{
  "revision": 2,
  "updatedAt": "2026-11-20T15:57:48.841Z",
  "project": {
    "id": "agents",
    "name": "Digital employees",
    "brand": "LIVE4K",
    "kind": "internal-project",
    "availability": "internal-only",
    "website": "https://agents.live4k.media/",
    "summary": "LIVE4K’s internal platform for agent teams, tasks, preparation and approvals. We use the experience to design controlled business processes and limited IT tools.",
    "what": "We explore how teams of AI agents can receive tasks, prepare materials and use clearly bounded tools. The platform is an internal LIVE4K project and a source of implementation experience.",
    "why": "Useful automation needs more than a model response. It needs a clear task, access limits, review points and a way to distinguish a proposal from an action that changes another system.",
    "who": [
      "Businesses exploring controlled AI automation",
      "Teams with repetitive research or preparation work",
      "IT owners defining tool access and approval responsibilities"
    ],
    "where": "Our internal environment. Any customer-facing implementation, access or deployment is a separately agreed project.",
    "when": "When you want to assess a bounded workflow for AI assistance and can define inputs, responsibilities and acceptance criteria.",
    "how": [
      "Identify one task, its inputs and its limits.",
      "Agree tools, permissions, outputs and human review.",
      "Test the proposed flow on an appropriate limited scope.",
      "Review quality, failure cases and the conditions for any next step."
    ],
    "differentiators": [
      "Internal use informs the design of real workflows.",
      "Agent work connects to our software, integrations and infrastructure experience."
    ],
    "benefits": [
      "Practical experience for designing an AI pilot",
      "Clearer separation of preparation, review and action",
      "A basis for defining tool permissions and operating responsibilities"
    ],
    "capabilities": [
      {
        "id": "capability-1",
        "name": "Agent teams and tasks",
        "description": "Organise work around defined tasks and cooperating agent roles.",
        "status": "owner-described",
        "evidenceIds": [
          "editorial-en"
        ]
      },
      {
        "id": "capability-2",
        "name": "Preparation of materials",
        "description": "Use agents to prepare information and proposals for further review.",
        "status": "owner-described",
        "evidenceIds": [
          "editorial-en"
        ]
      },
      {
        "id": "capability-3",
        "name": "Approvals",
        "description": "Place human approval at the steps that require it.",
        "status": "owner-described",
        "evidenceIds": [
          "editorial-en"
        ]
      },
      {
        "id": "capability-4",
        "name": "Bounded tools",
        "description": "Design narrowly scoped IT tools rather than unrestricted access to business systems.",
        "status": "owner-described",
        "evidenceIds": [
          "editorial-en"
        ]
      }
    ],
    "useCases": [
      {
        "audience": "Operations teams",
        "need": "Reduce repetitive preparation",
        "solution": "Assess a limited agent-assisted workflow with review before consequential steps.",
        "expectedBenefit": "A practical basis for deciding what to automate."
      },
      {
        "audience": "IT teams",
        "need": "Define safe tool access",
        "solution": "Design narrow interfaces and explicit approval boundaries.",
        "expectedBenefit": "A clearer operating model for a pilot."
      }
    ],
    "limitations": [
      "The described platform is internal; the public profile does not provide access or imply a ready-made subscription.",
      "New tools, customer deployments and integrations require separate design and validation.",
      "Outputs and external actions need the review and responsibility agreed for the use case."
    ],
    "tags": [
      "Digital employees",
      "ai-workforce",
      "customer-relationships",
      "infrastructure"
    ],
    "relatedProjectIds": [
      "software",
      "infrastructure",
      "remarketagent",
      "forge"
    ],
    "details": {
      "aliases": [
        "Digital employees",
        "AI agents for controlled business workflows"
      ],
      "domainIds": [
        "ai-workforce",
        "customer-relationships",
        "infrastructure"
      ],
      "positioning": {
        "category": "AI agents for controlled business workflows",
        "elevatorPitch": "LIVE4K’s internal platform for agent teams, tasks, preparation and approvals. We use the experience to design controlled business processes and limited IT tools.",
        "marketRole": "We explore how teams of AI agents can receive tasks, prepare materials and use clearly bounded tools. The platform is an internal LIVE4K project and a source of implementation experience.",
        "offeringBoundaries": [
          "The described platform is internal; the public profile does not provide access or imply a ready-made subscription.",
          "New tools, customer deployments and integrations require separate design and validation.",
          "Outputs and external actions need the review and responsibility agreed for the use case."
        ]
      },
      "readiness": {
        "commercialStatus": "internal-only",
        "documentedState": "We explore how teams of AI agents can receive tasks, prepare materials and use clearly bounded tools. The platform is an internal LIVE4K project and a source of implementation experience.",
        "notEstablished": [
          "The described platform is internal; the public profile does not provide access or imply a ready-made subscription.",
          "New tools, customer deployments and integrations require separate design and validation.",
          "Outputs and external actions need the review and responsibility agreed for the use case."
