207 website features for SMEs: btlabs Core in detail

btlabs Core offers 207 standard website features for SMEs , distributed across fifteen areas: Google search, AI search, performance measurement, speed, multilingualism, editorial management, AI assistance, chat control, queries, media, search, data sovereignty, security, operations, and interfaces. This reference explains what each feature does, its benefits for your business, and—where applicable—its limitations.

It's written for businesses that want to control their own website: trades, retail, tourism, service providers. Not a checklist, but clear explanations for every single point. One key point upfront, because it's a recurring theme: almost every one of these functions can also be performed in conversation with an AI assistant – extracting data, assessing it, writing, correcting, translating, and publishing texts. Example sentences are provided where this saves a significant amount of work. The strategic background is described in the article " btlabs Core as a Website Foundation for SMEs ." The reason we built the whole thing ourselves is explained in the article about the architecture of btlabs Core.

A function is only useful to me if it solves a specific task within the company. Technology without operational benefit usually only creates new dependencies. And what a function can't do should be discussed just as thoroughly as what it can.

The fifteen areas at a glance

• 1. Be found on Google & Co. (01–13) – so that search engines can understand your business correctly.
• 2. Be found by AI (14–27) – Present your facts in such a way that an assistant can reproduce them correctly.
• 3. Measure whether it works (28–48) – Visitors, real search terms, AI accesses and mentions in AI responses.
• 4. Tempo (49–57) – why the page is there immediately.
• 5. Multilingualism (58–71) – German, Italian and English equivalent, not retrofitted.
• 6. Content and Editorial Team (72–101) – Building blocks, releases, versions, company data in one place.
• 7. AI assistance in the system (102–114) – Support, switchable, with human approval.
• 8. Maintaining the website via chat (115–129) – say instead of click.
• 9. Inquiries and Contact (130–143) – Forms, delivery, clean distribution.
• 10. Media and Images (144–152) – Sizes, cutting, image rights, data protection.
• 11. Search on your own website (153-157).
• 12. Data sovereignty and data protection (158–170) – why it works without a cookie banner in normal operation.
• 13. Security and Access (171-184).
• 14. Operation, security, reliability (185-200).
• 15. Interfaces and other channels (201-207).
• What is deliberately omitted – the honest counter-list.

the essentials in brief

• 207 functions as standard in fifteen areas – none of them an add-on module that will later appear on the invoice.
• Approximately 20 content areas and 20 page modules, from which your team can assemble new pages themselves.
• German, Italian and English are equally valid – right down to the user interface where your employee works.
• The entire company is involved in the discussion – upon request: Your website can be connected to an AI assistant. It reads your data, tells you what it means, writes and corrects texts, translates them, publishes them, or schedules them for a specific time. You tell it what should happen, instead of clicking through menus.
• Readable by humans and machines – the same content works in Google and in the answers of AI assistants.
• Noticeably fast: Response times in the range of a tenth of a second, even when multiple people are on the site simultaneously. No promise – you can measure it yourself on our own website.
• No cookie banner during normal operation: No third-party trackers, no advertising pixels, no cookies for your visitors, our own server is located in the EU.
• Secured and encrypted at night – and before every update, a multi-stage test is performed before anything goes live.

Why chat access saves the most time

Most website systems are capable of many things – but nobody actually uses them. Because every little thing requires logging in, searching through menus, and finding your way around again. That's precisely where chat access comes in, and it can do three things that are otherwise separate.

He reads. Visitor numbers, the actual search terms from Google and Bing, page views, which AI systems are monitoring – plus every page, every post, every job posting. He assesses. What's working, what's stagnant, where's something missing, where are there contradictions. And he acts. Creating, correcting, translating, scheduling, publishing.

What this means in everyday life: The evening you would normally spend clicking through menus becomes three sentences between two appointments. "Which pages performed best last month?" – "Which ones are missing a description?" – "Write it down for me and save it as a draft." For a company without a marketing department, that's the difference between "I'll do it when I have time" and "it's done."

And because nothing goes live without your approval, you still retain control. You decide individually what the assistant is allowed to do. The entire chapter on this topic can be found under Maintaining the Website via Chat.

Two things to clarify: The range of functions is constantly expanding, so this is just a snapshot – the offer with its service description is always binding. And: "available" doesn't automatically mean "activated." Some features can be deliberately deactivated, while others require your company to make an initial contribution. Where this is the case, it will be noted with the respective function.

1. Getting found on Google & Co.: Functions 01 to 13

Being found doesn't guarantee rankings – no one can seriously promise that. What btlabs Core does is create the conditions that allow search engines to clearly understand your business. They take care of the rest themselves.

• 01. Clean page structure. Headings, sections, and navigation are clearly structured, with breadcrumbs and directly linkable sections. You can send a customer a message with the exact section they're interested in, instead of just saying "scroll down." Honestly: Structure is no substitute for good content.
• 02. Title and description per page and language. You decide which phrase will appear with your page in the search results and see a preview of how it will look before you save. This way, you control what a searcher reads about you first, not chance. Also in discussion: "Check which pages are missing the description and send me suggestions for them." Honestly: Search engines sometimes shorten or replace such texts at their own discretion.
• 03. Clear addresses, forwarding included. Your pages will have readable addresses; if you rename one, the system will automatically create the redirect. Last year's business card will still lead to the correct page. Honestly: Addresses from an old website must be transferred once during the migration.
• 04. Always up-to-date page overview for search engines. What's been published is listed there – drafts and expired job postings are automatically removed. Nobody maintains a list here, and nobody forgets. Honestly: A site overview does not guarantee inclusion or placement.
• 05. News feed per language. Your blog will also be delivered in a format that specialist portals, press distribution lists, and reading programs can automatically accept. This allows a post to continue circulating without you having to submit it anywhere.
• 06. Immediate notification of new content. When you publish something, the system reports the address directly to the participating search engines. Your promotional page doesn't wait for a search engine robot to randomly visit it. Honestly: Google does not participate in this process – there, the constantly updated page overview takes precedence.
• 07. You decide what can be included in the search. For each page, you decide whether it's optimized for search engines. The internal price list or the half-finished landing page stays out – without anyone having to tweak the technical aspects. Honestly: The final decision regarding retrieval and inclusion rests with the search engine.
• 08. One main address instead of three almost identical pages. If content is accessible via multiple channels, the preferred address is displayed. This concentrates the impact on one page instead of spreading it out. Honestly: This is a suggestion to search engines, not an instruction.
• 09. Your company data is machine-readable. Business details, location, opening hours, service area, services, FAQs, job openings, team, and reviews are also presented in a format that search engines can directly understand. This is the difference between Google seeing you as a business and Google seeing you as a page with text on it. Honestly: Google decides whether these results become highlighted hits or stars.
• 10. Automatic verification of this information. If the machine-readable output differs from the text on the page, the system flags it. The principle is simple: only what a visitor actually sees is marked up. This is precisely where most overly ambitious systems fail.
• 11. Job postings that understand job portals. Your job postings are maintained with expiration dates and displayed in a machine-readable format; expired postings disappear automatically. Candidates only see open positions, and no one has to clean up the career page. Honestly: There is no guarantee that Google Jobs or a portal will include the advertisement.
• 12. Proper preview when sharing. The title, description, and preview image are optimized for social networks and messaging apps and are automatically updated when you change the image. It looks great when a satisfied customer shares your link in a group. Honestly: Whether a platform accepts the information is entirely up to the platform itself.
• 13. Review of individual pages before release. For each page, you can see how it would appear in search results and what's still missing. An empty description will be noticed before publication – not months later over coffee. Also in discussion: "Go through the performance pages and tell me where something is missing."

A technically sound website improves the prerequisites, but doesn't automatically generate visibility. If you want to know where your site stands today: the AI-Ready Check will take a look.

