AI Disclaimer

GrantSpark uses artificial intelligence to help you find grants more quickly. This page explains exactly what that means, where the AI is reliable, and where you must make your own judgement before relying on a result.

What our AI does

When you run a match, GrantSpark filters our grant database by hard eligibility rules (such as geography, organisation type and employee size) and then asks an AI model to assess fit quality and suggest a recommended next step for each surviving grant.

The AI returns a fit score (0–100), a recommended decision (Apply / Consider / Skip), a short list of reasons it matched, risks to watch out for, and suggested next steps. Every result is shown with the reasoning that produced it. We do not believe in black-box scoring.

What our AI does NOT do

GrantSpark is decision support, not legal, financial or grant-writing advice. We do not guarantee that any grant we surface is the right fit, that you will be eligible, or that you will be awarded funding.

Specifically, you should be aware that:

• AI models can be wrong. The model occasionally misreads eligibility text, mis-classifies an organisation, or over-states a match. Always read the funder’s own eligibility criteria before applying.
• Our grant data may be out of date. Funder websites change frequently. We refresh our database regularly, but a grant may have closed, changed criteria, or moved deadlines between our last check and your match. The grant’s source URL is the canonical truth — always click through and verify.
• Eligibility is your responsibility. You know your own organisation, finances and project better than we do. A match is a suggestion to investigate, not a confirmation that you will pass the funder’s assessment.
• No guarantee of funding. Grant awards are competitive and entirely at the funder’s discretion. GrantSpark cannot influence funder decisions and we make no claim — express or implied — about your likelihood of success.

How we reduce errors

We try to minimise AI mistakes through a few specific design choices:

• Hard eligibility filters (geography, audience type, employee size, match-funding requirement, deadline) are applied deterministically before the AI sees a grant — so the model can’t accidentally recommend a grant you’re fundamentally ineligible for.
• The fit score you see is a blend of a deterministic rule score (60%), the AI’s fit score (30%) and the AI’s self-reported confidence (10%) — so a low-confidence AI guess can never on its own produce a high overall score.
• Every match shows the reasons for the recommendation and the risks to verify, not just a number.
• Grant data is timestamped and re-checked periodically. Stale grants are removed from active matching.

Before you apply

For every grant you intend to apply for, you should:

1. Click the source URL and read the funder’s own eligibility criteria and guidance in full.
2. Confirm the deadline directly on the funder’s page.
3. Check whether the funder has updated any criteria, restricted geographies, or paused the programme.
4. If you’re unsure about a specific eligibility point, contact the funder directly. Most funders welcome questions.

Application drafting

When we add AI-assisted application drafting, the same principles apply: any text generated by GrantSpark is a starting point you should review, edit and verify before submission. You are responsible for the content of any application you submit.

Reporting an issue

If a match is clearly wrong — an obvious eligibility miss, a closed grant still showing as active, a hallucinated requirement — please tell us at hello@grantspark.co.uk. We use these reports to improve the matching engine.

See also our Terms of Service, Privacy Policy and Cookie Policy.