# AI Automation in Practice: How We Use Make.com to Automate Our News Publishing for TutKit.com > URL: https://4eck-media.de/en/blog/ai-automation-in-practice-how-we-use-make-com-to-automate-our-news-publishing-for-tutkit-com/ > Language: en > Description: There is currently a lot of talk about AI automation. In many articles it sounds as if you only need to connect a few tools and the business practically runs itself. The reality i… --- There is currently a lot of talk about AI automation. In many articles it sounds as if you only need to connect a few tools and the business practically runs itself. The reality is different. Automation can save an enormous amount of time, but only if it is properly planned, connected via the right interfaces, and equipped with suitable control points. That’s exactly why we’re starting this blog series: **AI automation in businesses**. We don’t want to explain in theory what might be possible someday. We show what we actually use ourselves. We build automations for our own portals and websites, test them under real conditions, continuously improve them, and then transfer these experiences to client projects. A good example is our new news automation for [TutKit.com](https://www.tutkit.com/). TutKit is our e-learning portal for creatives, freelancers, businesses, trainers, and anyone working with digital content. There, we want to regularly publish current news, especially on topics like AI, design, marketing, software, creative workflows, online business, and digital tools. The problem: Regular news is valuable, but it costs time. A lot of time. ![Illustration des automatisierten News-Workflows: KI-Recherche, Texterstellung, Bildgenerierung und Veröffentlichung auf Website und Social Media](https://4eck-media.de/wp-content/uploads/2026/07/ki-automatisierung-workflow-news-veroeffentlichung-make.avif "KI-Automatisierung in der Praxis – Automatisierter News-Workflow mit Make.com") We need to find topics, check sources, assess relevance, write an article, create a featured image, publish the article, write social media posts, post on Facebook and LinkedIn, and finally make sure nothing gets published twice. For individual articles, that’s manageable. For several news items a day, it quickly becomes a burden. This is exactly where our Make.com automation comes in. ## The problem: News costs time, especially when it needs to appear regularly Many companies know this: the blog should become more active. The website should feel alive. Social media should be updated regularly. Google should see that new content is being created. Customers should notice that the company stays up to date and that the news section doesn’t look outdated with old posts. The desire is clear. Implementation often fails due to everyday workload and the resources available. An editor or marketing employee would have to complete several steps every day: 1. Find relevant news 2. Check whether the topic fits the target audience 3. Capture the source and context 4. Write the article 5. Create or find an image 6. Create the post in the CMS 7. Write social media texts 8. Post on Facebook and LinkedIn 9. Document the publication 10. Prevent a topic from appearing twice Especially the last point sounds trivial, but it’s important in practice. As soon as several topics are researched and prepared in parallel, you need a clean status. What is just a suggestion? What has been approved? What has already been published? What was discarded? What is still waiting for an image? Without structure, automation quickly takes on a life of its own. A system might then post something twice, publish content without approval, or pick up a topic that is current but doesn’t fit the brand. Our goal was therefore not: „AI should simply publish everything.“ Our goal was: **AI should take over the time-consuming steps, but editorial control stays with us.** From our point of view, that’s an important difference. ## Our solution: Two Make.com scenarios with Google Sheets as the control center We deliberately split the automation into two scenarios. The first scenario handles the research. It searches for new topics twice a day and stores suitable news suggestions in a Google Sheet. The second scenario handles the publishing. It checks the sheet several times a day to see whether we have approved any posts. For us, approval simply means: a “1” is entered in the corresponding column. This gives us a very simple but effective approval process. No complicated editorial system. No new tool. Just a Google Sheet that every employee understands immediately. The sheet is the heart of the automation. It contains, for example: | Column | Meaning | | --- | --- | | Research date | When was the news item found? | | Topic/Title | What is it about? | | Short description/Source | Summary and source reference | | Approval | Marked with a “1” if it may be published | | Publication date | When was it published? | | Image URL/Status | Status of the featured image | | Status | New, approved, published, or faulty | The advantage: we can decide on topics quickly. We can see at a glance what the AI has found. We don’t have to research everything manually, but we keep control over what gets published. ## Scenario 1: The automated news research The first