STEALTH). RI should also establish new centres with economically-critical MNCs. Supported by RI but privately run as Focussed Research Organisations, these new labs could help anchor FDI.

What should Research Ireland stop doing?

The first thing RI should stop is its over-reliance on peer review. NIH Deputy Director Mike Lauer found that peer review can sort proposals into broad groups, but has no power to rank them. Still, SFI and IRC relied heavily on peer review despite its limitations. Keyword searches would find reviewers, to be remunerated by a modest fee. But keywords didn’t guarantee reviewers had the right expertise. And there was no mechanism to verify they conducted the review themselves. AI has worsened peer review further. Now LLMs are being used to write and assess proposals. Other methods, like industry panels or even lotteries, might be better. “Golden tickets” can help ambitious proposals survive modified peer review via panels.

The second thing RI should stop is spreading its resources too thinly. One of RI’s goals is to “support research and innovation in all fields and disciplines.” At first glance, this could spread limited resources too thinly, with little impact in each area. To prevent this, priority should go to projects that can bring returns to the State and generate income. This would then make it possible to increase the research budget over time. Promising candidates include precision fermentation, cellular agriculture, neuroscience, biotechnology, and materials science. These fields can be further accelerated with AI for research. Research support for mature industries is also essential.

The third thing RI should stop is its over-reliance on certain metrics. Anecdotally, some respondents suggested that SFI and IRC engaged in low-value, high-volume reporting. Better metrics are needed to judge the quality of research. In particular, metrics measuring novelty and commercialisation should be developed.

The fourth thing SFI did badly was properly support full-time researchers. The majority of principal investigators are hosted by Irish universities. They have significant teaching commitments. SFI and IRC awarded grants to these applicants, but (anecdotally) failed to appreciate their dual role. RI should develop awards to fund full-time researchers at host institutions or develop new centres with fully dedicated research staff.

The fifth thing SFI did badly was support PhD researchers. Most PhD students in Ireland receive €18,500 per annum. An independent report gave advice on how to better support PhD researchers. It said Ireland should raise the stipend, aiming for €25,000 if funding allows. These recommendations should be put in place.

The sixth thing SFI did badly was support the university spin-out system. SFI did not, some say, deliver the expected number of spin-outs. This may be due to a lack of entrepreneurial people in the centres. RI should explore developing or joining a system like the UK Catapult Network. It should also take on board key ideas from the UK’s review of university spin-out companies.

The seventh thing SFI did badly was establish the National Challenge Fund (NCF). The NCF was promoted as Ireland’s Advanced Research Projects Agency (ARPA). It didn’t work out that way. The NCF lacked many key features of the ARPA model. In particular, it lacked the appropriate focus and management. The focus should have been use-inspired research on time-limited problems. This is called Pasteur’s Quadrant. The research should have been overseen by empowered programme directors. Program directors manage constant flux — building, replanning, changing tack, and moving talent in and out as project needs shift. The NCF did not feature these essential components.

What should Research Ireland start doing?

The first thing RI should do is establish a dedicated metascience unit. A small team implementing metascience findings could significantly increase Ireland’s innovation ROI. Even a modest increase would translate to millions in economic value. It should adopt the same mandate as the UK DSIT Metascience Unit. This includes the following. Running large grant programmes to grow Ireland’s metascience community and support research that helps policy and R&D. Designing and testing funding methods through experiments like large randomised trials, data analysis, shadow tests, and policy pilots to learn what works best. Sharing findings so that DFHERIS, RI, and others can use the best evidence in their funding decisions.

The second thing RI should do is build links with next-gen global funders. RI should seek to partner with Schmidt Futures, Renaissance Philanthropy, Speculative Technologies, and Convergent Research. The new agency should join the Research on Research Institute consortium to avoid being left behind. It should learn from spaces like Newlab. It should also implement funder playbooks from those institutions, particularly: Research Leader’s Playbook, Mid-scale Science, Policy Entrepreneurship, and a checklist for funder experiments with partial randomisation.

