Check your AI potential The free AI Potential Check
Your AI maturity and up to five use cases that fit your company best. First results instantly in your browser. Unlock euro ranges and the roadmap with your email address. We only read public pages.
- AI Act (EU) 2024/1689
- GDPR
- Destatis
- Eurostat
- Bitkom
- Studies
What the check reads and what you get
First your website, then a few short questions. From these the check works out your maturity and the use cases that fit your company.
Your website
What does the web say about your company?
Home page, legal notice, careers page and sitemap: from these we prefill industry, business model and shop system and spot signals such as open positions, a missing chat widget or many product pages.
Your answers
Where does your team lose time?
Systems, time sinks, rough volumes, state of your data and start date. Rough estimates are fine, and you can change your answers in the result at any time.
Instantly in your browser
How ready is your company for AI?
AI maturity from 0 to 100 in five areas, up to five fitting use cases with department and a preliminary classification under the EU AI Act, plus first findings.
With your email
What is the freed-up time worth?
Euro ranges and hours per use case, a roadmap, notes on data protection and the EU AI Act, and every assumption to adjust. We also send you the report by email.
What a result looks like
A made-up company, calculated with the same formula as the check: Wholesale, 50–249 employees, online shop on Shopware, ERP SAP Business One.
- AI maturity and classification instantly
- Use cases sorted by value, feasibility, effort and your answers
- Euro ranges and hours once you enter your email
47/100
Building up
€57k–110k
per year, typicalValue of the working time freed up, typical, before licence and running costsFreed-up working time: about 150 hours per month, typical
- 1 · Sales
Capture orders from email and PDF
€16,000–28,000
per year, typicalValue of the working time freed up, typical, before licence and running costs - 2 · E-commerce
Write product copy from product data
€11,000–21,000
per year, typicalValue of the working time freed up, typical, before licence and running costs - 3 · Customer service
Sort service emails and suggest replies
€10,000–20,000
per year, typicalValue of the working time freed up, typical, before licence and running costs - 4 · Sales
Prepare B2B quotes from enquiries
€12,000–29,000
per year, typicalValue of the working time freed up, typical, before licence and running costs - 5 · Purchasing
Check and chase order confirmations
€6,600–13,000
per year, typicalValue of the working time freed up, typical, before licence and running costs
How the check calculates
Your question isn't here? Ask us directly
How is the euro value calculated?Formula
Volume × minutes per task × AI share ÷ 60 × 12 × hourly rate
For each use case the check takes your monthly volume or, if you leave it empty, a typical value for your industry and size. Minutes per task and the share the AI takes over are cautious values from studies and vendor figures. The hourly rate is €45 employer cost per hour including non-wage costs, the average across all businesses in Germany. Smaller businesses and trades are often lower.
The range runs from the cautious to the ambitious end of the chosen scenario (cautious, realistic or ambitious). So that no use case grows unrealistically large, the check caps the hours per use case and for the top list by company size. You can change the hourly rate, the scenario and the volumes in the report.
The result is the value of the freed-up working time, typical, before licence and running costs.
How is the AI maturity score calculated?0 to 100
Five areas with up to 20 points each. From 80 points the check shows “AI-ready”, from 60 “Good foundation”, from 40 “Building up”, below that “Getting started”.
- Data: How structured your data is and whether ERP and CRM bring it together.
- Systems: Which systems you use and how well they can be reached via interfaces.
- Processes: How clearly time sinks and volumes are named and how digital your workflows are.
- Team: How your team uses AI today.
- Governance: Whether there are approved tools, rules and a framework for AI.
Use cases are sorted by value, feasibility, effort and your answers. If a use case needs something that is still missing, such as an ERP or digitally captured data, it moves down.
What does the scan read?Public pages
Our server only fetches publicly accessible pages of your website, within a fixed time limit:
- Home page
- robots.txt and sitemap
- Legal notice (Impressum)
- Careers page and at most one page of a job board linked there
- MX records of the domain in DNS
From these it reads:
- Industry, business model and shop system for the prefill
- Company name, legal form and city from the legal notice, shown only to you
- Open positions and their titles
- Chat widget, traces of CRM tools and the website's languages
- Product pages in the sitemap and missing translations
- Microsoft 365 or Google Workspace from the MX records
How does AI classify your website?With redaction
Our server sends an AI model by TypeSafe AI only the language, title, meta description, headings, menu items and a short text excerpt of your home page, plus job titles and paths from the sitemap. Before that it removes email addresses, phone numbers, street addresses, the company name from the legal notice, headings and menu items that look like a person's name, as well as sentences containing role or title words such as managing director, owner, Mr, Ms or Dr, quotes and legal text. Other names your home page shows publicly may remain. The model classifies the website, for example by industry and business model, to prefill the questions. You check and correct the prefill before anything is calculated.
If the AI model does not answer, fixed keyword rules classify the website. The check calculates every figure in the result with fixed formulas, not with the AI model.