Make scenario runs twice a day. Each time, it searches for ten potential news topics that could be relevant for TutKit. ![Make-Szenario zur KI-News-Recherche mit AI Web Search, Iterator und Google Sheets](https://4eck-media.de/wp-content/uploads/2026/07/ki-automatisierung-make-szenario-news-recherche.avif "KI-Automatisierung: Make-Szenario für die automatisierte News-Recherche") This isn’t about collecting just any AI news. That would be too imprecise. The topical fit with our users is essential for us. For TutKit, topics that concern creatives, designers, marketers, content teams, trainers, freelancers, and digital businesses are especially interesting. Typical topic areas include, for example: - AI tools for design, text, video, and marketing - New features in creative software - Developments at OpenAI, Anthropic/Claude, Adobe, Canva, Figma, Google, and similar providers - Trends around automation, online marketing, and e-learning - Relevant changes to platforms, workflows, and digital business models So the scenario doesn’t just search for “any old news,” but for news that can later offer real value to our readers too. The process roughly looks like this: 1. Make starts the scenario automatically. 2. The web search, or AI research, looks for suitable news. 3. The results are structured and prepared. 4. Each news item is given a title, a short description, and a source. 5. The data is written to a Google Sheet. 6. The status is initially set to “New.” ![Einstellungen des Google-Sheets-Moduls im Make-Szenario zur Speicherung der KI-Rechercheergebnisse](https://4eck-media.de/wp-content/uploads/2026/07/ki-prozesse-google-sheets-modul-einstellungen-make.avif "KI-Prozesse in Make: Google-Sheets-Modul für die News-Recherche") This creates a fresh pool of possible topics twice a day. We no longer have to scroll through multiple sources ourselves, search newsletters, or manually collect trends. The research is prepared for us. The decision stays with us. This is a very important point with automations like this: the AI provides suggestions. The human decides what fits the brand. ## Why the Google Sheet works so well ![Google-Tabelle als Redaktionsplan mit KI-recherchierten News-Themen und Freigabespalte](https://4eck-media.de/wp-content/uploads/2026/07/ki-news-redaktionsplan-google-tabelle-freigabe.avif "KI-News-Redaktionsplan in Google Sheets mit Freigabe-Workflow") You could also build a full-fledged editorial system. You could develop your own dashboard. You could program an approval interface with roles, permissions, and status logic. For the start, however, that would have been over-engineered. Google Sheets has several advantages for this purpose: it’s set up quickly. Everyone on the team understands it immediately. It connects easily with Make.com. Every row is a potential post. Every column has a clear purpose. The status is transparent. Changes are visible immediately. And above all: approval is extremely simple. A “1” in the approval column is enough. That sounds almost too simple, but it’s exactly why it works in everyday use. Good automation doesn’t have to look complicated. It has to reliably do what it’s supposed to do. For us, that means: - News suggestions come in automatically. - We review the topics. - We set a “1” for suitable topics. - Publishing then runs automatically. - The status is then updated. This creates a lean editorial process that doesn’t create any additional bureaucracy for our team. ## Scenario 2: An approved news item becomes a published post The second Make scenario is the actual publishing process. ![Make-Szenario für die automatisierte News-Veröffentlichung mit ChatGPT, KI-Bildgenerierung, API-Anbindung, Facebook und LinkedIn](https://4eck-media.de/wp-content/uploads/2026/07/ki-automatisierung-make-szenario-news-veroeffentlichung.avif "KI-Automatisierung: Make-Szenario für die automatisierte News-Veröffentlichung") It checks several times a day whether there are approved news items in the Google Sheet. Importantly: not all approved news items are published immediately, one after another. The scenario is set up to check eight times a day for new approvals. The advantage: if five news items are approved in the morning, they don’t all appear at once. Instead, they are spread out over the course of the day. To visitors, it looks as if an editor is regularly publishing new content. That’s exactly what matters for a portal like TutKit. The site feels alive. New content doesn’t appear in blocks, but in a natural rhythm. The process of the second scenario looks like this: 1. Make checks the Google Sheet. 2. It looks for rows where approval has been set. 3. It checks whether the status is not yet “published.” 4. The news information is passed to the AI. 5. The AI creates a news article of about 400 words from it. 6. A suitable featured image is then created using OpenAI DALL-E. 7. The finished post is published on TutKit.com via API. 8. Short social media posts are then created. 9. These are published on Facebook and LinkedIn. 10. The correct row in the Google Sheet is updated. 