The third thing RI should do is build grants around new datasets and AI tools. Researchers at the Arc Institute have built a machine-learning model called Evo 2. It was trained on DNA from over 100,000 species across the Tree of Life. Evo 2 can spot patterns in gene sequences that would take years for scientists to find. It can also detect harmful mutations in human genes and design new genomes as long as those found in simple bacteria. RI should encourage Irish researchers to use resources like Evo 2 and AlphaFold.

The fourth thing RI should do is pioneer self-driving labs. Self-driving labs (SDLs) combine machine learning, robotics, and lab automation to speed up discovery in chemistry and materials science. They run experiments in a loop, learning and improving with each cycle. This helps researchers find new molecules faster and in greener, more sustainable ways. SDLs are already being used in areas like energy and healthcare. A roadmap is needed to help non-experts adopt these tools, expand their use, and overcome current limits. Their wider rollout could accelerate progress on major global challenges.

The fifth thing RI should do is optimise for leveraging FP10. The next EU research programme (FP10) will likely follow the advice in the Heitor report. To get the most EU funding for Ireland, Research and Innovation (RI) should align Ireland’s research system with the goals in that report.

The sixth thing RI should do is build a UK-and-Ireland research database. In 2023, the Government gave over €1 billion to research and development. Nearly 75% of this was outside SFI’s remit, so looking at their funding alone doesn’t give the full picture. To make good decisions, we need clear data on all public research funding. A national database should be managed by RI and include all publicly funded research. Working with UKRI would help us learn from a similar system and support cross-border projects. If we start now, the database can help us track progress over time. Without it, how will we know if RI is successful?

Progress Ireland’s responses to public consultation

What do you think Research Ireland (i.e. Science Foundation Ireland (SFI) & the Irish Research Council (IRC)) has been doing well and what should Research Ireland continue doing?

SFI’s Frontiers programme should continue by splitting it into two clearer streams with stable, predictable, rolling calls modelled on NIH R35 and R01. The NSF partnership should prioritise staff exchanges and shared infrastructure like the US National AI Research Resource. Innovate for Ireland should be strengthened but with tighter alignment to fewer national priorities and clearer post-PhD pathways. Access Charge Plans (ACPs) should be required and linked to an updated, online LIRE database. Successful Centres should be granted greater autonomy and assessed using spin-outs and patents. New centres should be modelled on Bell Labs and BBN. New centres as Focused Research Organisations should be developed with strategic MNCs.

What do you think Research Ireland should change/improve or should not carry forward from its predecessors SFI and the IRC?

Research Ireland should reduce reliance on peer review by testing alternatives like industry panels, lotteries, and golden tickets. Reviewer selection must improve, with better checks and less keyword-matching. It should be AI-proofed. Funding should be concentrated on fewer, high-impact areas like biotech, neuroscience, and materials science, especially where AI can help. Metrics should shift to focus on novelty and commercialisation. RI should fund full-time researchers or centres with dedicated staff. It should also join a system like the UK Catapult Network and adopt key recommendations from the UK spin-out review. Any successor to the NCF should be a true ARPA, likely focused on AgTech. PhD stipends should be increased to circa €25,000 pa.

What new actions would you like to see Research Ireland progress with the opportunity given by the formation of the new agency and new remit?

Establish a metascience unit to test and improve funding methods, grow Ireland’s metascience community, and inform RI strategy with evidence. Partner with next-gen funders like Schmidt Futures and Renaissance Philanthropies, and join the Research on Research Institute consortium. Use funder playbooks and learn from innovation spaces like Newlab. Encourage the use of AI models like Evo 2 and AlphaFold. Promote self-driving labs to speed up discovery in energy, health, and materials. Align Ireland’s research system with the Heitor report to maximise FP10 funding for Ireland. Build a shared UK-Ireland research database to guide decisions and track RI impact.