R&D operations
What EUR 1.57M in Business Finland approvals taught me about AI R&D framing
Patterns that repeat across four approved Business Finland AI R&D cases, from EUR 70k to EUR 957k — what separates a fundable framing from an unfundable one.
Short answer: Across four Business Finland approvals shaped between 2021 and 2026 — EUR 70k, EUR 957k, EUR 187k, and EUR 357k, totaling EUR 1.57M — the pattern that predicted success wasn’t project size, sector, or company maturity. It was whether the founder could state the technical uncertainty in one honest sentence before we started writing anything. Every approval had that. Every near-miss or slow case was, at its root, a framing problem that took longer than it should have to fix.
The track record, briefly
Four approvals, four different shapes: a EUR 70k early-stage Tempo grant in 2021 for a tightly scoped feasibility test; a EUR 957k multi-year applied-AI R&D programme in 2024, combining applied research with productization work packages; a EUR 187k focused feature build in 2025, one clean uncertainty tested over a defined timeline; and a EUR 357k platformization project in 2026 with an internationalization component attached. Different company stages, different sectors, different euro figures. The thing they had in common wasn’t visible in the numbers.
Pattern one: the uncertainty was never invented after the fact
In every successful case, the technical uncertainty existed before the application did — it was a real, pre-existing open question the team was genuinely unsure about, not something reverse-engineered from “what would make this fundable.” Applications built by starting with the target budget and working backward toward a plausible-sounding uncertainty read differently to evaluators, and they read differently for a structural reason: the uncertainty section in a genuine case is specific and slightly uncomfortable to write, because it means admitting you don’t know if something will work. Manufactured uncertainty sections tend to be confident and vague at the same time — technically worded, but short on the actual moment of not-knowing.
Pattern two: project size followed project shape, not ambition
The EUR 957k approval wasn’t bigger because the company was more ambitious than the EUR 70k company — it was bigger because the work honestly required a multi-year, multi-work-package programme to resolve. The EUR 70k grant wasn’t a “starter” size in any diminished sense; it was exactly the size the actual uncertainty justified, and it turned out to be enough to unlock what came after. Every attempt to push a budget above what the project’s real scope supported produced a weaker application, not a bigger one.
Pattern three: the strongest applications separated the uncertainty from the plan
A recurring early draft mistake, across nearly every case, was letting the technical implementation plan crowd out the uncertainty statement — pages of confident technical detail with the actual open question buried in a sentence near the end. The fix was almost always structural, not about generating new content: state the uncertainty plainly and first, then let the plan follow as the method for testing it, not as proof the answer is already known.
Pattern four: co-funding and instrument fit got decided too late, not too early
In cases where the process ran long, it was rarely the technical narrative that caused the delay — it was the company figuring out its co-funding approach, or which specific instrument fit the project’s actual stage, after the application was mostly written rather than before. The projects that moved fastest had the instrument and funding-stack question settled before the writing started.
What this means if you’re earlier in the process
None of these patterns are about being a bigger company, having a bigger budget, or picking a more exciting sector. They’re about sequencing and honesty: find the real uncertainty before you write anything, size the project to what that uncertainty actually requires, and settle the instrument and co-funding question early rather than late. The four approvals that add up to EUR 1.57M didn’t get there by being different kinds of companies. They got there by following that order.
FAQ
Does a bigger project have a better chance of approval? No — size follows the project’s real scope. A well-scoped EUR 70k application can be stronger than a padded EUR 500k one.
How do you know if your uncertainty is “real” rather than manufactured? If you can honestly describe what a negative result would look like — the approach not working — and you’re not sure in advance which way it goes, that’s a real sign. See the ten-signs checklist linked below for the fuller version of this test.
Do these patterns hold across different AI project types? Yes — the four cases span different sectors and project shapes, and the framing patterns held across all of them, which is part of why they’re worth naming as patterns rather than one-off advice.
The one-sentence version
Across EUR 1.57M in approved Business Finland AI R&D funding, the pattern that mattered wasn’t size or sector — it was a real, pre-existing technical uncertainty, sized honestly, with the instrument and co-funding question settled before the writing started.
Related: What kinds of AI projects does Business Finland actually fund? · Ten signs your AI project is ready for Business Finland funding · How to write the R&D uncertainty section of a Business Finland application