11. The status is set to “published.” This completes the entire process. What used to be many individual manual steps now runs as a continuous automation pipeline. ## The most important safeguard: no publishing without approval We could have built the automation so that every news item found gets published automatically. Technically, that would be possible. Content-wise, though, it wouldn’t be a good idea. Especially with news, there are several risks: a source can be inaccurate. A topic can be irrelevant to the target audience. A story can be too technical. A development can still be unclear. A topic may already have been published. Or the news simply doesn’t fit the desired tone of the portal. That’s why the approval step is so important. We don’t use AI as an uncontrolled autopilot, but as a strong assistant. The AI researches, structures, writes, formulates social posts, and triggers publications. The editorial decision stays with us. This is also the approach we recommend for client projects. Automation should be applied where it reliably reduces workload. Humans should stay involved where judgment, responsibility, tone, and brand understanding are required. It’s exactly this combination that makes AI automation so valuable in a business. ## Why the status in the sheet is so important One small detail is especially important in workflows like this: at the end, it must be saved in the Google Sheet that the article has been published. That sounds like a side step. In fact, it’s one of the most important points of the entire automation. Without a clean status, the scenario could find the same row again on the next run. The same news item would then be written again, published again, and posted to social media again. That obviously must not happen. That’s why, after successful publication, the exact row is updated. The status is changed. The publication date can be recorded. If needed, the URL to the finished post can also be entered. This way the system knows: this news item is done. For us, this is a good example that automation isn’t just made up of AI prompts. The real value comes from clean process logic. What data comes in? What conditions apply? When is something executed? What happens on success? What happens on error? How do we prevent duplicate actions? Only once these questions are resolved does a gimmick become a productive workflow. ## The role of AI in writing the news The AI doesn’t just get a keyword. It receives structured information from the sheet: topic, short description, source, context, and, if applicable, further notes. From this, it creates a post of around 400 words. For news, this length makes sense. The post should be short enough to be read quickly, but long enough to provide context, meaning, and classification. The prompt matters a lot here. The quality of the output depends heavily on how clearly the instructions are formulated. ![KI-Prompt im Make-Modul AI Web Search für die automatisierte Erstellung redaktioneller News-Beiträge](https://4eck-media.de/wp-content/uploads/2026/07/ki-prompt-automatisierte-news-erstellung-make.avif "KI-Prompt für die automatisierte News-Erstellung in Make") The prompt must, for example, define: - How long the post should be - Which target audience is addressed - What tone is desired - Whether the post should sound factual, casual, or advisory - Which terms should be avoided - How closely the source may be drawn on - That no invented facts should be added - That the benefit for TutKit readers should be highlighted This last point matters especially to us. A news item isn’t automatically relevant just because it’s new. It has to be put into context for readers. When Adobe releases a new feature, we’re not just interested in the fact that this feature exists. We’re interested in: What does this mean for designers, photographers, content teams, or trainers? Does it save time? Does it change workflows? Does it make a tool more attractive? Do creatives need to adjust to it? This classification turns a simple announcement into a useful post for us and our readers. ## Automatically creating featured images with DALL-E ![OpenAI-Modul in Make mit Prompt zur KI-Bildgenerierung moderner Beitragsbilder](https://4eck-media.de/wp-content/uploads/2026/07/ki-bildgenerierung-beitragsbild-openai-prompt.avif "KI-Bildgenerierung: OpenAI-Prompt für automatische Beitragsbilder") After the text comes the image. For many editorial teams, this is an underestimated time factor. Writing a short post is one thing. Finding or designing a suitable featured image is often almost just as time-consuming. You have to consider image rights, find a suitable visualization, follow format requirements, possibly use design templates, and make sure the image fits the style of the portal. In our workflow, the featured image is created automatically with OpenAI DALL-E. Here too, the prompt matters. The image should fit the topic without looking cluttered. It should match the news without being misleading. It should work as a featured image and look good on the portal. Especially for AI, software, and workflow topics, abstract or illustrative images are often more suitable than classic stock photos. Automatically generated images can work very well here if the style is clearly defined. Here too: automation takes work off our hands. At the same time, we can continue to refine the process later, for example with fixed image styles, format specifications, or different prompts