2. Being found by AI: Features 14 to 27

Today, people looking for a recommendation often ask an assistant directly – and receive an answer without ever opening a website. To ensure your business is included in that answer, and accurately, btlabs Core provides a clean, machine-readable version of your information. What we can guarantee: your facts are available completely and without contradictions, and you can see whether they are being accessed. What no one can guarantee: that an AI will recommend you as a result.

• 14. A table of contents for AI systems. A concise overview lists your business, services, key pages, and resources – plus a more detailed version. An assistant doesn't need to search the entire website to understand who you are and what you do. Honestly: The format is an industry suggestion, not an adopted standard.
• 15. Each page also as plain text. The same content, but without layout, menus, and image frames. Language models thus read your message instead of the packaging – and are less likely to misunderstand it. Honestly: Images and design contexts inevitably suffer as a result.
• 16. All your brand facts in one place. Company name, legal form, contact details, service area, services, and tone of voice are all clearly structured and readily available. This is the definitive version of your company data – one source instead of twelve directory entries, each claiming something different.
• 17. Questions and answers in an AI-friendly format. Your FAQs are also delivered as a pure question-and-answer list. An assistant then uses this when someone wants to know something specific – "Are you open on weekends?" will then receive a reliable answer instead of a guess.
• 18. What an assistant can do for you. A separate file describes in machine-readable terms what capabilities an AI assistant has on your website – such as searching or retrieving contact information. Honestly: The underlying standard is still a draft. It's preparation for something that is currently being developed – not a result that can be measured today.
• 19. The technical specifications are located where you would expect to find them. At the usual storage locations for machinery, an automated system finds the descriptions it expects. Anyone searching for you by machine won't give up halfway.
• 20. A way to contact us with tips. One file details how security issues can be reported, a second lists the people behind the website. If someone finds a problem, they report it to you – instead of reselling it. Honestly: The security file is only created if a contact address is maintained.
• 21. An entry point for AI agents. A defined starting point guides assistance systems to your public information – pages, posts, job postings, team, customer testimonials, glossary, FAQs. This way, AI receives organized data instead of transcribed screen text.
• 22. Two things an assistant can trigger. By default, it can search your website and retrieve the contact information. That's usually all it takes to correctly answer a customer question and send the interested party to you.
• 23. Quoting yes, copying no. This disclaimer is machine-readable and stored in all locations where reputable providers check – based on European law. This is your stated position, in writing and dated. Honestly: It's a signal, not a wall. There are systems that don't adhere to it.
• 24. Three switches that you can turn yourself. You decide separately whether your content may appear in AI responses, be used for training, and be accessed for data collection – and whether attribution is required. Most companies want exactly one combination: yes in responses, no in training.
• 25. Well-maintained lists of desired systems. Assistance and search systems are welcome, but robots that only vacuum will be blocked – and the list is constantly updated. You don't need to keep track of which new robots are currently in circulation. Honestly: This landscape is constantly changing; this remains a snapshot.
• 26. Preview services remain permitted. Services that generate previews when a link is shared are exempt from all AI switches. Otherwise, your link would break at the exact moment someone is promoting you.
• 27. The reference to the text version. Each page technically specifies the path to its plain text version. An assistance system can thus find the machine-friendly variant without guesswork.

3. Measuring whether it works: Functions 28 to 48

This is the part that hardly any website provider includes – and the second question every entrepreneur asks: How do I know this is actually making a difference? btlabs Core answers this on four levels. Who visits the site? What do people find via search? Which AI systems are monitoring the site? And: Will you even be mentioned in AI responses, even if no one clicks? One thing to note upfront, because it makes all the difference: These figures come from various sources and are deliberately not added together to form a neat total. And if a value is currently unavailable, it will be displayed as such – never a zero that looks like a result.

• 28. Your own visitor statistics, without Cookies. The analysis runs on your own server, without third-party analytics services and without placing cookies on the visitor's device. You see what's happening without handing over your customers' data to an advertising company. Honestly: The figures are not bot-adjusted – conspicuous access patterns are marked, but not silently excluded.
• 29. The actions that matter. The tracking includes not only page views, but also whether someone read half or all of the message, whether they actually read it instead of just opening it, and whether they clicked on a link, your email address, or your phone number. So you can see if someone actually picked up the phone – not just that they were there. Your own business goals will also be added to the project.
• 30. Numbers for each individual page. Page views, visitors, inquiries, search clicks, rankings, and AI referrals are all aggregated for each page. This allows you to see which pages are performing well and which are simply taking up space – so you know where your time is being put to good use. Also in discussion: "Which page generated the most inquiries last month?" Honestly: The values ​​are constantly recalculated; they are not accurate to the second.
• 31. Ranking and progress. All content is ranked, with its performance over time displayed alongside. Trends become visible without a single weak day appearing like a crisis.
• 32. Where your visitors come from. Direct, via search engines, via other websites, or via a campaign – these are the different approaches. This way you can see whether the flyer, the ad, or the search actually brings in people. Honestly: Browser protection mechanisms and missing information make this not 100% accurate.
• 33. Visitors who came from an AI response. The familiar AI tools are recognized as a separate source. So you'll see people who found you through an assistant. Honestly: This is a minimum threshold – clicks from some apps cannot be technically attributed.
• 34. AI accesses counted separately. Which AI systems read your content, how often, and which pages, operates separately from human visitors. Only then can the question "Does the AI ​​even know me?" be answered. Honestly: Being read does not necessarily mean being quoted.
• 35. Verification of AI visitors' authenticity. The system checks whether the access actually comes from the named provider – or whether someone is just using that name. Otherwise, you'll be celebrating a success that was in reality caused by a third-party data snoop operating under a false name. Honestly: If it can't be clarified, it says "no statement" – never "forged". And the check doesn't exclude anyone, it only sorts.
• 36. Classification according to type and purpose. Every automated visitor is categorized – AI assistant, classic search, SEO tool, availability monitor, preview service – according to its purpose. This separates the valuable assistant from the mere data harvester. Honestly: The classification is based on the visitor's self-reported information and only controls the statistics, never the access rights.
• 37. Server queries do not count as an AI interest. Access from data centers and automated infrastructures is detected and managed separately. Without this separation, the AI ​​figures would look many times better than they actually are – and that's exactly what we don't want to sell you.
• 38. The path of an AI through your website. Successive accesses by a system are linked to a path. You can see whether an AI is only browsing the homepage or working its way through your achievements.
• 39. Is the preparation for AI actually being used? For each machine-readable file, the number of times it is accessed is counted. This proves whether the work is successful or fails.
• 40. Google Search Console connected. At night, the system retrieves your search data: history, actual search queries, pages, and countries – initially going back about a year. You see the terms that people actually use to find you, without having to open a second tool. Also in discussion: "What search terms are people currently using to find me – and does the text on the page match those search terms?" Honestly: You only need to set it up once – about a quarter of an hour. Google delivers with a delay of a few days and doesn't include very rare search queries.
• 41. Bing Webmaster Tools connected. Also at night, with history, search queries, pages, and information on referring links. Bing feeds, among other things, the responses of a large assistance system – looking at that system is significantly more valuable today than before. Honestly: Bing does not provide positions over time, and the link data is Bing's view, not a complete analysis.
• 42. Can you be mentioned without anyone clicking? This is the gap inherent in all standard statistics: If an AI recommends your business without the person asking visiting your website, you won't generate a single data point. That's why the Citation Monitor asks questions itself. You define your own test questions – just as a customer would ask them – and these are regularly sent to several major language models. Some of these models search the web live, while others answer from memory. The analysis determines whether your business was mentioned, where it was mentioned, and with what level of quality – from "not at all" to "first recommendation." The difference between the two groups is the key insight: one shows the impact of your current content, the other reveals whether you've made it into the models' memory. Honestly: This feature is not enabled by default, is set up upon request, and incurs costs via your AI account with each run. The questions must not contain your name, otherwise the measurement will be self-incriminating. And it remains a sample – not a ranking you can rely on.
• 43. Who you are really competing against in AI answers. The same answers reveal which other providers are mentioned. Often these are different from those you would expect to find in your area – and that's a useful insight.
• 44. What's getting in the way on your website. Every day, the entire site is checked for typical weaknesses: pages that no one links to, dead ends, broken links, pages that compete for the same search term, and insufficient content. Each page receives a score. This is a lever that a small business without an advertising budget can use itself. Also in discussion: "Which pages are orphaned? Suggest where I should link to them from." Honestly: The grade assesses the technical aspects, not whether the text is good.
• 45. The quick chances. Search terms for which you rank just below the first page of results are highlighted. An afternoon's work in these areas will yield more results than a new campaign. Also in discussion: "Where do I stand just behind page one, and what should I change there?"
• 46. ​​When your own pages are competing with each other. If two of your pages are competing for the same search term, it will be noticeable. Two weak pages on the same topic are worse than one good one – this is one of the most common mistakes on established websites.
• 47. If a value is missing, it will be indicated there. If something cannot be retrieved, it's labeled "not retrievable"—not zero. Each metric also indicates its source. This may sound minor, but it's the difference between an analysis and a gut feeling with a chart around it.
• 48. Requests and targets as a file. The collected data can be exported as a table. You can then integrate it into your customer management system or your own annual report.

