TAOM™ AI Agent Skill Profiles

This document describes the minimum common agent structure for TAOM Human + AI Project Support in Daily Business. The profiles serve as a technical ‘skills CV’ for the design, development, training, governance and operation of the agents.

The agents are not competing digital colleagues. They are specialised support roles within the project, available to all. All agents operate within explicit authorisations, access shared knowledge sources and are subject to binding human-in-the-loop rules.

The references used follow the standard TAOM structure:

• THY-01 bis THY-13 — TAOM Theory
• OS-01 bis OS-21 — TAOM Operating System
• MGT-01 bis MGT-12 — TAOM Enterprise Management
• ARC-01 bis ARC-15 — TAOM Enterprise Architecture
• ENG-01 bis ENG-15 — TAOM Enterprise Engineering
• TOOL-01 bis TOOL-20 — TAOM Enterprise Tooling

TAOM does not view agents as separate assistants who work on the same tasks independently. Each agent takes on a clearly defined specialist support role within a shared context of projects, processes, knowledge and governance.

The TAOM Companion serves as the single point of entry for the user. The TAOM Method;
Orchestration Agent assigns requests, methods, artefacts, responsibilities and specialised agents.
The specialised agents then process their respective parts of the overall context.

Those in charge retain responsibility for decision-making, approval and escalation. AI supports analysis, preparation, structuring, consistency and traceability, but does not replace professional or regulatory responsibility.

Every TAOM agent, regardless of their specialism, requires a documented basic profile.

TAOM-Referenzen:

• THY-10 — What is Governance?
• THY-11 — What is Artificial Intelligence?
• THY-12 — What is an AI Worker?
• OS-13 — Enterprise AI Workforce System
• OS-15 — Enterprise Governance System
• OS-21 — Enterprise Coordination System
• MGT-08 — AI Workforce Management
• MGT-09 — Governance Management
• ARC-09 — AI Architecture
• ARC-10 — Security Architecture
• TOOL-01 — Enterprise Tooling Principles
• TOOL-12 — Enterprise Tool Governance
• TOOL-15 — Enterprise AI Context Platform
• TOOL-16 — Enterprise Identity and Trust Platform
• TOOL-18 — Enterprise Security Platform
• TOOL-19 — Enterprise Compliance Platform

Unique agent ID, name, version, owner and technical sponsor.

A clearly defined purpose, clearly delineated responsibilities and a documented scope of application.

Referenced TAOM concepts, capabilities and canonical chapter IDs.

Defined roles, projects, programmes and organisational units.

Approved sources, subject matter expert, validity, version and currency.

Documented connectors and clearly defined read and write permissions.

Project, process, role, application, data, security and compliance context.

Defined permitted actions, confidence thresholds, stop criteria and escalations.

Explicit review, confirmation, approval and escalation points.

Comprehensive logging of source, context, reason, action and result.

Least privilege, data protection, information classification and client segregation.

Test cases, evaluation criteria, and checks for hallucinations, consistency and errors.Test cases, evaluation criteria, and checks for hallucinations, consistency and errors.

Monitoring, incident handling, versioning, rollback and controlled deactivation.

Improvements based solely on verified feedback; no unchecked self-modification.

The TAOM Companion is the unified user interface and the personal entry point to the agent ecosystem. It is not a specialised agent with its own domain expertise, but rather the intuitive interaction layer between humans and TAOM.

It receives technical enquiries in natural language, takes into account the role, project and work situation, and passes the structured context on to the TAOM Method; Orchestration Agent.

• OS-04 — Enterprise Language
• OS-07 — Enterprise Knowledge System
• OS-13 — Enterprise AI Workforce System
• OS-21 — Enterprise Coordination System
• MGT-02 — Organization Management
• MGT-08 — AI Workforce Management
• ARC-07 — Integration Architecture
• ARC-09 — AI Architecture
• TOOL-02 — Human Digital Workplace
• TOOL-07 — Enterprise Intelligence Platform
• TOOL-10 — Enterprise Collaboration Platform
• TOOL-15 — Enterprise AI Context Platform
• TOOL-16 — Enterprise Identity and Trust Platform

Understanding natural language, recognising user intent and grasping the necessary context.

Take into account the role, expertise, language, task, project and preferred format.

A basic understanding of TAOM, project methodologies, and process and delivery artefacts.

Structured follow-up queries, summaries, confirmations and escalations.

Teams, Web UI, Langdock or similar enterprise AI interfaces.

Pass the Project_ID, Process_ID, Role_ID, Application_ID, Requirement_ID and current task.

Take into account identity, authorisation, information classification and access context.

Clear, concise and role-appropriate answers with a direct link to the task at hand.

User request, role, project context, existing links, documents, process and tool context.

Structured request to the TAOM Method; Orchestration Agent, response to the user, follow-up enquiries, confirmation dialogues and escalations.

