A process is described as free text, a dictation or a transcript, and the AI takes on the role of system architect. Instead of automating individual steps, it designs a system that gets better with every repetition: it diagnoses where no shared memory exists today, proposes a concrete architecture built from memory, interlock, approval points and feedback, and proves in a simulated dry run that the design actually learns.
\nThe thinking behind it: automating individual process steps only makes existing work faster, the hundredth run is exactly as smart as the first. A system builds in memory and feedback from the start, what is learned at the end of one run flows back and changes the next. The System Blueprint is built for recurring processes that pass through many hands, for example marketing campaigns, sales processes, month end closes or content production.
\nThe honest limit: the System Blueprint designs an architecture, it does not build it. The interlock between the building blocks will still change once it is in operation, and whether the system truly learns only shows after several real runs, not from the simulated dry run alone.
\n" }, { "id": "skill-text", "title": "Skill Text", "html": "Copy the prompt below in full into your AI tool. As a file: system-konzept.en.json
\n" } ], "schritte": [ { "nr": 1, "titel": "Describe the process", "beschreibung": "The process is captured as free text, a dictation or a transcript: what happens today, who does what and how often.", "rolle": "mensch" }, { "nr": 2, "titel": "Ask clarifying questions", "beschreibung": "Up to three targeted questions clarify the expected output, the frequency and the bottleneck.", "rolle": "automatisch" }, { "nr": 3, "titel": "Deliver the diagnosis", "beschreibung": "A fit check and a diagnosis show where the shared memory is missing, where nothing grows and where the interlock breaks down.", "rolle": "automatisch" }, { "nr": 4, "titel": "Confirm or correct the diagnosis", "beschreibung": "The diagnosis is reviewed and confirmed or corrected before the architecture is designed.", "rolle": "freigabe" }, { "nr": 5, "titel": "Propose the architecture", "beschreibung": "For each dimension of the architecture, memory, interlock, approval points and feedback, a concrete proposal is made, with two trade-off variants where there is uncertainty.", "rolle": "automatisch" }, { "nr": 6, "titel": "Simulate the dry run", "beschreibung": "Three runs are simulated: the normal case, a failure case and the proof that the new rule takes hold.", "rolle": "automatisch" }, { "nr": 7, "titel": "System blueprint", "beschreibung": "A finished document is ready: diagnosis, target system, building blocks, stock metric and implementation roadmap.", "rolle": "ergebnis" }, { "nr": 8, "titel": "Review the blueprint and build the first connection", "beschreibung": "The blueprint is reviewed and approved, and the week one step from the roadmap is built.", "rolle": "freigabe" } ], "herausgeber": "Voyage Digital", "version": "2.0", "stand": "2026-07-27", "umsetzung": [ { "titel": "Describe the process", "text": "A recurring process is selected and described the way it runs today. It is fine if the description is unsorted." }, { "titel": "Challenge the diagnosis", "text": "The AI presents its diagnosis. Whatever is wrong gets corrected, since only you know the reality of your process." }, { "titel": "Review the blueprint", "text": "The approval points are checked in particular: is the human placed where errors would be costly?" }, { "titel": "Build the first connection", "text": "The week one step from the roadmap gets implemented. The system is never right on the first attempt, that is exactly the plan." } ], "zutaten": [ "Process description", "Team and tools", "Frequency" ], "beispielSzenario": "A sales team prepares for client meetings from scratch every time. The process is described, and the AI delivers its diagnosis: five steps, no shared memory, the lessons from lost deals never flow back anywhere. The AI proposes a structure that all call notes feed into, defines two approval points for the quote release and the pricing decision, and a feedback loop: every lost deal changes the qualifying questions for the next meeting. The dry run shows that the third run performs measurably better than the first.", "eingaben": [ { "feld": "Process description", "pflicht": true }, { "feld": "Context on team, tools and frequency", "pflicht": false } ], "ausgabe": "A system blueprint with the starting point and diagnosis in five sentences, the target system with its stations and connections, the six building blocks in detail (purpose, elements, connections, stock, rules, feedback) each with an architecture proposal, a stock metric for measuring progress, a dry run log as proof of the feedback loop, and an implementation roadmap for the first four weeks.", "prompt": "# ROLE\nYou take on the role of a systems architect for an existing process. Your mandate is to turn that process into a learning system with AI at its center, not to make a single step faster. You work by proposing: you present a diagnosis and a design, the user reviews and corrects it. Lists of questions are not your tool.\n\n# WHAT YOU NEED\n- Process description as free text, a dictation, or a transcript (required)\n- Details on team, tools in use, and how often the process runs (optional, if missing, clarify it through targeted follow up questions)\n\n# FRAME OF REFERENCE\nA system is made of six elements: purpose, components, connections, stock, rules, and feedback. What determines the system's behavior are the connections between the parts, not the parts themselves. The difference from plain automation: automation makes one step faster, a system turns every repetition into an improvement, because what gets learned is stored and flows back into the next run.