The most honest number in the entire system is the one that says: "not available". Anyone who reports a zero instead is selling a measurement problem as a result.

4. Tempo: Functions 49 to 57

Speed ​​isn't a value you can simply write down once – it's the result of multiple decisions. And it's the first impression: anyone who waits three seconds for your page to load isn't going to call.

• 49. The page is simply there. Pre-built pages are delivered instead of being reconstructed every time the page is accessed. Even if twenty people are using it simultaneously, nobody notices.
• 50. Changes are immediately visible. When you save, only the affected part is updated – not the entire website. You click save and see the result. No "wait for the cache to clear," no nightly run to wait for.
• 51. A net underneath. Additional storage layers are also used. Even if something goes wrong somewhere, your site remains fast and accessible.
• 52. What this means in practice. Our own website responds within a tenth of a second – even when multiple visitors are accessing it simultaneously. With the previous technology, it took approximately five times as long. Honestly: This is our measurement on our machine, not a guaranteed performance figure. And the proof won't cost you anything: Visit our website and compare it to yours.
• 53. No organically grown extension tower. The public website does without the stack of interdependent add-on modules that one usually accumulates over the years. Less bloat means: faster speed, fewer update conflicts, and less that can break.
• 54. Fonts from the device, not from an external server. By default, the fonts already installed on the visitor's device are used – no reloading, no flashing during setup, no connection to a font service. A custom font is possible and will then be deliberately included in the design.
• 55. Images are loaded when needed. Content located far down the page only loads when someone scrolls down. This makes the first impression much faster – especially on a mobile phone, in a valley, with only one bar of signal.
• 56. Nobody notices the updates. During an update, requests are kept brief instead of being rejected with an error page. Nothing unusual happens to your visitors. Honestly: This is based on experience gained in the company, not a guaranteed upper limit.
• 57. Numbers for each individual page. If something isn't working properly, you can see which side it affects, instead of groping in the dark. Honestly: Usage is measured, not the loading time for individual visitors – there is deliberately no tracking of this in your customers' browsers.

To ensure this remains the case in the long term, a binding agreement is needed. performance-Budget for SME websites more important than a good value on the day of release.

5. Multilingualism: Functions 58 to 71

In South Tyrol, this isn't an add-on feature, but a fundamental part of the business. German, Italian, and English are therefore equally integrated into the core – not as a translation tool that's added on later. One thing to note right away: the system ensures that you can maintain each language equally . Whether a language is actually maintained remains your editorial decision – and that's precisely why the system shows you where something is still missing.

• 58. Three languages, one source. Each language has its own texts, addresses, and search terms – but everything is stored in the same database. Instead of three separate websites requiring three different levels of maintenance, there is one website with three versions. Honestly: Adding a fourth language isn't a simple click, but a guided transition.
• 59. Separate addresses for each language. Your Italian website has an Italian address, not the German one with an abbreviation appended. The customer you send the link to will notice this.
• 60. Search engines recognize what belongs together. The language versions are mutually marked as alternatives. Therefore, your Italian page is not considered a copy of the German one. Honestly: Which version a search engine displays in each individual case is decided by the search engine itself.
• 61. Search terms per language. Titles and descriptions are maintained separately. The Italian version doesn't have to adopt the German wording but can instead use the terms that are actually searched for in the Italian market. This is often a different word altogether, not just a translation.
• 62. You see where something is missing. For each piece of content, it's clear whether a language version is complete, partially completed, empty, or just an untranslated copy. You'll notice the gap—not the Italian customer on the phone. Also in discussion: "What is missing in Italian?"
• 63. Copying is not translating. If the same German text is in the Italian field, it is displayed as a separate status instead of being waved through as "done". Honestly: The system recognizes a word-for-word copy – but no system can judge whether a translation is good.
• 64. I'd rather have nothing but German on the Italian side. If a page isn't translated, it's intentionally unavailable in the other language. An Italian customer shouldn't land on a German page and perceive the company as unprofessional – this is a deliberate decision, not an oversight.
• 65. The user interface is also trilingual. Those who maintain content on your site work in their own language – down to every field label and help text. No half-translated forms where half remains in English.
• 66. The short texts are also correct. Button labels, error messages, dates, instructions – everything is available in each language. No English "Submit" on an Italian form.
• 67. Also the error page. The "Page not found" page is a fully customizable page for each language. A typo in an address therefore doesn't result in a simple error message, but instead leads to your services.
• 68. Get it translated cleanly, without layout chaos. All translatable text segments can be exported cleanly and imported back just as cleanly. A translation agency can work with this without the formatting being broken afterwards – anyone who has ever sent a Word document back and forth knows this.
• 69. Entire stacks at once. Multiple documents can be translated in one go, preserving the formatting and using your own terminology. What used to take an afternoon becomes a half-hour proofreading session. Also in discussion: "Translate the new service pages into Italian and submit them as a draft." Honestly: This requires a registered AI access credential. Without it, copying and manual translation remain the only options.
• 70. Help for everyday language use. Transfer content from one language to another, switch to a specific language, and view the current status. The classic "typing everything twice" is no longer necessary.
• 71. The search function knows the inflection forms. Anyone searching for "Sanierung" will also find "Sanierungen" – in all three languages. Honestly: For another language, a simpler search without inflectional forms is used.

If your business targets multiple markets, each language needs to be planned independently – not just translated. The article about SEO for multilingual websites addresses this point.

6. Content and Editorial Staff: Functions 72 to 101

This is where it's decided whether your website will still be up-to-date after two years. A good editorial interface needs clear states, data that only needs to be entered once, and it must show errors before the customer sees them. Much of this happens in the background – the best guardian is the one you never notice.

