Business Finland

How an R&D grant consultant shapes an AI project before you write the application

Pre-application shaping: the diagnostic, use-case discovery, feasibility framing, and decision-workshop stages that produce fundable R&D projects. Week-by-week breakdown and signs your project isn't ready yet.

Short answer. Yes — and if your project is a broad AI ambition rather than a concrete R&D question, this pre-application shaping is often the highest-leverage part of the entire engagement.

Most companies come to a grant consultant with the same artifact: a slide deck describing what they want to build with AI. Sometimes it’s a two-page memo, sometimes a mature product roadmap. What it almost never is, on arrival, is a Business Finland-shaped R&D project.

The word that matters is shape. A grant application is a container. If the thing you pour into it is not already in R&D shape, the container will not save it — no matter how well the paragraphs are written. This is why the interesting work happens before anyone opens the application template.

This piece describes what that pre-application shaping actually looks like: what has to exist before writing starts, how a senior consultant produces those things week by week, and how the shaping decisions change the size of the grant you end up with.

The two things that must exist before you start writing

An AI R&D application only survives evaluation if two conditions are true on the day writing begins. Both are structural, not editorial. No amount of drafting fixes their absence.

1. A real R&D question with genuine technical uncertainty

Business Finland does not fund AI projects; it funds R&D projects that happen to use AI. The distinction matters. An “AI project” is a product ambition — we want to add an AI assistant to our platform. An R&D question is a specific technical uncertainty you cannot resolve by reading documentation and integrating an off-the-shelf model — we do not yet know whether a domain-adapted model can classify these regulatory documents accurately enough at cost X, and answering that requires a new training regime we intend to develop.

The test is blunt: if a competent senior engineer could deliver the project by picking existing tools and integrating them, it is not R&D. It is engineering. Engineering can be excellent, valuable, and profitable — and it will not be funded by an R&D grant.

Shaping work turns the fuzzy version into the sharp version. It replaces “we want to use AI to do X” with “we do not know Y, and finding out requires A, B, and C — none of which can be bought off the shelf”.

2. A project shape Business Finland can score

Evaluators do not read applications freely; they score them against a framework. That framework rewards:

• Distinct, named work packages with deliverables that map to R&D uncertainty (not to product milestones)
• A team and budget where personnel cost lines correspond to who is actually doing the research
• An impact story connecting research outputs to commercial outcomes, employment, exports, and the Finnish ecosystem — with numbers a reviewer will believe
• A timeline that is aggressive enough to be credible and slack enough to survive AI-project reality

A shaped project has these elements already decided before the writing begins. An unshaped project has them invented on the fly during drafting — which is why so many first-time applications read as if they were assembled backwards from a template.

What project-shaping actually looks like, week by week

Shaping is not a workshop deck; it is a series of decisions made together with the company’s technical and commercial leadership. At BRNSFT, this is packaged as the AI Opportunity Sprint — a 2–4 week engagement whose only job is to produce a project that is ready to be written.

The rhythm below is typical, not fixed. Faster runs happen with repeat clients; longer ones happen when the underlying business question needs to be renegotiated.

Week 1 — Use case discovery

The first week is diagnostic. Sessions with the CEO, CTO, and whoever owns the commercial roadmap surface the full set of candidate use cases the company is considering, along with the ones already ruled out.

The output is not a shortlist yet. It is an inventory: which problems are genuinely painful, which have clean data, which have a customer willing to sponsor the outcome, and which are quietly nice-to-haves that will die on the vine six months in. The point is to see the whole landscape before narrowing.

For AI projects specifically, this week is where the boundary between R&D use cases and integration use cases gets drawn honestly. Most companies discover, uncomfortably, that half their candidate list is integration work.

Week 2 — Feasibility framing

The second week narrows. Each surviving candidate use case is stress-tested against three lenses:

• Technical uncertainty: where does the not-yet-solved problem actually live — in the model, the data pipeline, or the evaluation setup?
• Data feasibility: is the training and evaluation data reachable, labelled, and legally usable inside the grant window?
• Commercial pull: is there a real customer or market willing to pay for the outcome, or is this a research idea in search of a use case?

At the end of week 2 the candidate list has usually collapsed to one or two projects. The rest are archived — sometimes for a later grant cycle, sometimes permanently.

Week 3 — Decision-maker workshop

The third week is the pivot from analysis to commitment. A single working session with the founders and technical leadership converts the surviving shape into a decision: this is the R&D question, these are the work packages, this is the rough team allocation, this is the timeline.

The workshop’s purpose is not to invent the project — the previous two weeks did that. Its purpose is to force the decision that most companies avoid: this and not that. Grant applications collapse under the weight of “we could also do…”. Shaping is how “also” gets deleted.

Week 4 — Handover to authoring

The final week — sometimes compressed into a few days — produces the artifact the writing stage needs: a one-page project shape covering R&D question, work packages, team, budget skeleton, and impact hypothesis. From here, the application drafting has something concrete to compress into evaluator-register prose.

If a company insists on skipping shaping and starting with drafting, this artifact gets improvised during writing — badly, under time pressure, and without the technical stress-tests. That is the failure mode this stage exists to prevent.

