AI grants 2026: FENG, Dig.IT and KFS

Most companies that write to us about an AI grant already have the project in mind. They know what they want to deploy — the budget is the blocker. That makes it a good moment to check funding, but here’s the bad news: most grant guides are bureaucratic noise that never answers the one question that matters — does my specific AI project qualify?

Below are three programmes that can realistically fund an AI implementation in a company in 2026. No generalities. Each fits a different situation, and in two of them a costly misunderstanding is easy to make.

Call status changes. As of 28 July 2026, the Dig.IT pilot is closed, ARP has published a list of 26 projects and announced two further calls in 2026. The PARP SMART Path call for SMEs closed on 31 March 2026, while KFS runs through local labour-office calls until funds are exhausted. Before recommending a programme, we check the current schedule.

Three programmes, three different situations

FENG / SMART Path — when you have a novelty element

Poland’s flagship grant programme (European Funds for a Modern Economy). The call for SMEs closed on 31 March 2026. The programme funds R&D projects leading to innovation — aid intensity depends on the type of work, company size and premiums (25–50% base, up to 60–80% with premiums). And here’s the catch the guides stay quiet about: this is not the track for buying an off-the-shelf model, SaaS or AI tool. An AI deployment qualifies only as part of an R&D project leading to an innovation.

If you want to roll out an existing model or a SaaS subscription with no novelty element, this is probably not your programme. If you’re building something the market doesn’t have yet — it may be ideal.

Dig.IT — the cleanest fit for a first implementation

Digital-transformation grants for SMEs in manufacturing and production services (Agency for Industrial Development, under FENG). Scope: PLN 150–850k, up to 50% of eligible costs. This is a fit for a first AI implementation as part of IT solutions — no mandatory R&D. One condition: at least five closed full financial years.

For most companies that simply want to deploy AI in a process (rather than invent a new technology), Dig.IT will be a simpler route than SMART Path.

What the pilot showed: 146 applications requested a total of PLN 57,280,894.35. The published list covers 26 funded projects receiving PLN 11,221,918.74. Examples include an AI/IoT manufacturing deployment, production and sales automation using AI, and OEE monitoring with process analytics. ARP plans two further calls in 2026; the pilot and the two new rounds are expected to have a combined allocation of PLN 100m. This is the position as of 28 July 2026 — confirm the next timetable and call documentation on the official Dig.IT website of the Agency for Industrial Development.

A pre-check before the next Dig.IT call

The first round showed that a sound technology idea is not enough. ARP identified recurring problems with business eligibility, presentation of financial data and complete documentation. Before drafting the application, answer four questions:

Is the company and its activity within the eligible group of manufacturing or production-services SMEs?
Do the financial data and five closed full financial years support eligibility for the programme?
Does the project describe a specific process change, implementation scope and measurable outcome rather than merely buying “AI”?
Does the documentation connect technical scope, costs, schedule, integrations, security and maintenance?

Semitora can run this pre-check as the technical delivery partner: we assess use-case fit, data, architecture, integrations, evaluations and budget. We do not replace a grant adviser — the company or a specialist partner remains responsible for the formal application, declarations and settlement.

KFS — for training, not implementation

The National Training Fund finances employee training — it may cover AI competences if they fit KFS priorities and the role’s duties. It covers up to 90% of the cost for firms with up to 9 employees and up to 70% for larger ones (capped at 200% of the average wage per participant). And here’s the first important condition: it’s a grant for training people, not for deploying a system.

The second condition, key from 2026: KFS-funded training must be delivered by a provider listed in the BUR development-services registry. The old training-institution register (RIS) no longer applies here. Not everyone who runs training can deliver it under KFS — check this before you sign anything.

The most common mistake: confusing an implementation grant with a training grant

This is a mistake that costs months. A company applies expecting to “get a grant for AI,” only to find the chosen programme funds something other than what it needs:

You want to deploy an AI system → look at Dig.IT or SMART Path (if you have an R&D element). KFS won’t help here.
You want to train your team on AI (e.g. for AI Act requirements) → that’s KFS. But the training must be delivered by a BUR-listed provider.

These two needs often go together, but different programmes fund them, with different rules.

Four terms, no noise

FENG / SMART Path — Poland’s flagship grant programme (European Funds for a Modern Economy). The call for SMEs closed on 31 March 2026. The programme funds R&D projects leading to innovation; an AI deployment qualifies as part of such a project, not as buying a ready-made tool. Funding depends on the type of work and premiums (25–50% base, up to 80% with premiums).

Dig.IT — digital-transformation grants for SMEs in manufacturing and production services (Agency for Industrial Development, under FENG). PLN 150–850k, up to 50% of costs. A fit for a first AI implementation in a company.

KFS — the National Training Fund. Co-funds employee training (up to 90% for firms with up to 9 employees, up to 70% for larger ones). From 2026 the course must be delivered by a provider listed in the BUR development-services registry. This funds training, not implementation.

Eligibility (kwalifikowalność) — whether a given cost can be covered by the grant. The key question before applying: not every AI cost (e.g. a SaaS subscription) is eligible in every programme.

What Semitora does — and what it doesn’t

Let’s say it plainly, because a lot of grant firms promise everything. We are not a grants agency. We help with what we know — the technical part:

matching a programme to your real AI project,
assessing eligibility (what qualifies and what doesn’t),
preparing the technical part of the application: AI project description, architecture, budget.

The formal submission and settlement are handled by your team or a specialist grants adviser. With KFS we help define which AI competences your team needs — but we don’t deliver the training itself as a BUR-listed provider.

Compliance-first: compliance is part of the project, not an add-on after it

It’s easy to describe the implementation in an application and leave out what makes it last: AI Act risk classification, oversight and a plan to maintain quality. Yet those are what separate a “we’ll deploy AI and see” project from a considered one. Compliance and governance aren’t a cost bolted on afterwards — they belong in the project’s technical scope from the start. In the technical part of the application, which we help prepare, we include them alongside the implementation itself: AI Act risk classification and governance and evaluations.

Next step

If you have an AI project and you’re wondering whether it qualifies — write to us. We’ll send back a preliminary eligibility assessment and point you to the programme that best fits your implementation. Programme details: funding page.