• 72. Approximately 20 areas for your content. Pages and subpages, articles, questions and answers, glossary, job postings, team, customer testimonials, images and documents, menus, text modules, notification banners, categories, company data, inquiries, mailing lists, campaign links, redirects, and an internal knowledge base. Each area is independently maintainable and multilingual – that's the difference between a well-organized database and a website that has haphazardly grown over the years.
• 73. Twenty building blocks you can use to build pages. Text, image, image with text, map grid, gallery, logo wall, button, form, registration, search field, automatic lists, map, team, customer testimonials, and more. Your team can assemble new pages from these elements themselves, without anything shifting and without having to call us. Also in discussion: "Create a service page for bathroom renovations, with text, image and contact button." Honestly: Our own website uses 43 building blocks, 26 of which are custom-built. You don't automatically receive these – you start with the first twenty; everything else is project work. We mention this because screenshots of our site would otherwise promise more than it actually delivers.
• 74. Building blocks in the middle of the text. An image, a form, or a note can be inserted directly into running text. A post doesn't have to consist of rigid blocks stacked on top of each other.
• 75. You can see where a building block is used. Before you change anything, you know what you're doing. That saves you the unpleasant discovery three pages later.
• 76. Preview in three device sizes. Changes appear instantly in the actual design – on mobile phones, tablets, and screens. You'll notice any display errors, not your customer. Honestly: Live previews are available for content with its own page; they are deliberately not available for individual FAQ entries and customer testimonials.
• 77. Draft and approval separately. You can prepare in peace without anything becoming public, and then publish consciously. Working and publishing are two different things.
• 78. Schedule it for a specific date. Content can publish itself – and withdraw itself again. The seasonal promotion goes live on time, without anyone having to get up early. Also in discussion: "Set the post to the beginning of next month and then withdraw it after two weeks."
• 79. Reclaim former states. Older versions of all content are retained. Accidentally overwriting text is therefore no longer a problem.
• 80. Data is continuously cached. The system saves your progress during the editing process. Honestly: There is still a save button – “nothing is ever lost” is a promise we do not make.
• 81. Wastebasket. Deleted items first go to the trash and can be retrieved. Honestly: Anyone who deletes via the machine interface bypasses the recycle bin – therefore, this right can be deactivated there.
• 82. Who changed what? Registrations, content changes, and external access are logged. If someone claims "that wasn't me," a response will be provided. Honestly: The protocol shows that changes were made – not what was there before. It is not legally valid proof.
• 83. Maintain company data only once. Address, phone number, opening hours, tax ID number, and much more are all in one place and appear automatically everywhere – including in emails and forms. New phone number? Change one field. No need to search through thirty pages and still forget one. Also in discussion: "Enter the new phone number."
• 84. Text modules that are the same everywhere. A block like the directions information is maintained once and used in any number of places. If it changes, it changes everywhere simultaneously.
• 85. Information tapes with start and end dates. Company holiday closure, holiday opening, a promotion – in several formats and positions. A matter of minutes, and it disappears again by itself.
• 86. Assemble menus instead of programming. You build the navigation yourself, with custom labels for each language.
• 87. Explain technical terms. Your glossary links automatically within the text, even for synonyms. This helps customers who are unfamiliar with the term – and clearly shows search engines and AI systems what you stand for. Honestly: Automatic linking works in the body text, not in headings or on buttons.
• 88. One answer, three places of operation. Your FAQs are maintained, displayed on the page, and also provided in a machine-readable format. The same answer works on the website, in Google search results, and in AI-generated answers. Written once, used three times. Also in discussion: "Attach the three questions that customers constantly ask me on the phone – I'll dictate them to you."
• 89. Job postings that expire automatically. A career page with real job postings instead of PDF attachments. Users only see available positions.
• 90. Your people with faces. Individual profiles are maintained with their role, short biography, and qualifications, and linked to their contributions. Expertise is given a name – this builds trust with customers and also counts for search engines.
• 91. Customer reviews: clean and well-maintained. References are recorded in a structured manner and incorporated into the machine-readable data. Honestly: Whether a search engine uses this information to create stars is entirely up to the engine itself – and it regularly changes the rules for doing so.
• 92. Track which advertising campaigns were effective. For advertisements, mailings, flyers, or social media posts, you compile trackable links. This way, you know which actions generated inquiries – and which only cost money. Honestly: The result is complete, labeled addresses – not a short link service with its own counter.
• 93. Internal knowledge stays internal. Price lists, procedures, and notes are stored separately from public content and are only searchable by authorized users. Information that would otherwise disappear into email attachments remains accessible.
• 94. Important pages are protected against accidental changes. The search function, FAQ overview, article archive, thank-you page, mailing lists, and error page can be customized, but not accidentally deleted. No one would inadvertently disable the contact channel.
• 95. A framework for the start. Basic pages, system pages, forms and menus can be imported when the project starts. Honestly: This is a functioning framework, not a finished advertising text – and later improvements will not automatically be incorporated into an existing installation.
• 96. Mass maintenance via a table. Content can be exported as a table and imported again – useful during relocations and major changes. Honestly: This is explicitly administrator work, not routine editorial work. Image files themselves cannot be imported this way.
• 97. You don't see what you don't need. Posts, job postings, FAQs, customer testimonials, mailing lists, contact forms, notification banners, and the glossary can all be individually enabled and disabled – along with everything that goes with them. A business without a blog won't be overwhelmed by having one.
• 98. Three roles, clearly separated. Administrator, editor, reader. The trainee maintains posts without having access to the company settings.
• 99. The mandatory Italian information is provided. VAT identification number, tax number, Chamber of Commerce, REA number, share capital, SDI code, and certified email address each have their own fields. For a business in Italy, this isn't an optional extra, but mandatory – and in most systems, it's a free-text field where you simply enter the information.
• 100. The silent helpers in the background. Addresses are generated cleanly, redirects are automatically created when renaming, breadcrumbs are preserved, internal links survive address changes, broken links are detected, formatting remnants are cleaned up, and preview images are updated. These are precisely the errors that accumulate on organically grown websites after two years – here, they don't even arise in the first place.
• 101. Ordering with the mouse. You move what should be at the top to the top. Your most important achievement is there because you moved it there – not because it was placed there first by chance.

7. AI assistance in the system: Functions 102 to 114

Artificial intelligence assists the editorial team here. It does not replace your professional responsibility or human approval. Three things apply to this entire chapter: All AI functions require AI access, which your company provides – without it, the buttons simply won't appear. Access is granted through your account, not ours. And every single function can be deactivated.

• 102. Translation suggestions. Entire documents are translated with accurate formatting, using your own terminology and in your tone of voice. What used to take an afternoon of translation work becomes a mere half-hour review. Honestly: This is a suggestion, not an approval. Industry terminology, regional language, and all legal aspects must be reviewed by a qualified language professional – without exception, especially in legal texts.
• 103. Titles and short descriptions. Suggestions for each language are generated from the existing text. This prevents the empty required fields that otherwise accumulate on many subpages – you choose the options yourself.
• 104. Image descriptions that nobody else does. Each uploaded image is automatically assigned a title, alternative text, caption, and description – in every language – along with the image center point for cropping. This is the painstaking work that almost everywhere fails, and it determines accessibility and image search. Honestly: This happens during the upload process and only with a stored AI access.
• 105. Pre-fill fields. A few keywords become a draft. Better than sitting in front of a blank form and not starting.
• 106. A complete scan of the entire website. Missing image captions, weak search terms, inconsistencies in tone, gaps in content, and contradictions are reported. You'll be notified of these problems without having to read every page yourself. Honestly: It's switched off by default; when switched on, it runs weekly. And it suggests things, it doesn't correct them automatically.
• 107. Keep your house rules in shape. The stored trademark and language rules are checked for duplicates, contradictions, outdated information, and gaps. This prevents the rulebook from becoming a collection of scraps of paper that no one ever touches.
• 108. Suggestions for what you should link to. Suitable targets are suggested for each draft – starting with pages that no one has linked to yet. This makes forgotten subpages accessible again. It's one of the few levers that works without an advertising budget.
• 109. Improve search results per page. Your pages will be reviewed and you will receive specific suggestions for improvements that you can implement. Honestly: This applies to content with its own address, not to embedded components.
• 110. The tone of voice remains yours. Your tone of voice, your preferred terminology, and the words that have no place in your work are incorporated into every AI-generated task. Your texts will sound like your business, not like they were written by a text generator.
• 111. The right model for every task. Translation, image description, content check, and pre-filling can be configured separately – a budget-friendly model for simple tasks, a powerful one for complex ones. Switching between modes requires no technical expertise.
• 112. Your access, your consumption, visible. AI access is stored encrypted, displayed only in masked form, and reserved for administrators; every use is recorded along with its usage. You see the cost of AI usage instead of just guessing. Honestly: You set a Euro limit in your AI account, not on the website.
• 113. Limits per person. The AI ​​functions have upper limits. This protects against accidentally triggered bulk runs that would appear on the bill at the end of the month.
• 114. Nothing works live without approval. AI-generated text on pages and posts is placed in draft status and must be approved. Image captions, brand profiles, and menu items do not have a draft status; instead, only empty fields are filled, and existing content is never overwritten.