depending on the category. ![Einstellungen der KI-Bildgenerierung in Make: Qualität Low, WEBP-Format und Kompression 70 für optimierte Beitragsbilder](https://4eck-media.de/wp-content/uploads/2026/07/ki-bildgenerierung-webp-qualitaet-einstellungen.avif "KI-Bildgenerierung: Qualitäts- und WEBP-Einstellungen in Make") Important: so that the images don’t cost too many tokens, we deliberately chose Low quality, since this is completely sufficient for featured images. To keep the final image from becoming too large, we chose WEBP with 70 compression as the format. ## Publishing via API to TutKit.com The next step is publishing on TutKit.com. Make.com uses an API interface for this. The finished post isn’t copied and pasted manually. Instead, title, text, featured image, category, status, and other data are automatically transferred to TutKit. That’s a big advantage over semi-automated solutions. Companies often save time writing, but then lose that time again entering everything into the CMS. A truly good workflow doesn’t end with a text document. It delivers the content to where it’s actually needed. In our case, that means: the post lands directly on [TutKit.com](https://www.tutkit.com/). ![Übersicht der per KI-Automatisierung veröffentlichten News-Beiträge auf TutKit.com](https://4eck-media.de/wp-content/uploads/2026/07/ki-automatisierung-veroeffentlichte-news-uebersicht-tutkit.avif "KI-Automatisierung: Automatisch veröffentlichte News-Beiträge auf TutKit.com") Depending on the system, different publishing logics can be used. A post can go live immediately. It can be created as a draft. It can be assigned to a category. It can be given tags. It can be linked to an image. It can even be scheduled for a later time. ![Automatisch per KI-Workflow und API veröffentlichter News-Beitrag auf TutKit.com](https://4eck-media.de/wp-content/uploads/2026/07/ki-automatisierung-news-beitrag-tutkit-cms.avif "KI-Automatisierung: Per API veröffentlichter News-Beitrag auf TutKit.com") For client projects, this part is often especially interesting. That’s because almost every company works with different systems. Some use WordPress, others Shopify, WooCommerce, PrestaShop, TYPO3, HubSpot, or their own systems. What matters isn’t the name of the system, but the question: is there an interface? And if so, what can be automated through it? ## Automatic social media publishing on Facebook and LinkedIn ![Automatisch erstellter Facebook-Post als Teil der KI-automatisierten News-Veröffentlichung](https://4eck-media.de/wp-content/uploads/2026/07/ki-automatisierung-social-media-post-facebook.avif "KI-Automatisierung: Automatischer Social-Media-Post auf Facebook") After the article, the process isn’t finished yet. A new post also needs to be seen. That’s why the scenario creates short social media texts and publishes them on Facebook and LinkedIn. Here too: the social post isn’t just a copy of the article. It has to be short, capture the benefit, and drive clicks to the post. The automation can, for example, generate the following: - A short intro - A compact summary - A note on why the topic is relevant - The link to the post on TutKit.com - Optionally matching hashtags This automatically extends the reach of the post. The article doesn’t just live on the blog, but is also distributed directly through additional channels. This step is enormously valuable, especially for businesses. Many publish blog posts but forget about distribution. Or they plan to share the post on social media later. Later, often nothing happens. Automation ensures that this step is carried out reliably. ## Why we spread the posts throughout the day One especially nice effect of the automation is the publishing rhythm. If we approve several news items in the morning, they shouldn’t all appear right after one another. That would look unnatural. Visitors would feel like everything was dumped out at once, and then nothing happens for the rest of the day. That’s why our scenario checks the sheet several times a day. This way, approved posts can be published spread out over the day. This has several advantages: - The website feels more active. Visitors regularly see new content. Social media channels aren’t flooded with several posts at once. Interactions are spread out better. The editorial team doesn’t have to sit at the computer at fixed times. - For users, it feels as if someone on the team is preparing a new news item every hour. In reality, the editorial team only approved the topics beforehand. Automation takes care of the rest. This is exactly the kind of relief that becomes noticeable within a company. ## The result: more output, less routine work, better predictability What does an automation like this actually deliver? For us, the biggest advantage is the combination of speed and control. We regularly get new topic suggestions. We no longer have to start from zero. We quickly decide which topics are interesting. The AI turns them into posts. The image is created automatically. Publishing runs via the API. Social media is factored in. The status is documented cleanly. As a result, we don’t just save time writing. We save time at many small points that add up to a lot. These small work steps are often underestimated in everyday life. Copy once. Upload