Where do the assumptions come from?41 sources
Statistics, studies and vendor figures from which we cautiously derive minutes per task, the share the AI takes over and hourly rates:
- Labour costs per hour worked, Germany: Destatis, Pressemitteilung Nr. 148 vom April 2026 (2026)
- Labour costs per hour worked by economic sector: Destatis, Pressemitteilung Nr. 148 vom April 2026 (2026)
- Labour costs per hour worked by company size: Destatis, Fachserie 16 Heft 1, Arbeitskostenerhebung 2020 (2022)
- Companies using AI (10 or more employees): Destatis, Unternehmen mit Nutzung von KI nach Beschäftigtengrößenklassen (2025)
- Companies using AI and reasons against it: Destatis, Pressemitteilung Nr. 444/2024 (2024)
- Companies using AI in the EU: Eurostat, Use of artificial intelligence in enterprises (2025)
- Companies with 20 or more employees using AI: Bitkom, Erstmals nutzt die Mehrheit der Unternehmen KI (2026)
- Companies with 20 or more employees using AI: Bitkom, Durchbruch bei Künstlicher Intelligenz (2025)
- AI assistance for customer service agents in chat: Brynjolfsson, Li, Raymond: Generative AI at Work (2025)
- Writing tasks with a language model: Noy & Zhang, Science (über MIT News) (2023)
- Knowledge work by consultants with GPT-4: Dell'Acqua et al.: Navigating the Jagged Technological Frontier (2023)
- Time saved with AI chatbots, all office tasks: Humlum & Vestergaard: Large Language Models, Small Labor Market Effects (2025)
- Time saved with generative AI, self-reported: St. Louis Fed: The Impact of Generative AI on Work Productivity (2025)
- Microsoft 365 Copilot, time saved per day: UK GDS: Microsoft 365 Copilot Experiment, Cross-Government Findings (2025)
- Microsoft 365 Copilot, time saved per task: UK Department for Business and Trade: M365 Copilot evaluation (2025)
- Copilot in the laboratory: Microsoft Research: AI and Productivity Report, 1st ed. (2023)
- Email work with generative AI, field experiment: Dillon et al.: Shifting Work Patterns with Generative AI (2025)
- AI assistant in customer service, vendor figure: Klarna, Pressemitteilung (2024)
- Chatbot resolution rate by request type: Gartner, Pressemitteilung vom 15.06.2023 (über Digit.fyi) (2023)
- Resolution rate of an AI agent in support: Intercom, Fallstudie auf claude.com (2025)
- Time saved with AI, employees in Germany: Indeed-Umfrage, berichtet von DER BETRIEB (2025)
- Processing time per incoming invoice in SMEs: IW Consult: Wenn sich Digitalisierung rechnet (2025)
- Processing time per invoice: Mittelstand-Digital Wertnetzwerke, ZUGFeRD (2012)
- Cost and lead time per invoice: Ardent Partners: State of ePayables 2024 (Bottomline) (2024)
- Manual order entry: Hobson & Company, zitiert von TradeCentric
- EDI versus paper: cost difference per document: GS1 Austria, Broschüre Order2Cash (2011)
- Order entry from email: Lleverage, Blog (2025)
- Handling time per call in a call centre: SQM Group: Call Center Industry Operational Benchmarking Data (2021)
- Handling time by channel: Gladly, CX Performance Indicator Benchmarks
- Staff time per appointment booking: Accenture-Bericht, zitiert von JBJS OrthoBuzz (2014)
- Minutes per product description: Pleasepoint, Success Story
- Cost per returned item: Universität Bamberg, Forschungsgruppe Retourenmanagement (2019)
- Cost per return shipment: FashionUnited, Bericht zur Bamberger Studie (2019)
- Internal handling cost per return: EHI, Versand-, Verpackungs- und Retourenmanagement im E-Commerce (über VerkehrsRundschau) (2025)
- Minutes per document for classification: KI-Syndikat, Use-Case-Seiten (2026)
- Delivery note capture: imagetotable.ai, Leitfaden (2026)
- Reading time per CV: StepStone Österreich und MindTake (2018)
- First screening of an application: Kienbaum/Staufenbiel Jobtrends 2017 (über Haufe) (2017)
- Meeting load in Germany: HR WORKS / bilendi, Umfrage (2025)
- Share of working time spent searching for information: KMWorld / Feldman (IDC)
- Effort for the monthly management report: WHU Controller Panel mit ICV: Management Reporting in turbulenten Zeiten (2022)
From check to measurement
The check calculates with typical values. How much of it is in your company only shows with real volumes and handling times from your processes.
For companies that want to decide on real numbers
Measuring the potential with real numbers: interviews with your team, volumes and handling times from your processes and a technical review. At the end you get an assessment and prioritisation of the use cases and a roadmap.
For teams that should use AI safely day to day
Our workshops are one building block for developing and documenting your team's AI literacy (Art. 4 EU AI Act).
For specific questions about your result
Tell us which use cases interest you. We'll tell you which one makes a good starting point and what the next step would be.

Slawa Ditzel
Executive Partner
Reply within one business day
- You send your result or your questions
- We go through the use cases with you
- You get a concrete next step
30 minutes · no pitch · honest assessment
Learn more: AI potential analysis · AI workshops · AI consulting
AI Potential Check FAQ
Your question isn't here? Ask us directly
How accurate is the AI Potential Check?
What happens to my data?
What do I see instantly and what after the email?
Does the check work without a website?
Is the EU AI Act classification legal advice?
Does our team need AI knowledge already?
What does the check cost?
Measure your AI potential with real numbers
Send us your result or tell us where your team loses time. We'll tell you which use case makes a good start and what an AI potential analysis would look like for you.