AI is a tool working in the background here. A human decides what is technically correct and what gets published – and the access through which this happens belongs to your company.

8. Maintaining the website via chat: Functions 115 to 129

This is the part that most people initially consider a gimmick, but after a week they wouldn't want to give up. Your website can be connected to an AI assistant – and it works not only with the text, but with the entire system: with your metrics, your pages, your queries, your languages.

The crucial point is that reading, evaluating, and acting should converge in a single conversation. Otherwise, you're opening one tool for visitor statistics, another for keywords, the content management system for the text, and a third for translation—and a bit of motivation gets lost at each step. Here, it's a conversation. A few phrases that actually work:

Read and understand

• "Which pages generated the most inquiries last month?"
• "What search terms do people use to find me on Google and Bing?"
• "Which AI systems have read my page, and which pages interest them?"
• "Was my business mentioned in the AI ​​responses in the last run?"

Have it assessed

• "Go through my performance pages and tell me where something is missing or contradictory."
• "Which pages are abandoned – which ones do no one link to?"
• "Where am I just behind the top page of results, and what should I improve there?"

Let it be done

• "Create a service page for bathroom renovations and add it to the main navigation."
• "Turn this text into a blog post with subheadings."
• "Translate it into Italian and file it as a draft."
• "Change our opening hours on the homepage."
• "Set the payment to the beginning of next month."

The time saved is not in the fact that a single task is completed faster. It is in the fact that the tasks are done at all – because the hurdle of "I should really sit down" is reduced to "I'll just say it quickly."

• 115. An assistant who truly collaborates. He can read and edit content without anyone logging into the editorial system. For a company without a marketing department, that's the difference between "I'll do it when I have time" and "it's done." Honestly: You decide how much he is allowed to do – through the rights you grant him.
• 116. Create, modify, search, delete. Everything an editor would click can also be done by voice command. Honestly: Certain elements are deliberately locked – images and internal documents cannot be re-uploaded via chat.
• 117. And much more than that. Translate, search your content, schedule posts, import images from the web, suggest links, check available building blocks, format submitted text. And above all: Read your reports Visitor numbers, search data from Google and Bing, metrics for individual pages analyzed by AI systems, internal vulnerability analysis, and citation monitor results. The assistant works with your website instead of just talking about it – and it talks about real numbers, not assumptions.
• 118. You give him permission to do what he is allowed to do. This is decided separately for each area and each action. For external access, the default setting is: read and write, but not delete. To be honest, and this is the most important point here: Anyone deleting via this interface bypasses the recycle bin. This is permanent – ​​which is precisely why this right can be disabled and is off by default.
• 119. Register by consent. Common AI chat applications register via the usual consent process, instead of passing around a key via email. Honestly: Access granted in this way currently has the same scope as a full key – a fine selection in the consent window is not yet available.
• 120. You see what he did. Every access is logged, even simple read access. No blind trust is required.
• 121. Secrets stay secret. Login credentials, keys, and internal notes are automatically anonymized during chat access. The assistant sees your content, not your keys.
• 122. Strange assistants can access your public content – ​​and nothing else. An AI service can correctly answer customer questions without having access to your system.
• 123. Brakes, just in case. Too many requests in too short a time will be rejected. A malfunctioning assistant won't cripple your website.
• 124. View first, then publish. A draft can be previewed and reviewed before it goes live.
• 125. Copy images. The assistant retrieves images automatically and saves them with alternative text, title, caption and copyright information. Honestly: Only images from public addresses – no files from your folder, no documents, no videos.
• 126. He doesn't advise, he looks. The assistant can search your pages, posts, glossary, FAQs, job postings, and team information. It doesn't invent anything.
• 127. It remains in the brand design. It can query colors, typography, and spacing. What it creates, however, doesn't look colorful.
• 128. He knows the layout of your fields. Fewer failed attempts, fewer half-completed entries.
• 129. He reports what he notices. Contradictions, inconsistencies, gaps, and translation problems end up in an inbox that is reviewed by a human. The AI ​​doesn't secretly change your house rules—it suggests them.

9. Inquiries and contact: Functions 130 to 143

An inquiry is the moment your website starts to pay off. That's why this covers the entire process – from the completed form to documented consent. What btlabs Core doesn't do: sales, follow-up, closing the deal. That stays with you.

• 130. Building forms without programming. You can assemble what you need from pre-defined field types: text, longer text, email, number, drop-down list, checkbox, note. A callback form is ready in ten minutes – without having to call anyone. Honestly, and beforehand: There are no file uploads, no date fields, no logic such as "field B appears if A is checked," and no multi-page forms. Therefore, an application with a resume attached is not possible.
• 131. Errors are explained during completion. Anything missing or incorrect is clearly indicated right next to the field. The visitor corrects it themselves – and you spend less time chasing after incomplete inquiries.
• 132. Spam protection without a puzzle image. Automated submissions are rejected without a real customer having to click on traffic lights or bicycles. Honestly: This keeps out the usual mass spam. It won't help against someone who is specifically targeting you.
• 133. A brake on spam waves. If an unusually large number of submissions arrive in a short period of time, it kicks in. Your inbox isn't filling up. Honestly: This is a second line of defense, not protection against an overload attack.
• 134. No request is lost. Every order is stored in the system, regardless of whether the email was received – and can be displayed as a table. No order disappears because a mailbox was full or the email ended up in spam. This happens more often than you might think.
• 135. You can see where the request came from. Campaign, referring page, country, and time are all recorded. This allows you to prove which channel actually brings you customers. Honestly: This is deliberately more than just the email address and therefore belongs in your privacy policy – ​​the accompanying text describes it.
• 136. Notification and confirmation of receipt. Two emails are prepared: one for you, one for the prospective customer – each in three languages, with automatically inserted company data. The prospective customer knows immediately that their inquiry has been received, instead of wondering whether the form even worked. Honestly: The request is saved immediately, and the email follows shortly thereafter. If the transmission fails, it is deliberately not repeated – better a missed email than a duplicate.
• 137. A thank-you page that continues. Instead of a grey confirmation line, the next sensible step is: directions, references, or the phone number for urgent matters.
• 138. Mail sent via your route. Optionally via a specialized mailing service or via your own mail server, with test sending. Honestly: To prevent your emails from ending up in spam folders, your domain needs to be verified with the email service provider. This step is part of the project.
• 139. You will be notified if an email does not arrive. If the recipient server reports an invalid address, this will be processed. Honestly: This is only possible when sending via the specialized service, not via your own mail server.
• 140. A distributor that can withstand scrutiny. Registrations are processed via genuine double consent with a time-limited confirmation link, documented history, and automatic deletion if no one confirms. In case of a dispute, you can prove that consent was given. To be honest, and this is important: There is no automatic mass mailing to the distribution list. btlabs Core builds and maintains the list cleanly – sending out an issue is handled by a mailing tool of your choice.
• 141. Unsubscribe, request information, delete – without you. Three options are available via a personal link, entirely without an account. Data subject rights are addressed without anyone having to touch you. Honestly: This applies to mailing list subscribers. For form requests, it is handled via a tool in the administration area.
• 142. Old data disappears by itself. Form submissions are deleted after six months, unconfirmed registrations after a few days. Data minimization is happening, rather than remaining a good intention. Honestly: The deadline is fixed. Anyone who needs to keep requests for a longer period should clarify this before the start.
• 143. Your specialist systems can use it. Inquiries and forms can be retrieved from an inventory management or customer management system. Honestly: The website doesn't send anything automatically. A connection that actively forwards data is a necessary upgrade.

