Frequently asked questions

The AI Tool Pyramid groups a continuously growing catalogue of tools into thematic tiers – from general-purpose assistants through image, video, audio, coding and automation to science, medicine, politics, philosophy, literature, mental health, finance and sustainability. The current number of tiers and tools is shown right below the headline on the start page. A tier's width reflects how many tools it contains, which creates the even pyramid shape. The tiers are not a ranking: a wider tier is not "better" than a narrower one.

The levels are grouped into seven thematic blocks, marked in the pyramid by colour families and divider rows. Block 1 “Assistance & Knowledge Work” (levels 1–4) holds general assistants, research, text and translation as well as office, notes and meetings. Block 2 “Creation & Media” (levels 5–7) bundles image and design, video and motion, plus audio, voice and music. Block 3 “Engineering & Operations” (levels 8–13) spans software development, automation and IT operations through gaming and cybersecurity to Industry 4.0 and robotics. Block 4 “Business & Organisation” (levels 14–19) covers law, finance, people, sales, real estate and logistics. Block 5 “Science” (levels 20–23) covers literature and evidence research, life sciences, physics and simulation as well as astronomy. Block 6 “Health & Environment” (levels 24–27) bundles medicine, nutrition and wellbeing, sports, climate and agriculture. Block 7 “Society & Culture” (levels 28–30) links society, politics and media with culture, art and literature and with philosophy, ethics and education. The level is a thematic classification, not a rating.

Every tool page states whether German and English are supported, distinguishing between the interface (the tool's menu language) and content (the language you can work in). Additional languages sit in a collapsible section. Vendors keep expanding coverage, so this information comes without guarantee.

Beginners are best served by starting with a general AI assistant and one concrete, recurring task from their own day — summarising texts, drafting emails or structuring ideas. Only once it is clear where the time savings are does it pay to move to specialised tools for image, video, research or automation. From the outset, avoid entering personal or confidential data into free consumer plans. The AI helper on this site suggests matching tools from a short description of your task.

AI language models generate answers from statistical patterns and can state factually wrong information in entirely convincing language. Numbers, quotations, legal statements, medical claims and literature references are especially prone to this. The work only becomes reliable when the tool supplies sources and those are spot-checked. For anything that needs evidence, research tools with citations are preferable to general assistants.

The filter bar above the pyramid narrows down the catalogue: • Category – thematic focus, e.g. image, audio, automation or finance & controlling. • Pricing – free, open source, freemium or paid. There is also a “Free forever” toggle for every tool you can keep using without paying (not just a trial). • Entry price – cheapest paid monthly plan, grouped into up to €10, €10–30, over €30 and on request. • Privacy – GDPR / EU, US cloud or unclear. Filters can be combined and are reflected in the address bar, so a filtered view can be shared as a link or bookmarked. If you prefer describing what you need in your own words, use the AI Tool Finder.

Yes. The plus icon on a tool card adds it to the comparison. As soon as at least two tools are collected, compare mode puts them side by side – pricing model, privacy classification, languages, use case and tier. The selection is also part of the URL and can be shared; the X removes individual tools again.

The AI Tool Finder answers questions about the tools in this catalogue. Simply describe your task in your own words – for example “Show me tools that turn PDFs into narrated videos”, “Which transcription tool runs on EU servers?” or “What is a free alternative to …?” – and you get matching suggestions with direct links to the detail pages. It only draws on the content of this site and is free to use without signing up. So the service stays available for everyone there are fair hourly and daily usage limits plus protection against very rapid bursts of requests.

There is no single best free image tool, because the offerings differ substantially in visual style, resolution, usage rights and daily limits. Photorealistic motifs call for different models than illustrations, logos or product shots. A useful comparison therefore asks three questions: how many images per day are free, is there a watermark, and may the results be used commercially? The image generation category lists every image tool with pricing details, and the free AI tools collection additionally filters for permanently free and open-source applications.

For academic research, specialised literature and evidence tools are far more suitable than general chat assistants, because they work directly on indexed scholarly databases and back each statement with a citable source. General assistants write fluently but occasionally invent references. A combination works best: research and source verification in the specialised tool, summarising and phrasing in the assistant. The science tiers of the pyramid group tools for literature search, evidence appraisal, life sciences, physics and astronomy.

The right AI assistant depends mainly on whether writing quality, live web research, data protection or integration with your existing software matters most. Heavy writers look at language quality and context length; researchers need citations and web access; anyone handling personal data needs EU processing and a data processing agreement. Because the models change monthly, current pricing and privacy details are more useful than leaderboards. The comparison mode places up to three assistants side by side with prices, languages and privacy details.

AI video splits into two families: generative video models that turn text or images into short clips, and avatar tools that build a talking presenter from a script. Generative models suit visual storytelling, advertising and art but are limited in length, consistency and lip sync. Avatar tools fit training, product videos and multilingual explainers. Both families have their own tiers in the pyramid so the two are not mixed up during selection.