        ],
        "reviewedOn": "2026-11-20"
      },
      "providerRoles": [
        {
          "organization": "LIVE4K",
          "role": "Design and implementation within the agreed scope.",
          "scope": "We explore how teams of AI agents can receive tasks, prepare materials and use clearly bounded tools. The platform is an internal LIVE4K project and a source of implementation experience."
        }
      ],
      "audienceProfiles": [
        {
          "id": "audience-1",
          "segment": "Businesses exploring controlled AI automation",
          "decisionRoles": [
            "The responsible business or technical owner"
          ],
          "endUsers": [
            "Businesses exploring controlled AI automation"
          ],
          "pains": [
            "Useful automation needs more than a model response. It needs a clear task, access limits, review points and a way to distinguish a proposal from an action that changes another system."
          ],
          "triggers": [
            "When you want to assess a bounded workflow for AI assistance and can define inputs, responsibilities and acceptance criteria."
          ],
          "fitConditions": [
            "A clearly bounded process and authorised owner",
            "Appropriate sample inputs, review criteria and access restrictions"
          ]
        },
        {
          "id": "audience-2",
          "segment": "Teams with repetitive research or preparation work",
          "decisionRoles": [
            "The responsible business or technical owner"
          ],
          "endUsers": [
            "Teams with repetitive research or preparation work"
          ],
          "pains": [
            "Useful automation needs more than a model response. It needs a clear task, access limits, review points and a way to distinguish a proposal from an action that changes another system."
          ],
          "triggers": [
            "When you want to assess a bounded workflow for AI assistance and can define inputs, responsibilities and acceptance criteria."
          ],
          "fitConditions": [
            "A clearly bounded process and authorised owner",
            "Appropriate sample inputs, review criteria and access restrictions"
          ]
        },
        {
          "id": "audience-3",
          "segment": "IT owners defining tool access and approval responsibilities",
          "decisionRoles": [
            "The responsible business or technical owner"
          ],
          "endUsers": [
            "IT owners defining tool access and approval responsibilities"
          ],
          "pains": [
            "Useful automation needs more than a model response. It needs a clear task, access limits, review points and a way to distinguish a proposal from an action that changes another system."
          ],
          "triggers": [
            "When you want to assess a bounded workflow for AI assistance and can define inputs, responsibilities and acceptance criteria."
          ],
          "fitConditions": [
            "A clearly bounded process and authorised owner",
            "Appropriate sample inputs, review criteria and access restrictions"
          ]
        }
      ],
      "systemModel": {
        "principle": "We explore how teams of AI agents can receive tasks, prepare materials and use clearly bounded tools. The platform is an internal LIVE4K project and a source of implementation experience.",
        "components": [
          {
            "id": "component-1",
            "name": "Agent teams and tasks",
            "purpose": "Organise work around defined tasks and cooperating agent roles.",
            "evidenceIds": [
              "editorial-en"
            ]
          },
          {
            "id": "component-2",
            "name": "Preparation of materials",
            "purpose": "Use agents to prepare information and proposals for further review.",
            "evidenceIds": [
              "editorial-en"
            ]
          },
          {
            "id": "component-3",
            "name": "Approvals",
            "purpose": "Place human approval at the steps that require it.",
            "evidenceIds": [
              "editorial-en"
            ]
          },
          {
            "id": "component-4",
            "name": "Bounded tools",
            "purpose": "Design narrowly scoped IT tools rather than unrestricted access to business systems.",
            "evidenceIds": [
              "editorial-en"
            ]
          }
        ],
        "inputs": [
          {
            "id": "input-1",
            "name": "Project brief",
            "description": "A clearly bounded process and authorised owner",
            "required": true
          },
          {
            "id": "input-2",
            "name": "Requirements and responsibilities",
            "description": "Appropriate sample inputs, review criteria and access restrictions",
            "required": true
          }
        ],
        "outputs": [
          {
            "id": "output-1",
            "name": "Agreed outcome",
            "description": "A workflow proposal or separately agreed pilot",
            "format": "As specified in the project agreement."
          },
          {
            "id": "output-2",
            "name": "Handover",
            "description": "Defined controls and evaluation criteria",
            "format": "As specified in the project agreement."
          }
        ],
        "workflow": [
          {
            "id": "step-1",
            "name": "Choose a bounded use case",
            "description": "Identify one task, its inputs and its limits.",
            "customerContribution": [],
            "providerContribution": [
              "Identify one task, its inputs and its limits."
            ],
            "outputNames": [
              "Choose a bounded use case"
            ],
            "evidenceIds": [
              "editorial-en"
            ]
          },
          {
            "id": "step-2",
            "name": "Define control points",
            "description": "Agree tools, permissions, outputs and human review.",
            "customerContribution": [],
            "providerContribution": [
              "Agree tools, permissions, outputs and human review."
            ],
            "outputNames": [
              "Define control points"
            ],
            "evidenceIds": [
              "editorial-en"
            ]
          },
          {
            "id": "step-3",
            "name": "Prepare a pilot",
            "description": "Test the proposed flow on an appropriate limited scope.",
            "customerContribution": [],
            "providerContribution": [
              "Test the proposed flow on an appropriate limited scope."