an image once. Insert a title once. Write a social post once. Set a link once. Note the status once. Each of these sounds like just a few minutes. But with several posts per week or per day, this adds up to considerable effort. Automation helps us publish more evenly. It makes the process more predictable. It reduces manual errors. And it gives us more time for the tasks where real strategy is needed. ## Why clients benefit from our own experience We don’t build automations like this purely in theory. We use them ourselves. For clients, that makes a big difference. When we support businesses with AI automations, we don’t just bring tool knowledge. We bring real project experience. We know where workflows can fail. We know why status fields matter. We know why approval is often more sensible than full automation. We know how important clean prompts are. We know that interfaces don’t always work the way you initially imagine. And we also know: automation is never just a technical topic. It’s always a process topic too. Before connecting Make.com, OpenAI, Google Sheets, Facebook, LinkedIn, or a CMS, you need to understand the workflow: What is the starting point? Who decides? What data is needed? What errors can occur? What happens if an API doesn’t respond? What happens if an image isn’t created? What happens if a topic shouldn’t be published after all? How does the system recognize that something is done? Many companies start with the tool. We prefer to start with the process. Only once the process is clear do we build the right automation. ## What businesses can learn from this Our example of news publishing shows well how AI automation can be used sensibly. It’s not about replacing people. It’s about reducing routine work. People should assess, classify, decide, prioritize, and lead the brand. AI and automation should research, structure, prepare, formulate, transfer, and document. That’s a good mix. For businesses, this means: you don’t have to automate all of marketing right away. Often, a clearly defined process that takes a lot of time and recurs regularly is enough. That’s exactly where it’s worth starting. A good first automation process usually has these characteristics: - It repeats regularly. - It consists of several small manual steps. - It uses data from various sources. - It needs clear rules. - It can be safeguarded with an approval step. - It saves noticeable time when it runs reliably. Our news pipeline meets exactly these points. ## Further conceivable use cases for AI automation The logic behind our news automation can be applied to many other areas of a business. A few examples: **1. Automated blog preparation** Research topics, create outlines, gather sources, suggest internal links, and prepare drafts in the CMS. **2. Social media planning** Automatically turn new blog posts, products, or references into short posts for LinkedIn, Facebook, Instagram, or Google Business Profile. **3. Newsletter creation** Gather new content from the blog, shop, or knowledge base and automatically create a newsletter draft from it. **4. Recruiting communication** Create job ads from internal requirements, publish them on career pages, and prepare matching social posts. **5. E-commerce content** Enrich product data from spreadsheets, PIM systems, or shop systems, and use it to create SEO texts, short descriptions, or FAQ blocks. **6. Customer service** Detect recurring customer questions, generate suggested replies, and take the load off support staff. **7. Project management** Automatically turn emails, form inquiries, or meeting notes into tasks and document them in the project management system. **8. Reporting** Combine data from analytics, SEO tools, ad accounts, or shops and automatically create easy-to-understand management summaries. **9. Local marketing** Automatically turn new promotions, events, or offers into website announcements, Google posts, and social media posts. **10. Internal knowledge base** Structure, summarize, and make documents, records, and instructions findable for employees. The technical toolkit can be similar: Make.com, Google Sheets, OpenAI, APIs, MCP servers, webhooks, databases, CMS interfaces, social media integrations, and approval processes. The real difference lies in the process. ## Why Make.com is so well suited for this Make.com is particularly well suited for automations like this because many services are already connected and can be linked visually. You can see the workflow. You can identify which modules run one after another. You can set conditions. You can transform data. You can build different paths. You can call APIs. You can work with Google Sheets, OpenAI, HTTP modules, social media platforms, and many other services. This is attractive for businesses because many workflows can be implemented faster this way than with a fully custom development. Still, Make.com isn’t a set-and-forget solution. You need to know how to build robust scenarios. This includes, for example: - Clean filters - Clear status fields - Error handling - API tests - Data validation - Protection against duplicates - Logging - Sensible schedules - Clear prompts - Controlled approvals This is especially important for production processes. A demo is built quickly. A stable workflow for everyday use requires more care. ## What we pay attention to when building automations like this From our experience, there are a few points that are almost always important. - **First: Don’t automate too much at once.