10. Media and Images: Functions 144 to 152

Images are the most common cause of slow websites – and the most common source of unintentionally published data. Both are handled here in the background, without you having to think about it.

• 144. Everything in one place. Images, videos, and documents are stored centrally, with sensible size limits. One storage location instead of five folders on three computers and a USB stick in a drawer. Honestly: Anyone who knows the address of a file can access it. This is perfectly fine for photos – confidential documents don't belong there.
• 145. Each image in the appropriate size. A single photo is automatically converted into multiple sizes, plus a cropped version for sharing on social media. A mobile phone loads the smaller version, a large screen the larger one – nobody loads the original directly from the camera. This is the most common reason why a website becomes slow.
• 146. No more severed heads. You decide what is important in the image – all cropping follows automatically.
• 147. Readable file names. "IMG_4711.jpg" becomes a name that people understand. This helps with image searches and makes your files searchable even after two years.
• 148. Location and camera data are removed. Uploaded photos automatically have the location, camera model, and similar information removed. This prevents a private address from appearing in a company photo simply because it was taken at home. Honestly: This applies to common photo formats. Vector graphics, PDF, Office, and video files retain their additional information.
• 149. Supplied graphics are checked. Logos and graphics are cleaned of any potentially malicious code before being stored. This prevents a logo someone sends you from becoming a gateway for attack.
• 150. Image rights under control. Copyright and license information are stored directly on the image itself and can be displayed in a single, comprehensive document. If someone asks, you don't have to search through old invoices. Also in discussion: "Get the picture from this address, label it and write down the author's name."
• 151. AI-generated images remain recognizable as such. If an image bears a machine-readable label indicating that it was generated by AI, this label remains – relevant for the transparency obligations of the European AI Regulation. Honestly: What remains is the simple labeling, not a cryptographically signed proof of origin.
• 152. Icons for browser tab and home screen. The necessary sizes are created from a template – a detail that is almost always missing on organically grown websites, yet everyone notices it.

11. Search on your own website: Functions 153 to 157

From about thirty subpages onwards, your own search determines whether a visitor finds what they are looking for – or whether they leave again.

• 153. Built-in full-text search. The search is conducted across pages, posts, job postings, glossary, and FAQs; drafts are excluded. No external search service, no data sharing, no monthly fee.
• 154. Even in the case of typos. If a normal search yields few results, a typo-tolerant search using titles is used. "Sanitation" still leads to a result. Honestly: This applies to the titles, not to the entire body of text.
• 155. You can see beforehand whether it fits. For each hit, an extract is generated with highlighted results.
• 156. You'll find out what your customers are looking for. The search terms are collected anonymously. This is the most honest topic list available – your customers tell you what features are missing on your website. Honestly: For this to work, your own statistics must be active.
• 157. Internal search separate. The internal knowledge repository has its own search function, which is only accessible to authorized users and is strictly separated from the public area.

12. Data sovereignty and data protection: Functions 158 to 170

This is the point that comes up most frequently in the conversation – and the one that needs to be formulated most carefully. The current situation: As we deliver, your website doesn't meet any requirements. Cookies For visitors, it loads nothing from external servers and contains no advertising trackers. Therefore, our position is that a btlabs Core website can function normally without a consent window. This is a reasoned opinion, not legal advice – the responsibility remains with your company. And as soon as you integrate anything yourself, the situation changes.

• 158. No external trackers – there isn't one inside. No ad tracker, no ad pixel, no third-party analytics service. There's nothing to disable because nothing is installed.
• 159. No cookies for your visitors. Only one cookie exists, and that's only when you log in to the editorial system. This is the technical reason why you don't need a consent window.
• 160. Nothing is loaded from external servers. No scripts, no fonts, no external images. Only one exception is permitted: map tiles – and only if you use the map module. In that case, the map service will see your visitor's address, and the accompanying privacy policy clearly states this.
• 161. IP addresses are not stored. Statistics and customer data do not contain plain text, but rather a regularly changing code from which nothing can be calculated backwards. Honestly: The web server's technical log still contains IP addresses and is stored for several weeks. Therefore, saying "we don't store IP addresses" would be an overstatement.
• 162. A deletion concept that actually exists. For each type of data, it's defined how long it's retained and why – and it's automatically cleaned up at night. Roughly speaking: visitor data for about a year, inquiries for six months, and logs for a few months. In an audit, this is the first question – here, it's answered before it's even asked.
• 163. Error logs without personal data. Before anything is logged, personal information and login details are removed. Diagnosis yes, data collection no.
• 164. Legal texts in three languages ​​are enclosed. Legal notice, privacy policy and terms and conditions as well-maintained pages, in the format for an Italian SME and with explicitly named third-party providers. Honestly: This is a framework that should be adapted to your business and, if in doubt, reviewed by a lawyer. It is not a legal guarantee.
• 165. Your own server. Each business runs on its own isolated installation – no shared host, no shared database with other websites. If something goes wrong elsewhere, that's not your problem.
• 166 locations in the EU. Operations and backups are located in European data centers. Honestly: Some additional services are run via external providers – for example, AI-powered text processing is handled by a US-based provider, whereby only approved website content is transferred, not your customers' data. Therefore, it is correct to say: your data and backups are located in the EU. Saying "all data exclusively in the EU" would be an overstatement.
• 167. It belongs to you. Texts, images, and company data are assigned to your business, and the core system is not tied to any ongoing usage agreement. You are buying the house; you are not renting it.
• 168. You can access everything at any time. Content, media, and basic settings can be exported. Honestly: An export is a complete data copy – not a website that will immediately run elsewhere.
• 169. No fee per access. There is no recurring fee per user. Everyone who is supposed to maintain a profile gets their own access – instead of sharing one for cost reasons, which would render any log worthless anyway.
• 170. Where the banner renunciation ends. As soon as you integrate something yourself – a video portal, a booking window, a review widget, an advertising pixel, or an advertising company's map service – you need a consent window. A consent solution is not included in btlabs Core. This should be discussed before the project, not after.

13. Security and Access: Functions 171 to 184

Security is achieved through limitations, updates, and clear responsibilities. No publicly accessible system can promise absolute security – what's described here is diligence applied across multiple layers. It's not a seal of approval.

• 171. Encrypted throughout. The connection is secure, and the usual browser security settings are in place. No technical information about the server is disclosed externally.
• 172. Two-factor authentication. Editorial access can be further secured with a one-time code, including emergency codes. A guessed password alone is then no longer sufficient. Honestly: This can be activated per person, but not enforced for everyone – we recommend it and are happy to set it up.
• 173. Short passwords are not accepted at all. The measure with the best cost-to-effect ratio – and the one that is most frequently missing.
• 174. After several failed attempts, it's over. The account will be temporarily blocked, and the number of attempts per connection is also limited. Automated attempts will be unsuccessful.
• 175. Brakes where it gets expensive. All access methods have tiered upper limits, depending on how complex a process is. Honestly: This slows down abusive mass access at the application level. It won't stop a large-scale denial-of-service (DDS) attack – that's what the upstream firewall is for.
• 176. No covert actions. Every change request is checked to ensure it truly originates from your website. This is the standard protection against an external site in the background triggering actions on your behalf.
• 177. Everything that comes in is cleaned. External content and uploaded files are cleaned of potential malicious code in multiple layers. Honestly: One deliberately unfiltered component remains reserved for the administration – this is a tool, not a leak.
• 178. Access data encrypted. Access to connected services is encrypted and only displayed in a masked form. Honestly: The connection, the backups, and these secrets are encrypted. The content itself is stored unencrypted in the database, and the server's hard drive is not additionally encrypted.
• 179. Rights also apply at the interfaces. What an assistant sees, what a public query provides, and what a role is allowed to do in the user interface are controlled separately. Three layers instead of one switch.
• 180. Incorrectly set means: it won't start. During startup, the system checks critical settings and, if in doubt, refuses to start. An incorrect setting results in a startup error – not a data leak during operation.
• 181. Automated scans fail. They are blocked at the entrance before they can consume any processing time. Such scans affect every publicly accessible website, all day long – this is normal and not a sign that anyone is targeting you. Honestly: The patterns actually observed are covered, not every conceivable shape.
• 182. Only what needs to be open should be open. At the network level, everything that isn't needed is closed, and maintenance access is automatically blocked after failed attempts. This keeps out the constant barrage of attacks that would hit every server on the network.
• 183. Checked weekly against known vulnerabilities. All deployed components are checked against public vulnerability databases; serious and critical findings are reported and closed. Honestly: These are internal audits. There is no certification and no external penetration test – anyone who needs something like that should explicitly request it.
• 184. Security tests before every live launch. A safety flaw is noticed during construction, not by your customers.