            ],
            "outputNames": [
              "Prepare a pilot"
            ],
            "evidenceIds": [
              "editorial-en"
            ]
          },
          {
            "id": "step-4",
            "name": "Evaluate the result",
            "description": "Review quality, failure cases and the conditions for any next step.",
            "customerContribution": [],
            "providerContribution": [
              "Review quality, failure cases and the conditions for any next step."
            ],
            "outputNames": [
              "Evaluate the result"
            ],
            "evidenceIds": [
              "editorial-en"
            ]
          }
        ]
      },
      "delivery": {
        "engagementModels": [
          "A separately scoped discussion or evaluation; access is not implied."
        ],
        "operationModes": [
          "Operating model and responsibilities agreed for the actual use case."
        ],
        "prerequisites": [
          "A clearly bounded process and authorised owner",
          "Appropriate sample inputs, review criteria and access restrictions"
        ],
        "customerResponsibilities": [
          "A clearly bounded process and authorised owner",
          "Appropriate sample inputs, review criteria and access restrictions"
        ],
        "providerResponsibilities": [
          "A workflow proposal or separately agreed pilot",
          "Defined controls and evaluation criteria"
        ],
        "acceptanceChecks": [
          "Review the agreed functions and deliverables against the project brief.",
          "Confirm relevant interfaces, responsibilities and limitations."
        ],
        "scopeDrivers": [
          "The described platform is internal; the public profile does not provide access or imply a ready-made subscription.",
          "New tools, customer deployments and integrations require separate design and validation.",
          "Outputs and external actions need the review and responsibility agreed for the use case."
        ]
      },
      "integrations": [
        {
          "id": "integration-1",
          "name": "Business and IT tools",
          "category": "workflow",
          "direction": "bidirectional",
          "status": "proposed",
          "description": "Any new connector needs an explicit permission model and implementation scope.",
          "conditions": [
            "Confirm the actual interface, permissions and scope before implementation."
          ],
          "notIncluded": [
            "Compatibility with every third-party system is not implied."
          ],
          "evidenceIds": [
            "editorial-en"
          ]
        }
      ],
      "valueModel": [
        {
          "outcome": "Practical experience for designing an AI pilot",
          "mechanism": "Useful automation needs more than a model response. It needs a clear task, access limits, review points and a way to distinguish a proposal from an action that changes another system.",
          "indicators": [
            "Criteria agreed for the specific project."
          ],
          "measurementNote": "An expected benefit, not a measured result or a guarantee."
        },
        {
          "outcome": "Clearer separation of preparation, review and action",
          "mechanism": "Useful automation needs more than a model response. It needs a clear task, access limits, review points and a way to distinguish a proposal from an action that changes another system.",
          "indicators": [
            "Criteria agreed for the specific project."
          ],
          "measurementNote": "An expected benefit, not a measured result or a guarantee."
        },
        {
          "outcome": "A basis for defining tool permissions and operating responsibilities",
          "mechanism": "Useful automation needs more than a model response. It needs a clear task, access limits, review points and a way to distinguish a proposal from an action that changes another system.",
          "indicators": [
            "Criteria agreed for the specific project."
          ],
          "measurementNote": "An expected benefit, not a measured result or a guarantee."
        }
      ],
      "commercial": {
        "goodFitSignals": [
          "When you want to assess a bounded workflow for AI assistance and can define inputs, responsibilities and acceptance criteria."
        ],
        "poorFitSignals": [
          "The described platform is internal; the public profile does not provide access or imply a ready-made subscription.",
          "New tools, customer deployments and integrations require separate design and validation.",
          "Outputs and external actions need the review and responsibility agreed for the use case."
        ],
        "discoveryQuestions": [
          "What outcome should the project support?",
          "What environment and constraints apply?",
          "Who will review the result and manage its use?"
        ],
        "firstStep": "Start with a short conversation about your intended outcome, the existing environment and a useful first scope.",
        "notIncluded": [
          "The described platform is internal; the public profile does not provide access or imply a ready-made subscription.",
          "New tools, customer deployments and integrations require separate design and validation.",
          "Outputs and external actions need the review and responsibility agreed for the use case."
        ]
      },
      "relationships": [],
      "faq": [
        {
          "question": "Can we log in to the platform?",
          "answer": "The public profile describes an internal LIVE4K project. Customer access or a separate implementation must be discussed independently.",
          "evidenceIds": [
            "editorial-en"
          ]
        },
        {
          "question": "Can an agent act without approval?",
          "answer": "The appropriate approval rules depend on the task. We define those rules and tool permissions explicitly when designing a workflow.",
          "evidenceIds": [
            "editorial-en"
          ]
        }
      ],
      "evidence": [
        {
          "id": "editorial-en",
          "title": "English editorial adaptation of the reviewed public profile",
          "type": "editorial-analysis",
          "publicUrl": "https://live4k.media/projects/ai-agents-business-workflows/",
          "reviewedOn": "2026-11-20",
          "scope": "An English adaptation of the corresponding published Czech profile and its availability boundaries.",
          "limitations": "This adaptation is not a new operational test. The Czech source profile contains the detailed source record."
        }
      ]
    }
  }
}