Yes. European vendors are particularly strong in translation, language models, transcription, document processing and science, and often score with EU processing, German-language support and data processing agreements written to European practice. In areas such as generative video the offering is thinner. Every tool detail page states the researched company headquarters with city and country, which makes it easy to pick out European vendors.

Three kinds of AI tools help with programming: code completion inside the editor, chat assistants for explanation and debugging, and agents that carry out whole tasks across multiple files. For production use the decisive questions are whether your source code is used for training, whether a self-hosted variant exists and how well the tool integrates with your existing editor. The software development tier groups these tools with pricing and privacy details.

The privacy notes summarise what each vendor publicly documents: server location, handling of training data, data processing agreements and EU availability. They are guidance, not a legal assessment. For business use always review the vendor's own privacy policy and data processing agreement.

The tools that can be used in a GDPR-compliant way are those whose vendors document EU processing, offer a data processing agreement and do not use inputs for training without consent. That includes European vendors as well as business plans from large US vendors with EU data centres and standard contractual clauses. It is never the tool alone that decides, but the plan you choose and the contract you sign. The overview of tools with European data protection level collects every application whose vendor publicly documents EU processing; each tool detail page also states where data flows and which contract is required.

No. An EU server location is an important indicator, but not proof of legal compliance. What also matters is who operates the company, whether a parent company in a third country can access the data, whether a data processing agreement exists and whether inputs are used for training. Support and backup systems in particular often sit outside the EU. Each tool detail page reports these points separately so the data centre location alone does not decide the assessment.

Before deploying an AI tool in a business, four things are needed: a data processing agreement with the vendor, a documented legal basis for the data being processed, a clear internal rule about which data may be entered at all, and a check on whether inputs are used for training. For vendors outside the EU, suitable safeguards for the third-country transfer are added. It also helps to pseudonymise personal data before entering it and to keep the initial group of users small.

That depends on the vendor and the plan you book. Free consumer plans reuse inputs for training far more often than business or API plans, where training use is usually excluded contractually. Many vendors also offer an opt-out switch, though it often does not apply retroactively. The privacy section of each tool detail page states how the vendor handles training data and where the setting can be turned off.

Every tool is researched and assigned to one of four pricing models: free, open source, freemium or paid. The “Pricing and plans” section on each detail page also lists concrete plans – with the vendor's price, an approximate euro figure and notes on specifics such as usage-based billing, annual discounts or education plans. Extensive plan line-ups can be expanded via “Show all plans & details”. AI vendors change their plans frequently, and prices can differ by region, currency and VAT. The vendor's own pricing page, linked on every tool page, is therefore always authoritative.

No. The AI Tool Pyramid is freely accessible, without registration, advertising or a paywall. There is no partnership with the vendors listed and no affiliate links are used.

“Free” covers three very different models in the AI world: permanently free services, open-source software without licence costs, and freemium offers whose free tier is deliberately built as a step towards a subscription. Only the first two can be built into workflows permanently without a paywall; freemium limits usage through messages per day, output length, watermarks or project counts. Open source can also incur hosting or API costs. The free AI tools collection separates these cases cleanly.

Most individual AI subscriptions sit between roughly 10 and 30 euros per month per tool; team and business plans with data processing agreements and admin features are typically higher. Specialised professional tools in medicine, law or industry are often quoted on request only. Combining several tools quickly costs more than one all-in package, so it pays to clarify the actual need first. Each tool detail page lists the researched plans with the date of the last check.

Open-source AI tools can be used without licence costs, run on your own infrastructure and audited in the source code, so data never has to leave your own environment. That matters especially when personal or confidential content is processed. The trade-off is the effort for operations, compute and updates, plus the usual lack of support with guaranteed response times. Open-source tools are marked as such in the catalogue and listed separately in the free tools collection.

Tools are included when they are productively usable, publicly available and relevant in their field. In the scientific tiers a solid methodological foundation and transparent data are additional requirements. Deliberately excluded: short-lived wrapper products without their own value, and tools without an identifiable operator.

Not in the classic sense. No third-party analytics services such as Google Analytics, Matomo or Plausible are used, no tracking pixels, no advertising or third-party cookies, no recognition across visits and no profiling. No data is sold or passed on to advertising networks. For anonymous reach measurement the site itself records per page view: the page requested, date and time, a coarse city-level location, browser, operating-system and device family, plus where the visit came from (e.g. a search engine, a direct visit or campaign parameters in the link). Your IP address is neither stored nor logged. Automated access such as crawlers is filtered out during analysis based on user-agent, headers and behaviour patterns. Full details are in the privacy policy. Stored locally on your device are only your language preference, the acknowledged privacy notice and your recently viewed tools.

Yes, absolutely. Use the contact form and include the name, a link and a short description of the use case. It doesn't matter what field the tool comes from – suggestions for entirely new categories are welcome too. Privacy-friendly offerings are especially welcome.

The catalogue is maintained continuously: new tools are added, discontinued services removed and descriptions revised when features, pricing or privacy terms change. On top of that, automated checks regularly test every vendor link for availability and flag notable changes, so dead links and discontinued services surface quickly. Reports about outdated information are still always welcome.