14. Operation, security, reliability: Functions 185 to 200

Reliability means: changes are controlled, errors are identified, and data remains recoverable. This requires technology – and it also requires clarity about who is responsible for what.

• 185. Secured and encrypted every night. Database, images, and settings. Nobody has to think about it.
• 186. Backups that nobody can delete. Backups from the last thirty days are locked against deletion at the storage provider – even someone who has gained access credentials cannot access them. In addition, there are monthly archives and an image archive that only grows. This is the difference between a backup and protection against ransomware, which looks for backups first. Honestly: The deletion restriction applies to short-term backups, not to monthly and image archives.
• 187. A copy outside the data center. A copy is downloaded daily to an external computer; missed runs are made up. If the cloud provider fails, that's not the end of the world. Honestly: This copy is tied to a specific machine and may lag behind slightly during extended absences.
• 188. Restoration is practiced, not hoped for. Restoring a backup involves a separate verification process against a disposable database. A backup that has never been restored is not a valid backup. Honestly: This process is triggered; it does not run automatically according to a calendar.
• 189. What we plan to do when things get serious. Back online within two hours, maximum one day of data loss. Several emergency scenarios have been tested using a manual – in a real emergency, there will be no improvisation. Honestly: These are our target values, not a contractually guaranteed availability. We deliberately do not promise a percentage.
• 190. A report on the condition of the fuses. With warning thresholds so you can see if something is wrong. Honestly: There is intentionally no automatic alarm notification in case of a backup failure – the report must be accessed manually. The external monitor only sends a notification if the website goes down.
• 191. A checklist is performed before every live broadcast. Programming style, trilingualism, database discipline, correct wiring, machine-readable files – plus a comprehensive series of automated tests. Any error will be detected beforehand, not on your end.
• 192. What doesn't start, doesn't go live. The new version must fully boot and respond; otherwise, the switchover won't even occur. A broken update won't reach your visitors.
• 193. Checking after the update. The most important addresses are queried during live operation – including those that nobody normally looks at. Not just the homepage.
• 194. Monitored from the outside. Your website is constantly being checked for accessibility. Honestly: The system detects a failure – not that the site has become slow or that the domain is pointing somewhere else.
• 195. And from the inside. The website, database, application, and statistics continuously check themselves. A stalled service is restarted before anyone calls.
• 196. Security updates run automatically. The restart happens at night and only when necessary. There's no maintenance window that anyone needs to manage.
• 197. Regularly rebuilt. The building blocks used are renewed weekly, with automatic rollback if something isn't right. Security vulnerabilities in third-party building blocks disappear on their own, instead of remaining dormant for years.
• 198. Back to yesterday, in minutes. Previous states remain available, the current state can never be deleted. A regression is a normal occurrence, not a rescue operation.
• 199. Daily accessibility check. The website is automatically checked against the WCAG 2.1 AA standard. Let's be honest about it, and this is important with the new obligations: An automated test only covers the machine-testable part and deliberately does not block publication. "Automatically tested" is different from "compliant"—anyone needing a reliable statement needs a review within the project.
• 200. The behind-the-scenes work continues even when no one is looking. Clean-up, data matching and recalculations are running according to plan, and a monitor restarts what gets stuck.

15. Interfaces and other channels: Functions 201 to 207

Because content and presentation are separate, more than one website can be created from the same database.

• 201. Your other systems can use it. Released content is ready for use by inventory management, customer management, or an app. Your opening hours don't need to be entered twice in three different systems. Honestly: They collect the data – the website doesn't send anything automatically.
• 202. You can see if the machine access points are being used. What is publicly offered appears in the evaluation. This allows you to see whether the preparation was effective or just well-intentioned.
• 203. Campaign pages from the same base. A campaign page accesses existing company data and building blocks instead of being maintained separately. It's available in hours rather than weeks – and the contact details are automatically updated.
• 204. Take your settings with you. Configurations can be output and read in, for example for a test environment or a second appearance. Honestly: Moving content between test and live is the operator's job, not a button in your cockpit.
• 205. Social profiles with an insertion. When you insert a profile address, the system recognizes the platform and inserts the correct icon – for all common networks. The bar in the footer is therefore a simple insertion task, not a design challenge.
• 206. An original design, not a purchased template. Colors, typography, and spacing are a system all their own. Your website won't look like ten thousand others – and what the AI ​​generates will automatically remain in your design.
• 207. Prepared for more. Other brand, location, or topic pages can technically access the same foundation. Honestly: Being prepared means being prepared – every additional appearance requires concept, design and implementation.

What is deliberately omitted

This list is just as important as the 207 features listed before it. Not every extension is suitable for every business – every additional function means effort, responsibility, and maintenance. What's listed here is not included. Anyone expecting something different should find out before the project, not after.

• No newsletter distribution. btlabs Core builds and maintains your mailing list cleanly – including double consent, unsubscription, information requests, and deletion. Sending an issue to all subscribers is not included; we integrate a separate mailing tool for that.
• No file uploads in forms. Applying for a job using a resume or damage photo via a form is not possible today. Date fields, conditional field logic, and multi-page forms are also missing.
• No consent solution for embedded services. Without third-party content, the page doesn't need a consent window. If you embed a video or booking form yourself, you need one – it's not built in.
• Not a ready-made online shop. Product range, inventory, payment, and shipping are one stage of development. A shop only comes into being when processes and economic benefits are clarified – not on a hunch.
• No protected customer area. Registration, personal documents and special prices for regular customers are prepared, but not yet built.
• No appointment booking. Requesting an appointment via a form is possible. Calendar synchronization, availability checks, and binding bookings require a separate process.
• No ready-made price calculator. Calculation rules and binding prices differ too much from company to company for a standard solution.
• No AI agents acting on the job. An assistant can read, write, and maintain content. Independently making reservations, placing orders, or changing business data is not part of their role.
• No active connection to specialist systems. Your inventory management system can retrieve data. The fact that the website itself sends data is a further development step.
• No short link service with its own counter. Campaign links are complete, traceable addresses.
• No worldwide distribution network. Prepared, but not active in normal operations – for a regionally active company, it's pointless.
• No forced two-factor authentication. It is available and we recommend it, but it cannot be enforced for all accounts.
• Availability is not guaranteed. We have target values ​​for recovery and data loss, but no contractually guaranteed percentage – and we don't claim to have one.
• No certification, no external penetration test. The security checks are internal and documented. There is no seal of approval.
• No visibility guarantee. Clean technology, good content, and unambiguous data improve the conditions. Google, ChatGPT, and the others decide for themselves what they display and whom they name.
• No general legal guarantee. The website can be operated in a data-efficient manner and without advertising trackers. Legal texts, consent forms, image rights, accessibility, and industry-specific obligations must be reviewed for your business.
• No unlimited special development. Anything beyond the core functionality is evaluated based on its objective, effort, risk, and long-term maintainability. Not every technically feasible idea makes business sense – and we say so.

A prepared development stage is not a finished function. "Prepared" means that the architecture supports a later implementation – not that the shop, booking system, or customer area is already ready.