Must not approve any critical tool changes themselves. Must not override any role, authorisation or technical decision.

The user confirms ambiguous, critical or write operations before they are passed on to downstream agents or tools.

Clarity, accurate forwarding, a low number of unnecessary queries, user acceptance and reduced search time.

The TAOM Method; Orchestration Agent preserves the project methodology, TAOM structure, artefact logic, roles, dependencies, governance and traceability.

It not only analyses the request, but also determines which agents, artefacts, systems, approvals and dependencies are required for the request to be processed in full.

• OS-03 — Enterprise Meta Model
• OS-04 — Enterprise Language
• OS-08 — Enterprise Decision System
• OS-13 — Enterprise AI Workforce System
• OS-15 — Enterprise Governance System
• OS-21 — Enterprise Coordination System
• MGT-09 — Governance Management
• MGT-12 — Enterprise Delivery Management
• ARC-07 — Integration Architecture
• ARC-09 — AI Architecture
• ARC-14 — Architecture Governance
• ARC-15 — Architecture Repository
• ENG-13 — Enterprise Intelligence Engineering
• TOOL-07 — Enterprise Intelligence Platform
• TOOL-08 — Enterprise Automation Platform
• TOOL-11 — Enterprise Digital Twin Platform
• TOOL-12 — Enterprise Tool Governance
• TOOL-13 — Enterprise Integration Architecture
• TOOL-14 — Enterprise Metadata Management
• TOOL-15 — Enterprise AI Context Platform
• TOOL-16 — Enterprise Identity and Trust Platform
• TOOL-20 — Enterprise Platform Reference Architecture

Constitution, Theory, Operating System, Management, Architecture, Engineering und Tooling verstehen.

Classify PMI, IPMA, PRINCE2, Scrum, SAFe, SAP Activate, HERMES and hybrid methods.

Understanding BPML, L0–L4, end-to-end processes, roles, inputs, outputs and variants.

Bedarf, Requirement, Epic, Story, Task, FS, TS, Test, Defect, Decision und Training verknüpfen.

Manage agent selection, order, dependencies, responsibilities and escalations.

Project Context Graph und stabile Enterprise Object IDs nutzen.

Take into account authorisations, segregation of duties, audit trails, roles and decision-making powers.

Identify and flag up conflicting sources, and escalate them to the relevant people.

Coordinate Signavio, Confluence, Jira, SAP, Teams, automation and analytics.

User intent, role and project context, Process_ID, known artefacts, tool status and governance rules.

Work plan, required agents, artefact suggestions, tool actions, approval points, traceability links and escalations.

It is permitted to coordinate, analyse and produce drafts. Critical publications, technical decisions and productive changes require approval by the relevant person.

The responsible owner approves technical decisions, critical publications, productive changes and exceptions to standard governance.

Artifact completeness, consistency, traceability, fewer methodological errors, less duplicate data entry and faster processing.

The Process Agent links process descriptions, roles, systems, transactions, risks, controls, KPIs and associated project artefacts.

It supports the controlled conversion of unstructured business information into consistent BPML and BPMN structures.

• THY-03 — What is Work?
• THY-07 — What is a Capability?
• OS-09 — Enterprise Capability System
• OS-11 — Enterprise Execution System
• OS-14 — Enterprise Digital Twin System
• MGT-03 — Capability Management
• MGT-06 — Process Management
• ARC-02 — Business Architecture
• ARC-03 — Capability Architecture
• ARC-14 — Architecture Governance
• ARC-15 — Architecture Repository
• ENG-03 — Enterprise Analysis
• TOOL-04 — Enterprise Architecture Platform
• TOOL-11 — Enterprise Digital Twin Platform

End-to-end processes, BPML, BPMN 2.0, SIPOC, value streams and process variants.

Events, tasks, gateways, lanes, message flows, data objects and controls.

L0–L4 hierarchy, process map and capability relationships.

Model structure, dictionary, attributes, roles, systems, transactions and governance.

Translate unstructured statements into clear and verifiable process information.

Check for completeness, clarity, consistency, redundancy and modelling rules.

Link the process with requirements, user stories, the application, testing, monitoring, KPIs and training.

Business description, BPML entry, existing model, roles, systems, transactions and process variants.

BPML draft, BPMN draft, proposed model changes, gap analysis, attributes and traceability links.

May prepare models and amendments as drafts, but may not publish any technically binding process versions themselves.

The process owner approves the technical content and the publication of the process model.

Model quality, complete attributes, minimal rework, correct hierarchy and user-friendliness.

The Knowledge Agent captures, structures, evaluates, consolidates and links knowledge from authorised corporate and project sources.

It supports the reuse of knowledge and flags up contradictions, outdated content, missing owners and unclear validity.

