{ "version": "https:\/\/jsonfeed.org\/version\/1.1", "title": "Artöm Mazurchak: posts tagged systems", "_rss_description": "I live in Berlin. I built Biz-cen.ru in Russia, Lashoestring.com in the UK. I run a Telegram channel", "_rss_language": "en", "_itunes_email": "", "_itunes_categories_xml": "", "_itunes_image": "", "_itunes_explicit": "", "home_page_url": "https:\/\/www.mazurchak.com\/?go=tags\/systems\/", "feed_url": "https:\/\/www.mazurchak.com\/?go=tags%2Fsystems%2Fjson%2F", "icon": "https:\/\/www.mazurchak.com\/pictures\/userpic\/userpic@2x.jpg?1700581632", "authors": [ { "name": "Artöm Mazurchak", "url": "https:\/\/www.mazurchak.com\/", "avatar": "https:\/\/www.mazurchak.com\/pictures\/userpic\/userpic@2x.jpg?1700581632" } ], "items": [ { "id": "96", "url": "https:\/\/www.mazurchak.com\/?go=all\/the-fourth-revolution-ai-how-do-we-use-past-lessons-why-cant-we\/", "title": "AI’s Fourth Industrial Revolution: Why Old Product Playbooks Fail", "content_html": "

We’re living through a unique moment: the fourth industrial revolution, under the AI flag. Humanity has already gone through three industrial revolutions, and the worries people feel now aren’t new. How should we think so we stop worrying and, using what we know about past shifts, turn this moment to our advantage?<\/p>\n

\n\"\"\n<\/div>\n

Sam Altman often says<\/a> in his talks: “we’ll have to rethink and change how organizations and society function.” Let’s unpack what that means.<\/p>\n

Ford’s assembly line: a template for change<\/h2>\n

Let’s look at a past example of how to act to seize the moment. Before Ford’s assembly line, one worker assembled the entire car, doing many different operations.<\/p>\n

With Ford’s moving line, the process was rebuilt: the line moved, and each worker did one or two operations — productivity increased eightfold<\/b>.<\/p>\n

But what did people think about first, before they started building Ford’s assembly line? A change like that required huge investments: rebuilding production from scratch, retraining people with new skills, and setting new management standards.<\/p>\n

They started with the idea of a new organization design scheme:<\/p>\n

    \n
  1. Redefining the operations in the plant;<\/li>\n
  2. Reorganizing communication between workers;<\/li>\n
  3. Setting new roles for each team member.<\/li>\n<\/ol>\n

    Changing the organization design scheme<\/h2>\n

    In the end, they defined a organization design scheme. From this comes an important point:<\/p>\n

    The core feature of any industrial revolution is a organization design scheme.<\/b><\/p>\n

    These schemes aren’t “written in the sky” or hidden in secret books. In every industrial revolution, a specific group of people define the new scheme. They set a new norm, and from that norm new products were conceived.<\/p>\n

    Like any cultural change, a new norm takes time to be accepted by society. The key takeaway: first you establish the scheme, and only then do you start thinking about products.<\/p>\n

    From this follows another point:
    \nfast product success = product failure.<\/b><\/p>\n

    A fast win usually means the product fit into the old organization design scheme. Great products take time to mature. But the companies that set the new norm gain an undeniable advantage for the next decades and, in the end, win the competition.<\/p>\n

    How we work today<\/h2>\n

    Here’s, schematically, how work typically runs in a product company today. Suppose John takes on a task. John might be a product manager, a designer, or a project manager — that detail doesn’t matter.<\/p>\n

    \n\"\"\n<\/div>\n

    When John starts, he tries to understand the task and collects context:<\/p>\n

      \n
    1. He understands the company’s domain to some extent and knows the constraints of his department and his specific role.<\/li>\n
    2. He gathers information about the external context — competitors and customers.<\/li>\n
    3. He talks to the internal team to understand the task better.<\/li>\n<\/ol>\n

      Eventually, John forms a working picture of the task’s context. It’s incomplete, but good enough to start. His time for gathering context is limited: the context remains incomplete, yet he needs to proceed.<\/p>\n

      How we’ll work tomorrow<\/h2>\n

      In the new scheme, John has an agent that:<\/p>\n

        \n
      1. Remembers all relevant context: company constraints and plans; the department’s work and ideas; information tied to his role.<\/li>\n
      2. Exists not only for John but also on the side of internal teams, customers, and competitors.<\/li>\n
      3. Changes how we communicate: from “human–human” toward protocol-like human–agent and agent–agent interactions.<\/li>\n<\/ol>\n
        \n\"\"\n<\/div>\n

        Let’s use this scheme to see how product work could evolve.<\/p>\n

        Product example<\/h2>\n

        To see what kinds of products emerge from the new scheme, focus on the part where John used to “collect context.” We interview people who build products, analyze their work, and how their activities are set up today.<\/p>\n

        \n\"\"\n<\/div>\n

        The analysis shows: up to 60% of their time goes to gathering task context — and even then the context stays incomplete. From here comes a product idea: build a domain AI assistant for a specific person, in a specific department, for a company in a specific domain<\/b>. The goal is to radically increase the speed and quality of decisions.<\/p>\n

        How the assistant works:<\/p>\n

          \n
        1. It has access to role-relevant sources (docs, task tracker, knowledge base).<\/li>\n
        2. It sees context gaps and points out what’s missing.<\/li>\n
        3. It finds relevant past discussions (e. g., a Slack thread from six months ago) and routes to the knowledge holders.<\/li>\n
        4. It takes into account practices of direct and indirect competitors.<\/li>\n
        5. It acts within company policies and constraints.<\/li>\n<\/ol>\n

          Result: a dramatic productivity boost — less time searching, fuller context, faster and better decisions.<\/p>\n

          The AI assistant is just an illustration of how to derive a product from the scheme<\/b>.<\/p>\n

          An orchestra of products from one scheme<\/h2>\n

          We explored only one spot in the scheme that led to one product. But the scheme reveals more places where opportunities for new products appear:<\/p>\n

            \n
          1. Agents on the side of customers and competitors;<\/li>\n
          2. Agents on the side of internal teams;<\/li>\n
          3. Products that enable human–agent and agent–agent communication.<\/li>\n<\/ol>\n
            \n\"\"\n<\/div>\n

            When a company starts from the scheme, it can create several interlinked solutions at once. The scheme lets you build an orchestra of products that amplify each other.<\/p>\n

            A few books that explain how to think systemically about this topic are Idealized Design<\/a> by Russell Ackoff and Georgy Shchedrovitsky’s Organizational-Managerial Thinking.<\/p>\n

            Takeaways:<\/b><\/p>\n

              \n
            1. We’re in the fourth industrial revolution, and we must build products differently than in the last 100 years.<\/li>\n
            2. First define a new organization design scheme — then think about products.<\/li>\n
            3. Fast product success = product failure: a fast win means the product plugged into the old organization design scheme.<\/li>\n
            4. A new organization design scheme lets you invent several mutually reinforcing products.<\/li>\n<\/ol>\n
              \n