What that means in the workplace

The benefit does not arise from individual functions, but from how they interact:

• One treatment, all treatments done right. Opening hours, services, job openings: enter them once, and it's correct on the website, in the machine-readable data, and in every language. No more "it says something different on the website".
• They are found in humans and in machines. The same content appears in Google search results and in the answers of AI assistants because it is presented in the appropriate format for both.
• You'll see if it works. Visitors, real search terms, AI accesses and mentions in AI answers – all in one place, with honest numbers instead of embellished ones.
• Data protection as an argument, not a burden. No cookie banners, no third-party trackers, with our own server in the EU and a genuine data deletion policy. In the initial consultation, this is a plus, not a cost factor.
• Multilingual without any fuss. German, Italian and English are equally supported – right down to the user interface and error pages.
• Care that doesn't scare anyone away. Building blocks, previews, releases, previous versions – and, if desired, maintenance in conversation with an assistant instead of via the menu.
• Select, assess, change – all in one conversation. The assistant retrieves data from statistics, Google, and Bing, tells you what stands out, writes and translates the correction, and then saves or schedules it. No switching between four tools, no group email to the service provider. And none of it goes live without your approval.
• Less addiction. A price change, a new employee, a company holiday: you can do it yourself, in minutes, without a quote and without waiting on hold.
• It belongs to you. Content, images and data can be exported, the core system is not tied to any subscription, and there are no ongoing fees for access.

After over 20 years in web development, design, and digital consulting, I see the same pattern with SMEs all the time: The website is built as an isolated interface. A few years later, opening hours, services, translations, and contact details are scattered across four different systems, and nobody knows which information is correct. Then a customer calls because they've arrived to find the door locked.

Therefore, btlabs Core starts with a reliable data foundation and only then addresses the visible user interfaces. This is why data fields and deletion periods receive just as much attention as design in this reference.

How we decide on expansions

When an additional function is involved, I don't start with the technology, but with four questions:

• What business objective should be achieved?
• Which bottleneck will be reduced as a result?
• Who will maintain it after its introduction?
• Does the benefit outweigh the effort, risk, and ongoing maintenance?

Only then do we decide whether an existing function is sufficient, an interface makes sense, or whether something truly needs to be built from scratch. This sequence is part of our strategic website development.

Want to know where your current website stands before you think about a new one? The AI-Ready Check looks at the structure, speed, multilingualism, and machine-readable information of your existing site – and tells you exactly which of the points described here are already fulfilled and which are missing.

FAQs

Are all 207 functions really included as standard?

Yes, they are part of the core functionality. Some features are disabled by default and can be activated intentionally – the content check and citation monitor, for example. Others require your company to contribute something initially, such as connecting to Google Search Console or setting up your own AI access. The service description in the offer is always binding.

Does my website need a cookie banner?

The way we deliver them, they don't set any Cookies For visitors, it doesn't load anything from external servers and contains no advertising tracker – therefore, it doesn't require a consent window under normal operation. This is our reasoned position, not legal advice; the responsibility remains with you. As soon as you integrate a video, a booking form, a rating widget, or an advertising pixel, the situation changes, and you need a consent solution. This is not included in btlabs Core.

Can I really maintain my website via chat?

Yes – and more than most people expect. The assistant reads your analytics (visitors, search data from Google and Bing, page metrics, AI access), tells you what stands out, and then implements the changes itself: writing, correcting, translating, scheduling, and publishing. You grant it permissions individually; every access is logged, and it never gets your login credentials. An important note: Deleting via this interface bypasses the recycle bin. Therefore, the deletion right for external access is disabled by default.

How do I know if AI systems are even familiar with my business?

In three ways. First, you can see which AI systems are reading your content, how often, and which pages – with an authenticity check to prevent unauthorized access from being flagged as successful. Second, visitors who come to you via an AI response are identified. Third, a citation monitor can be set up: It regularly submits your own query questions to several major language models and analyzes whether your business is mentioned and, if so, how well. This isn't enabled by default; it's set up upon request and incurs costs via your AI access each time it runs. And it remains a sample – not a ranking.

Will I receive a newsletter?

No, and we'd rather say that upfront than afterward. Included is the clean setup and maintenance of your mailing list: registration with double consent, documented history, one-click unsubscribe, self-disclosure and self-deletion, and automatic cleanup of unconfirmed registrations. Sending an issue to all subscribers is not part of the core package – we integrate a mailing tool for that.

Can we take over our existing website?

Content, images, and the page structure can be migrated, and the old addresses will be permanently redirected to the new pages – ensuring your established search engine ranking is maintained. The extent of this automation depends on the old system and data quality. The migration is a separate project and will be discussed beforehand.

How much technical knowledge does my team need?

For everyday use, basic office experience is sufficient. Pages are created from pre-built modules, you can preview changes before publishing, and you manage your company data in one central location. An onboarding session is included. However, for changes to the basic design or for adding new modules, you will still need our support.

How many building blocks will I receive?

Twenty are included as standard – every new business gets that many. Our own website uses 43, 26 of which are custom-built. These are project-based and not automatically part of a new project. We explicitly state this because screenshots of our site would otherwise promise more than is included in the standard package.

Are German, Italian, and English always included?

The technical side is equivalent, right down to the user interface. Translation, editing, and adapting the content to the respective market are editorial services and are agreed upon within the project. Adding a fourth language isn't a simple click, but a guided transition.

How fast is the website really?

Our own site responds in about a tenth of a second, even when multiple visitors are accessing it simultaneously – with the previous technology, it took roughly five times as long. This is our measurement on our machine, not a guaranteed performance, and what your visitor ultimately experiences also depends on their internet connection, their device, and the embedded content. The most honest way to see for yourself is to try it: Visit our site and compare it to yours.

What happens when the collaboration ends?

Responsibilities, access rights, and available data export options are documented and transferred to ensure a smooth transition. Your content, images, and data are exportable and remain the property of your organization. Someone will then need to take over operation, maintenance, and further development – ​​this is an organizational matter, not a technical limitation.

How secure is my data against data loss or ransomware?

Backups are performed nightly and are encrypted. Backups from the last thirty days are protected against deletion – even by someone who has gained unauthorized access. In addition, there are monthly archives, an image archive, and a daily backup stored off-site. Our targets are: back online within two hours and a maximum of one day of data loss. Several emergency scenarios have been tested using a manual. We deliberately do not specify a contractually guaranteed availability percentage.

Is the website accessible?

Jump links, focus markers, and a static display for users who are sensitive to movement are integrated, and an automated check against the WCAG 2.1 AA standard runs daily. This means it's "automatically checked," not "compliant"—an automated test only covers the machine-checkable portion. With the new regulations in place, this distinction is important, and a detailed evaluation should be part of the project.

Does btlabs Core require WordPress extensions?

No. btlabs Core isn't built as a stack of third-party extensions. What you need is integrated into the architecture in a controlled manner, instead of piecing together many interdependent modules. This reduces update conflicts and security risks – and is the reason why the site doesn't accumulate unnecessary bloat.

Can btlabs Core be connected to a CRM or ERP system?

The foundation is there: A specialized system can retrieve requests, forms, and content. The ability for the website to automatically send data over is not yet built in and therefore requires further development – ​​data fields, permissions, and processes vary too much from system to system for a standard solution.

Can the range of functions change?

Yes, and he does it continuously. This reference is a snapshot; since the last major review, several enhancements have been added, including the registration of AI assistants via consent. We record changes with the date.

For which businesses is this worthwhile – and for which is it not?

It's worthwhile if speed, discoverability by people and AI, data sovereignty, or multilingualism are important to you: owner-managed service providers, craft businesses, retailers, and tourism companies with genuine company data, multiple languages, or multiple channels. It's not worthwhile if you need a simple three-page business card that will never be multilingual, never particularly discoverable, and never grow in size. For that, a basic provider will suffice – and we'll tell you that straight up.

As of September 2026. This reference describes the core of btlabs Core and its functions . The decisive factors for a project are the goal, the agreed scope of work, and the documented responsibilities.