How the shaping work changes the funded outcome

Shaping is not a nice-to-have. It changes the size of the grant the company ends up with.

BRNSFT’s own progression across four approvals — EUR 70k (2021) → EUR 957k (2024) → EUR 187k (2025) → EUR 357k (2026), totalling EUR 1.57M — is not a story of writing improving. The prose was already fine in 2021. What changed was the shape of the projects being written about.

• The first application (EUR 70k) was a small, well-formed R&D question with a narrow scope. It fit a small instrument; it got funded at the size the shape justified.
• The EUR 957k application was a fundamentally different project — larger R&D uncertainty, multiple work packages, a real employment and impact story. The shaping work is what made a nearly seven-figure ask credible to evaluators.
• The subsequent EUR 187k and EUR 357k projects were shaped to specific instruments and stages, not to the maximum available number.

The lesson is not “always ask for more”. It is that the shape sets the ceiling. An application written about a small, narrow project cannot legitimately ask for a large grant. An application written about a well-shaped, multi-work-package R&D project cannot legitimately be scored at the size of a small one. Shaping decides which conversation you are having with Business Finland.

Signs your project is not ready to write yet

If any of the following are true when you sit down to draft, the writing will be papering over a structural gap.

• You cannot state, in one sentence, what specifically is unknown that this project will find out.
• Your “R&D component” is a description of features rather than of uncertainty.
• You have not decided who on the team is doing the research work as distinct from the product work.
• You cannot yet describe the project as three to five work packages, each with a deliverable that is not “shipped a version of the product”.
• The commercial impact story is aspirational rounding rather than something a reviewer will believe.
• You are considering more than one use case simultaneously and hoping the application will pick.
• The data you would need to run the R&D is not yet reachable or is legally ambiguous.
• The timeline is either a wish (fits in six months because it has to) or a hedge (twenty-four months to be safe) rather than the actual arc of the work.

Any two of these together is a signal to shape before writing. Four or more is a signal that writing now will produce a well-formed rejection.

Consultants who only write applications vs operators who shape projects

The market conflates these two roles constantly. They are not the same job.

Neither role is wrong. A strong, well-shaped project genuinely does not need weeks of shaping — a writer is enough. The mistake is hiring a writer when what you needed was an operator, and only discovering the difference in the rejection letter.

FAQ

How long does the shaping stage usually take? Two to four weeks of elapsed time, spread across roughly four to eight working sessions with the technical and commercial leadership. Faster is possible for repeat applicants who already speak the evaluation language; longer is common when the underlying business question needs to be renegotiated before the R&D question can be sharpened.

Is pre-application shaping billed separately from writing the application? Usually yes. Shaping and writing are different kinds of work and are best priced separately — a scoping fee for the shaping engagement, followed by a writing fee (often partly tied to approval) once the shape is settled. Bundling them into a single success-fee tends to compress or skip the shaping stage entirely, which is exactly the opposite of what you want.

Do we need a specific AI use case before the shaping starts, or can shaping help pick one? Shaping can absolutely help pick. Companies frequently arrive with three or four candidate use cases and no principled way to choose between them. Week 1 of the sprint is designed for exactly this case — the inventory, feasibility framing, and decision-maker workshop stages exist to narrow honestly rather than to write about whichever idea was most recent.

What if we already have a rough application draft — can shaping still help? Yes, and often more sharply. A rough draft is a diagnostic artifact: reading it reveals where the project shape is soft, which R&D claims will not survive evaluation, and which work packages are actually product milestones in disguise. Shaping in this case looks less like discovery and more like restructuring — but the fundamental question is still whether the project is fundable in its current shape.

Can shaping happen in parallel with writing, to save time? It can, but the compromise is real. Applications built parallel to shaping tend to inherit unresolved decisions — two possible use cases carried forward because the workshop hasn’t happened yet, work packages worded loosely because the boundary between R&D and product is still moving. Sequential is slower and produces better applications; parallel is faster and produces the second-draft feeling on the day of submission.

What happens if shaping concludes the project should not be submitted? It happens, and it is a feature of the process rather than a failure of it. Sometimes the conclusion is not yet — the shape needs another six months of internal work before it is fundable. Sometimes it is not this instrument — the project fits an R&D loan, a Sprint grant, or private capital better than a Business Finland R&D grant. Occasionally it is not at all — the AI ambition is genuinely engineering rather than research. Any of these is cheaper to learn during shaping than after a rejection.

The one-sentence version

The application is a container, the project is what you pour into it, and shaping is the work of making sure what you pour will fit.

If you want the argument about whether the consulting fee itself is worth it, that lives in a separate piece. If you want the full guide to how Business Finland R&D funding actually works, start with the long-form guide. If you want to see how BRNSFT works on the funding lane specifically, the Business Finland funding page covers the engagement shape; the operator page covers the same work for international teams outside the Business Finland instrument.

Related: Are Business Finland grant consultants worth the fee? · What does a grant consultant actually do for an AI project? · Business Finland R&D funding guide