NRW.BANK.Impuls KI: €25,000 AI Grant for SMEs in NRW

From 17 August 2026, small and medium-sized enterprises in North Rhine-Westphalia can apply for funding under the new NRW.BANK.Impuls KI programme. The state covers half the cost, up to €25,000. For many mid-sized companies, this is the most economical moment to finally take a structured approach to getting started with AI. At Ziya, we support you along the way: from use case identification through the prototype to a production solution on Ziya OS.

What It Is About: NRW.BANK Covers 50 % of the Cost

With the NRW.BANK.Impuls KI programme, NRW.BANK helps the North Rhine-Westphalian Mittelstand prepare for the deployment of artificial intelligence. In other words, exactly the phase in which most projects fail or never get off the ground at all: the phase between "we really should do something with AI" and a robust, tested use case.

The key facts at a glance:

Type of funding	Grant (non-repayable)
Funding rate	50 % of eligible expenditure
Maximum funding	€25,000 within two years
Minimum threshold	€5,000
Target group	SMEs as defined by the EU, based in NRW
Project duration	as a rule, a maximum of six months
Application	from 17 August 2026, digitally via the NRW.BANK customer portal
Repeatability	a maximum of twice within two years

In concrete terms: a project with a volume of €50,000 effectively costs your company €25,000. The other half comes as a grant, not as a loan and not as an equity stake.

Two Modules, and Both Match Exactly How We Work

The funding is divided into two modules. That is the genuinely interesting part, because this structure mirrors precisely the path we take in AI projects anyway.

Module A: Concept and Strategy

Conceptual consulting services are eligible for funding, in particular:

Identifying company-specific fields of application and defining objectives for AI deployment
Defining and prioritising concrete use cases, taking into account the technical and organisational prerequisites within the company
Establishing structures and processes for efficient AI management
Structures and processes for the responsible use of AI

That last point is not a fig leaf. The EU AI Act brings concrete obligations for companies: clarifying your role as provider or deployer, transparency and labelling duties, documentation. Anyone who addresses this only after rollout builds up technical debt that gets expensive. Governance belongs in the concept phase, and that is exactly what is eligible for funding here.

Module B: Build and Test a Trial Version

Funding covers building a test version of the AI system and carrying out the test. The goal: proof that the system actually delivers the intended benefit and meets the company's requirements.

In addition to external consulting and testing services, software licences, hardware rental and access to test infrastructure are also eligible here to a limited extent.

This is where Ziya OS plays to its strengths.

Why Ziya OS: A Tested Use Case in Weeks Rather Than Quarters

The classic route to an AI prototype looks like this: set up infrastructure, connect models, integrate data sources, build permissions, develop a frontend. And after three months you find out whether the idea holds up at all. In a funded project with a six-month runtime, that is a problem.

Ziya OS is an AI implementation platform. Data connectivity, model orchestration, rights and role management, agent workflows and auditability are already in place. We do not build a prototype from scratch, we configure use cases on a production-ready foundation. For an Impuls KI project, that means three very concrete advantages:

Several use cases testable in parallel. Instead of putting your entire funding volume on a single bet, we evaluate multiple candidates and let the data decide which one gets scaled.
The prototype is not a throwaway product. What is tested in Module B runs on the same platform on which it later goes into production. No rebuild, no technology break between pilot and rollout.
Data sovereignty from day one. On-premise, private cloud or hybrid: your data stays where it belongs. For companies with sensitive engineering, customer or health data, that is not an option but a prerequisite.

Which Use Cases Are Particularly Well Suited

The same applications keep emerging from our projects in the Mittelstand, all with clearly measurable benefits and therefore well suited to a proof of value under Module B:

Sales & Market Development

Quote generation: A draft quote is created in minutes rather than hours, from the enquiry, the history and your pricing logic.
Customer and lead research: Systematically identifying and qualifying suitable target customers from internal and external sources.
Discoverability in AI search: Your customers no longer just google, they ask language models. Anyone who does not appear there as an answer does not exist for a growing share of the market. We analyse and optimise your visibility in AI systems.

Operations & Administration

Reporting and document generation: Pull weekly, monthly and project reports from your systems automatically, in your format and in your language.
ERP integration: Create entries in natural language and query analyses without clicking through input masks. "How did product group 4 develop last quarter compared to the previous year?" An answer instead of a ticket to IT.
Workflow automation and agents: Map recurring process chains such as incoming invoices, complaints, quote approvals or onboarding end-to-end with AI agents, including human approval points.

Knowledge & Research

Internal knowledge search across contracts, technical documentation, standards and project archives.
Research and analysis use cases, for example in patent landscapes, tenders or regulatory material.

Most of these applications can be properly designed, tested and evaluated within the funding period.

Grant Applications Are a Craft, and We Have Practice

A good funding application is not a form you fill in. It is the written version of a well-thought-through project: clearly scoped, plausibly costed, with a comprehensible value proposition and a work plan that fits the timeline.

Ziya is itself an AI company that has secured public funding several times, including at considerably larger scale at federal and state level, and we have successfully navigated funding processes together with our customers. We know both sides of the table: the substance that makes a project viable, and the formal requirements that applications otherwise get stuck on.

For NRW.BANK.Impuls KI, that means concretely: we support you with the project description, the module allocation, the costing and the work plan, and then implement as a service provider what the application sets out.

Three Things You Should Bear in Mind Now

1. Do not start beforehand. The project must not have begun at the time of application, and must not begin before the funding decision is issued. Anyone who starts the project in July and applies in September loses the funding. Preparing is fine, commissioning is not.

2. The use case must be settled before the application goes out. An application that amounts to "we want to do something with AI" will have a hard time. What is required is a clear focus on an intended AI application, cleanly assigned to one of the modules. That groundwork is exactly what we do with you, in a compact scoping conversation, free of charge and without obligation.

3. Twice in two years is possible. An SME can draw on the funding up to twice within two years. That opens up the option of staggering Module A and Module B strategically: first strategy and prioritisation, then testing the strongest candidate. We factor that in from the outset.

Why This Is More Than a Grant

North Rhine-Westphalia has the largest industrial Mittelstand in Europe and at the same time a considerable gap when it comes to productive AI use. Not because ideas are lacking, but because a gap yawns between idea and operation: missing infrastructure, an unclear data situation, uncertainty around regulation and data protection, no budget for an open-ended experiment.

That is precisely the gap NRW.BANK.Impuls KI addresses. The funding takes the economic risk out of the exploration phase and makes it rational to do what is right anyway: first understand, then test, then scale.

We see this as a shared task. Digitalisation in the Mittelstand does not succeed through tools alone, but through partners who take responsibility for the outcome.

Let Us Talk About It

Are you considering an application and not yet sure which use case is worth it? Or do you already have a clear idea and are looking for an implementation partner who will actually deliver it within the funding period?

Get in touch with us. In an initial conversation we clarify together:

which applications offer the greatest leverage in your company,
whether and how your project can be mapped to Module A and/or Module B,
what a realistic time and cost frame looks like within the funding logic,
and how the solution transitions into regular operation after the test.

Ziya GmbH: AI implementation from Dortmund, for the Mittelstand in NRW.

Book a meeting now

All programme details as published by NRW.BANK. Only the NRW.BANK.Impuls KI guideline and the NRW.BANK FAQ are authoritative. Further information and the application process can be found at nrwbank.de - NRW.BANK.Impuls KI. This article does not constitute funding or legal advice.

NRW.BANK.Impuls KI: €25,000 in Grant Funding for Your AI Start - and an Implementation Partner That Delivers