** A clear starting point is better. One process. One goal. It can be expanded afterward. - **Second: Plan for human approval when content goes public.** Especially for blogs, news, social media, and newsletters, not everything should go live unchecked. - **Third: Structure data cleanly.** A good spreadsheet or database is often more important than the fanciest AI prompt. - **Fourth: Take status logic seriously.** New, approved, published, faulty, discarded. States like these prevent chaos. - **Fifth: Treat prompts like work instructions.** A prompt isn’t just a sentence. It’s part of the process and must be formulated precisely. - **Sixth: Think through error cases.** What happens if an API is unreachable? What happens if an image isn’t created? What happens if a social post fails? - **Seventh: Start small, then improve.** An automation gets better with real data. Only in everyday use do you see which special cases come up. ## Our conclusion: AI automation becomes powerful when it takes on real work Our news automation for TutKit.com shows very well where things are heading. AI automation is not an abstract topic of the future. It can take on concrete work today. Not at some point. Not only at large corporations. But directly within everyday business processes. For us, this means: regular news output, less manual routine work, better predictability, automatic image generation, API publishing, social media distribution, and clean documentation in a Google Sheet. At the same time, editorial control is preserved. We decide which topics get approved. Automation then takes care of the rest. At 4eck Media, we build exactly these kinds of solutions for businesses that want to make their processes more efficient. These can be marketing workflows, content processes, recruiting procedures, e-commerce tasks, internal reports, or recurring administrative processes. Our advantage: we don’t advise from theory. We test and use systems like these ourselves. What works for us, we can transfer to client processes with the right concept. Anyone who regularly loses time on recurring tasks should check whether an AI workflow could be built for it. Often, the greatest potential lies not in one big overhaul, but in a small process that eats up time every day. That’s exactly where good automation begins. ## FAQ: AI automation with Make.com, OpenAI, and Google Sheets Can AI really write complete news articles? Yes, if the input data is good and the prompt is clearly formulated. But control is important. We don’t let the AI publish topics arbitrarily. The research is prepared, the approval is done by us, and only then does the AI write the post. Why do you use Google Sheets as the approval hub? Because it’s simple, fast, and transparent. Everyone understands a spreadsheet. You can see the status of every news item immediately. Approval can be set with a single “1”. For many workflows, this simplicity is a big advantage. Why don't you publish every news item found automatically? Because not every news item found is relevant. Some topics don’t fit the target audience. Some sources are too thin. Some reports are too speculative. That’s why human approval remains important. How do you prevent duplicate publications? After publication, the corresponding row in the Google Sheet is updated. The status is set to “published”. The scenario checks this status before every new run. This prevents the same news item from being processed again. Why are the news items spread out over the day? If several posts appear at the same time, it looks unnatural and can overload social media channels. By checking several times a day, content appears in a regular rhythm. For visitors, this feels more alive. Can a workflow like this also be built for WordPress? Yes. The basic idea is transferable. Instead of TutKit.com, WordPress, WooCommerce, Shopify, TYPO3, or another system could be connected, provided a suitable interface is available. Is Make.com strictly necessary for this? No. Similar workflows can also be built with other automation tools or custom development. However, Make.com is a good starting point for many businesses because many services can be connected visually. Is a workflow like this also suitable for small businesses? Yes, especially so. Small teams often have little time for regular content work. A well-built automation can take a lot of the load off here, without needing to build a large editorial team right away. What does an automation like this cost? That depends on the scope. What matters is how many systems are connected, how complex the approval logic is, whether a CMS needs to be integrated via API, and how robust the workflow needs to be. It’s often worth starting with a small pilot process. Can 4eck Media build automations like this for clients? Yes. That’s exactly what this blog series is about. We develop AI automations for our own projects and transfer our experience to client processes. Anyone who wants to automate recurring tasks can work with us to find out where the greatest leverage lies.