The first question in every RFP is price. The short answer: the AI opportunity assessment starts at HUF 500,000 + VAT. An AI assistant or integration project is usually €15-40K. A document processing system runs €30-80K. Across our 46+ completed projects, the typical ROI timeframe is 6-12 months.

Here's what's in the price, what moves it, and what it costs to run the system after go-live.

How much does an AI project cost in 2026?

Three price bands cover most of our projects: assessment from HUF 500,000 + VAT, AI assistant or integration €15-40K, document processing system €30-80K.

AI opportunity assessment — from HUF 500,000 + VAT. We review your processes and deliver a prioritized list: what each automation brings, what it costs, and how fast it pays back. The output is a written plan you can use even if you continue without us. For SMEs, this is the entry point — the details are on the AI consulting for SMEs page.

AI assistant or integration project — €15-40K. This covers domain-specific AI assistants built on your internal knowledge base, customer service AI, and adding AI capabilities to an existing ERP or CRM system. At an insurance company, 78% of staff used the internal knowledge base we built within the first month.

Document processing system — €30-80K. Invoices, contracts, and unstructured documents processed with fine-tuned models and RPA integration. At a freight forwarding client, the system handles 85% of about 2,000 monthly invoices without human intervention; processing time went from 3 days to 4 hours.

After the assessment you get an exact quote — no surprises.

What moves the price?

Three factors: scope, data quality, and integrations.

Scope. How many processes the system touches, how many document types it handles, how many people work with it. A pilot narrowed to one use case and a system serving several departments are multiples apart in cost.

Data quality. Data preparation is one of the largest cost items in a typical AI project. Duplicates, missing fields, inconsistent formats — the messier the starting data, the more work before the model performs reliably.

Integrations. The AI model has to work with your existing systems: it needs to read from your ERP, write to your CRM, and respect your access controls. Integration is a significant share of the total cost, and every additional legacy system raises it.

What does it cost to operate?

An AI system keeps costing money after go-live: API fees, monitoring, retraining, maintenance. Monitoring and maintenance need to be budgeted from the first year.

Models degrade over time: the real world changes, new document formats appear, customer language evolves. Someone has to watch the performance metrics, retrain when accuracy drops, and update the data pipeline when source systems change.

Volume moves the running cost too: token usage for API-based models, GPU infrastructure and DevOps time for open-source models on your own servers. We set up a monitoring dashboard for every system we build — the client sees what the system costs and what it returns.

Is there grant funding for AI adoption?

Yes. The DIMOP Plusz-1.2.6/C-26 program gives Budapest-based micro and small businesses HUF 3-12 million in support at up to 90% intensity; submissions open on 1 September 2026 (source: palyazat.gov.hu).

In the call, AI on its own is not a supported development goal. AI is eligible as a subscription service under the cloud-services development goal — at a net unit cost of HUF 14,400 per user per month. Anyone advertising "90% AI grants" is promising more than the call contains.

The conditions: a Budapest site, at most 49 employees, de minimis funding; entry requires a digital-intensity measurement at kkvdigital.dkf.hu. The budget is HUF 2 billion — the budget of the previous B-26 call (for areas outside Budapest) was exhausted quickly, so it pays to arrive with a submission-ready application on opening day. The conditions and the eligible AI costs are collected on our AI grants for SMEs page. We covered the AI-eligibility details in a separate article: Is AI an eligible cost in Hungary's DIMOP grant?

When does AI adoption pay back?

Across our 46+ completed projects, the typical ROI timeframe is 6-12 months.

Automated processes reach a 40-45% efficiency gain. At a manufacturing client, defect detection went from 60% to 94%, at 1.8 seconds per board. The payback starts early: a proof of concept is ready in 2-3 weeks, and most projects go from kickoff to production in 6-12 weeks. The savings start the day the system goes live.

ROI needs measuring. Every project we deliver includes a measurement framework: time saved, error rate, and the system's full running cost, tracked from day one.

Where should you start?

With the assessment: fixed price, a written plan, a prioritized list. Source code, trained models, and data pipelines remain yours — no vendor lock-in. The AI ROI calculator gives a first payback estimate from your own numbers. Tell us where the time goes, and we tell you what can be done about it. If there is no viable use case, we say so. We don't sell AI for the sake of AI.