AI can help funders reduce administrative burden throughout the entire grant lifecycle. However, it’s important to distinguish between using AI to automate basic tasks and allowing it to materially influence important decisions.
The use of AI should mean that staff have more time to review applications and engage with grant seekers. It should help them to make more informed decisions rather than replace their judgment with automated recommendations that they can’t properly scrutinise.
Funders also need to understand what happens behind the scenes when AI is introduced into their grant management processes. Where is applicant data being processed? Who can access it? Can staff verify the information AI produces? How much control does the organisation have over the way the technology is used?
Instead of asking whether a vendor offers AI features, it’s important to ask the above questions – and more. In this article, we look at five critical questions to ask any funding management software supplier before selection.
Grant teams spend considerable time on work that needs to happen before they can make a funding decision. Information from different sources has to be brought together, applications and documents need to be checked for completeness and reviewed against eligibility requirements, and due diligence must take place.
AI can help with these tasks. It can handle things like:
These uses don’t transfer decision-making authority to a system.
The 2026 Charity Digital Skills Report found that 27% of surveyed funders were using AI in their application processes for similar purposes. Specifically:
However, only 8% reported using AI to check eligibility, shortlist, or score applications against criteria. The report also found that 61% of grantmaking trusts and foundations were concerned about the quality and accuracy of AI tools, while 48% had concerns around ethics and human rights.
These findings reflect a sector exploring the technology while exercising caution about how it affects funding decisions.
The caution mentioned above is reasonable. An AI-generated summary can help an assessor navigate a long application, but it might overlook an important qualification or misrepresent an applicant's plans. A due diligence summary can bring together information from public records, but it can’t automatically establish whether an organisation is suitable for a particular grant.
AI outputs should be treated as material for review, not as verified facts or conclusions in their own right.
This distinction ensures everyone gets a fair chance. Smaller organisations may not have teams dedicated to writing applications, nor extensive digital resources. If funders use AI to process applications, they should ensure that doing so does not disadvantage applicants whose work is less polished or whose evidence is presented differently.
Funders need to understand how AI systems handle information, what safeguards are in place, and how much control they retain. The following five questions can help establish whether a supplier's AI features are suitable for the organisation's needs.
When a supplier says its platform uses AI, that description alone doesn’t tell you much about how the technology handles your organisation’s data.
The AI feature might run within the grant management platform, or the platform might send information to an external AI provider for processing. Different arrangements can have different implications for confidentiality, security, data retention, and contractual responsibilities.
Ensure that the vendor explains the architecture clearly. Ask:
It’s also important to understand whether the arrangements differ between features.
The supplier should be able to explain these arrangements clearly and provide relevant documentation to support its answers.
Funders should also establish which protections are covered by the contract and which depend on the policies of external providers.
Even when an AI system operates within a secure environment, funders need to understand what information it can access. A system that summarises an application, for example, may need access to the application form and relevant attachments. It may not need unrestricted access to every applicant record, internal discussion, financial document or previous funding decision held by the organisation.
The principle should be to provide access to the information necessary for the task, rather than making all available data accessible by default.
Further questions:
The appropriate safeguards will depend on the task and the sensitivity of the information involved. Processing a publicly available list of registered organisations presents different considerations from analysing personal data or confidential research proposals.
The Royal Free Charity's AI position statement offers an example of this distinction. It permits AI to support administrative work while stating that confidential or sensitive application data will not be entered into AI systems without appropriate safeguards. Funders should expect their own suppliers to explain how comparable protections work in practice.
An AI-generated output is more useful when staff can check where its claims came from, and this is especially important when it comes to due diligence.
AI should make relevant evidence easier to examine; staff should not accept conclusions without checking them.
Ask suppliers:
If a supplier can’t demonstrate how staff can verify an output, the feature may not be suitable for tasks where errors could materially affect an application or funding decision.
Determining whether an application deserves funding often requires contextual knowledge and consideration of the funder’s priorities. The interaction between these factors can be complex; therefore, judgments can’t always be reduced to a checklist.
Ask suppliers whether AI can automatically determine eligibility, assign scores, or reject applications. Can staff review the underlying evidence, correct errors, and override outputs? Are AI-assisted actions recorded so the organisation can understand how a decision was reached?
AI adoption shouldn’t mean surrendering control over how online grant application software operates. Ask suppliers whether individual AI features can be disabled or restricted, rather than requiring the organisation to enable or abandon AI across the entire platform. Can access be limited to particular users or workflows? Can the organisation change its settings as its internal policies change?
It’s just as important to understand what happens when an AI feature is unavailable or no longer considered appropriate. Can staff continue the process manually? Will disabling the feature affect access to records, existing workflows, or other parts of the grant management process?
Also ask how you’ll be notified about material changes to the AI models, features, or data-handling arrangements so you can be sure that you’ll be able to reassess its suitability when needed.
Grantmaking organisations need clear internal rules for how AI is used once a system is in place, and the answers vendors provide during procurement should inform governance.
If a funder can’t establish what data an AI feature accesses, for example, it may need to restrict its use until that information is clear. Likewise, if outputs can’t be traced to their sources, the feature may be unsuitable for tasks where accuracy is especially important.
Testing is also crucial. Before introducing a feature, funders should evaluate it against realistic examples and establish whether it performs accurately enough for its intended purpose. They should also consider whether it treats different applicants fairly and whether its outputs could disadvantage organisations that communicate, operate, or document their work differently.
Governance should continue after implementation. Regular reviews can help identify changing risks and features that no longer meet the organisation's needs. The level of scrutiny should be proportionate to the sensitivity of the data, the role of the AI, and the consequences of getting something wrong.
Finally, funders should consider whether a proposed use of AI solves a genuine problem. Automating an unnecessarily complex reporting requirement, for example, may be less valuable than simplifying the requirement itself.
AI can help grantmakers spend less time processing information and more time making evidence-based funding decisions.
Before choosing an AI grant management software supplier, establish where the technology runs, what data it can access, whether its outputs can be verified, who retains decision-making authority, and whether you can control or disable individual features.
These questions determine whether AI strengthens the grantmaking process or introduces new risks that staff can’t adequately oversee.
To learn more about Flexigrant’s AI features, contact us today.