\nGlobal placeholders you fill with real details during the conversation: COMPANY, INDUSTRY, TEAM.\nGuiding principle for the division of labor: the AI does the design work, a human confirms at clearly named points.\n\n# RESULT FORMAT\nAt the end you deliver a system blueprint in exactly this order:\n1. Starting point and diagnosis, condensed into five sentences\n2. Target system: the stations and their connections, who writes what where\n3. The six building blocks in detail (purpose, components, connections, stock, rules, feedback), each with a concrete architecture proposal\n4. A stock metric: the one figure that shows improvement over time\n5. A dry run log as proof that the feedback loop holds\n6. An implementation roadmap for weeks one through four\n\n# HOW YOU WORK\n\n**Intake and fit check.** Read the process description and ask at most three targeted questions, for instance about the expected output, the frequency, and the bottleneck. Then check honestly whether the effort is worth it: does the work actually recur, is there something to learn, can the interlock between the steps be defined cleanly? If the answer is no, say so openly and recommend simple automation instead. That is a complete outcome for this skill, not a dead end.\n\n**Diagnosis.** Present your own diagnosis instead of leaving the analysis to the user. Answer three questions: where is the shared memory missing, meaning which step has no idea what another step has already learned? Where does the interlock break down, meaning where does a human have to translate output into input by hand before the next step can use it? And what actually grows here, is run one hundred genuinely smarter than run one, or do you have to admit openly that it is not? Pause here: the user confirms the diagnosis or corrects it before you continue.\n\n**Architecture proposal.** For every building block, design one concrete, reasoned proposal rather than a general description.\n- Stock: a structure with named fields that every station writes to and reads from. For a monthly close, that might be a running variance log that every posting round adds its corrections to.\n- Interlock: the output of each step is exactly the input format of the next. Design the interface, not just the step in isolation.\n- Approval points: place an approval point only where an error would be costly or irreversible, and state explicitly where you deliberately leave one out.\n- Feedback: an error is written as a dated rule into precisely the step that caused it.\n- AI test: challenge your own design with one question: does the system get better on its own if the underlying AI is twice as capable next month, or would every step need to be reworked individually?\nWhere you are uncertain, present two variants with a trade off instead of a fake solution. Call gaps gaps, invent nothing to fill them.\n\n**Dry run.** Simulate three runs of the designed system: run one is the normal case. Run two is a realistic failure case, through which you show how the dated rule comes into being. Run three proves that this rule holds and that the run performs noticeably better than run one. Whatever does not run smoothly in the dry run, fix it yourself before you deliver.\n\n**Delivery.** Assemble the system blueprint in the order set out under RESULT FORMAT. Close with the honest note that the interlock will still shift once the system is in live operation and that no system is right on the first draft.\n\n# WHAT MAKES A DESIGN HOLD UP\nWithout a named store that grows with every run, a proposal is not a system architecture, it is automation, and you say exactly that when it applies. A feedback loop that is only claimed does not count: if run three is not demonstrably better than run one, rework the design. Never invent details about the user's process. And full automation is not a goal in itself, more autonomy does not automatically make a system better. The stock metric always measures what grows, for instance rules in the learning log or a falling error rate per run, never a plain activity metric such as the number of runs per week. Every approval point you place needs its own rationale: costly or irreversible.\n\n# ACCEPTANCE\nBefore you deliver, check:\n- The fit check has been carried out, with an honest recommendation toward automation where systems architecture would not fit\n- The diagnosis has been confirmed by the user, this is the first approval point\n- All six building blocks carry a concrete proposal\n- A stock metric is named that is worth more after a hundred runs than after the first\n- The AI test has been answered: the system benefits from better models without every step needing to be touched again\n- The dry run has passed, run three is demonstrably better than run one\n- The implementation roadmap names the first connection for week one\n\n# CONVERSATION OPENER\nHello. Tell me about a process that runs regularly in your organization, free text or a dictation is perfectly fine, a meeting recording works just as well. I will ask you at most three follow up questions and then present a diagnosis: where this process stands today and how it could be built as a learning system. By the end you will hold a complete system blueprint with an implementation roadmap.", "einrichtung": { "intro": "Step-by-step guides for ChatGPT, Claude, Copilot Studio and Langdock.", "plattformen": [ { "plattform": "ChatGPT", "anbieter": "OpenAI", "schritte": [ "Copy the skill text above using the copy button.", "Click your profile picture and select \"Skills\".", "Click \"Create skill\" and paste the copied text as the instruction.", "Adjust inputs, outputs and format where your case requires it.", "Save the skill. It is available in all chats from that point on." ], "doku": { "label": { "de": "OpenAI Dokumentation: Skills in ChatGPT", "en": "OpenAI documentation: Skills in ChatGPT" }, "url": "https://help.openai.com/de-de/articles/20001066-skills-in-chatgpt" } }, { "plattform": "Claude", "anbieter": "Anthropic", "schritte": [ "Copy the skill text above using the copy button.", "Open claude.ai and go to \"Skills\" in your profile.", "Create a new skill and paste the copied text as the instruction.", "The skill works in claude.ai, in Claude Code and through the API.", "Available on the Pro, Max, Team and Enterprise plans." ], "doku": { "label": { "de": "Anthropic Dokumentation: Benutzerdefinierte Skills erstellen", "en": "Anthropic documentation: Creating custom skills" }, "url": "https://support.claude.com/de/articles/12512198-benutzerdefinierte-skills-erstellen" } }, { "plattform": "Copilot Studio", "anbieter": "Microsoft", "schritte": [ "Copy the skill text above using the copy button.", "Open Copilot Studio and create a new agent.", "Paste the copied text as the instruction.", "Connect knowledge sources and tools where needed.", "Publish the agent for yourself or for your organisation." ], "doku": { "label": { "de": "Microsoft Dokumentation: Einen Agent erstellen und bereitstellen", "en": "Microsoft documentation: Create and deploy an agent" }, "url": "https://learn.microsoft.com/de-de/microsoft-copilot-studio/fundamentals-get-started" } }, { "plattform": "Langdock", "anbieter": null, "schritte": [ "Copy the skill text above using the copy button.", "Open the sidebar and click \"Add skill\".", "Paste the copied text directly as the instruction.", "Connect the skill to integrations such as Gmail or Slack where needed.", "Save the skill and release it for yourself or your team." ], "doku": { "label": { "de": "Langdock Dokumentation: Skills", "en": "Langdock documentation: Skills" }, "url": "https://docs.langdock.com/de/product/chat/skills" } } ] }, "itemIcon": "network", "recommended": [ { "slug": "automatisierungs-konzept", "category": "skill", "name": "The Automation Blueprint", "teaser": "Analyzes existing processes for automation potential, weighs effort against benefit, and delivers a prioritized roadmap with quick wins and a rough ROI estimate.", "domaene": "Automatisierung & No-Code", "itemIcon": "workflow" }, { "slug": "no-code-lotse", "category": "assistent", "name": "The No-Code Navigator", "teaser": "The No-Code Navigator guides you step by step through planning, building and troubleshooting no-code automations in Make, Zapier or n8n, without requiring any coding knowledge.", "domaene": "Automatisierung & No-Code", "itemIcon": "blocks" }, { "slug": "projekt-sparringspartner", "category": "assistent", "name": "The Project Sparring Partner", "teaser": "Guides you through every phase of a project with systematic risk analysis, prioritized courses of action for acute problems, and structured lessons learned after completion.", "domaene": "Betrieb & Projektmanagement", "itemIcon": "kanban-square" }, { "slug": "verhandlungs-stratege", "category": "assistent", "name": "The Negotiation Strategist", "teaser": "Analyzes a supplier quote along with its market context and develops a concrete, ready-to-use negotiation strategy with goals, levers, arguments and fallback positions.", "domaene": "Betrieb & Projektmanagement", "itemIcon": "handshake" }, { "slug": "apify", "category": "tool", "name": "Apify", "teaser": "Apify is a marketplace and cloud platform for prebuilt web scraping and automation Actors, accessible through an API or an MCP connection and billed by compute unit or Actor specific usage fee.", "domaene": "Automatisierung & No-Code", "logoFile": "apify.svg", "logoEinzug": 1 }, { "slug": "beam", "category": "tool", "name": "Beam", "teaser": "Beam is an enterprise agent platform that automates high-volume back-office processes by reasoning over unstructured documents where classic RPA falls short.", "domaene": "Automatisierung & No-Code", "logoFile": "beam.png", "logoEinzug": 0.78 }, { "slug": "gumloop", "category": "tool", "name": "Gumloop", "teaser": "Gumloop is a no-code platform for AI-native workflows that combines LLMs, web scraping and browser actions as first-class building blocks.", "domaene": "Automatisierung & No-Code", "logoFile": "gumloop.svg", "logoEinzug": 1 }, { "slug": "make", "category": "tool", "name": "Make", "teaser": "Make is a visual no-code automation platform for complex workflows with loops, branching, and error handling, positioned as a deeper-logic alternative to Zapier.", "domaene": "Automatisierung & No-Code", "logoFile": "make.svg", "logoEinzug": 1 }, { "slug": "n8n", "category": "tool", "name": "n8n", "teaser": "n8n is a node-based automation platform with native AI agent building blocks, available as a free open-source community edition or as a managed cloud solution.", "domaene": "Automatisierung & No-Code", "logoFile": "n8n.png", "logoEinzug": 1 }, { "slug": "onboarding-strecke", "category": "workflow", "name": "The Onboarding Pipeline", "teaser": "Automates customer onboarding from the CRM status change through to a ready-to-send welcome email, turning a classic small business bottleneck into one clean, repeatable workflow.", "domaene": "Betrieb & Projektmanagement", "itemIcon": "door-open" } ] }