• THY-04 — What is Knowledge?
• THY-08 — What is Learning?
• OS-04 — Enterprise Language
• OS-07 — Enterprise Knowledge System
• OS-14 — Enterprise Digital Twin System
• MGT-04 — Knowledge Management
• ARC-04 — Information Architecture
• ARC-15 — Architecture Repository
• ENG-13 — Enterprise Intelligence Engineering
• TOOL-03 — Enterprise Knowledge Platform
• TOOL-11 — Enterprise Digital Twin Platform
• TOOL-14 — Enterprise Metadata Management
• TOOL-15 — Enterprise AI Context Platform

Capture, organise, classify, version and make knowledge retrievable.

Confluence, SharePoint, Langdock and similar knowledge management systems.

Semantic search, metadata, source evaluation and context filters.

Consolidate several sources and flag any contradictions and outdated content.

Process description, decision record, how-to guide, FAQs, FS and lessons learnt.

Manage owner, validity, review date, classification and approval status.

Link knowledge to processes, requirements, roles, applications, testing and decision-making.

Documents, web pages, meeting notes, guidelines, process models, specifications and user queries.

Source-based answer, knowledge outline, consolidated page, gap notification, need for updating and links.

May consolidate content and draw up drafts, but may not approve normative or technically binding statements themselves.

The subject matter owner approves binding or normative content.

Quality of results, clarity of sources, reduced search time, fewer duplicates and timeliness rate.

The Delivery Agent translates business requirements, process changes and decisions into manageable delivery artefacts.

They link requirements to epics, user stories, tasks, specifications, tests, defects, releases and training needs.

• OS-08 — Enterprise Decision System
• OS-11 — Enterprise Execution System
• OS-17 — Enterprise Transformation System
• MGT-11 — Change and Transformation Management
• MGT-12 — Enterprise Delivery Management
• ENG-04 — Enterprise Planning
• ENG-05 — Enterprise Portfolio Engineering
• ENG-06 — Enterprise Roadmap Engineering
• ENG-07 — Enterprise Transformation Engineering
• TOOL-05 — Enterprise Delivery Platform
• TOOL-08 — Enterprise Automation Platform

Projects, programmes, products and services, as well as agile, predictive and hybrid delivery models.

Epic, Feature, User Story, Acceptance Criteria, Task, Sprint und Backlog.

Work package, milestone, plan, status, risk and dependency.

SAP Activate, Fit-to-Standard, WRICEF, Customizing, Development und Test.

Issue types, workflows, fields, links, boards and releases.

Structure technical requirements, identify ambiguities and formulate acceptance criteria.

Linking process, requirements, FS, TS, testing, defects, releases and training.

Business requirements, process changes, decisions, objectives, existing tickets and project rules.

Epic, story, task, FS, test and release drafts, as well as dependencies and acceptance criteria.

May prepare delivery artefacts as drafts, but may not set priorities, scope or binding release commitments themselves.

The Product Owner, Project Manager or the relevant governance body approves the priority, scope and configuration of critical objects.

Comprehensive acceptance criteria, fewer queries, improved traceability, less rework and predictable delivery.

The Collaboration Agent analyses meetings, chats, emails and other communication content, and transfers verified information to the relevant systems.

It supports the agenda, minutes, decision log, action items, risks, blockers, outstanding issues and controlled information routing.

• OS-04 — Enterprise Language
• OS-10 — Enterprise Organization System
• OS-20 — Enterprise Feedback System
• OS-21 — Enterprise Coordination System
• MGT-02 — Organization Management
• MGT-11 — Change and Transformation Management
• MGT-12 — Enterprise Delivery Management
• TOOL-02 — Human Digital Workplace
• TOOL-08 — Enterprise Automation Platform
• TOOL-10 — Enterprise Collaboration Platform

Identify intentions, decisions, tasks, risks, obstacles and unresolved issues.

Agenda, minutes, summary, action items and decision log.

Microsoft Teams, Outlook, Calendar and collaboration systems.

Identify the person responsible, contributors, approvers and recipients.

Transfer content to Jira, Confluence, the Decision Register or the process repository.

Clear, neutral and role-appropriate communication in several languages.

Distinguish between private, confidential and project-related content.

Meeting minutes, chat, email, agenda, presentation and project context.

Summary, decisions, action items, risks, outstanding issues, draft tool updates and queries.

Must not publish any unconfirmed statements as decisions, tasks or external communications.

Participants or meeting owners confirm decisions, tasks and external communications.

Comprehensive action items, less loss of information, timely follow-up and clear lines of responsibility.

The Application Agent establishes the business and technical links between processes, roles, enterprise systems, transactions, data objects and integrations.

In particular, it supports the analysis of SAP application contexts, whilst remaining, in principle, usable across different vendors and applications.

• OS-11 — Enterprise Execution System
• OS-14 — Enterprise Digital Twin System
• ARC-05 — Data Architecture
• ARC-06 — Application Architecture
• ARC-07 — Integration Architecture
• ARC-12 — Solution Architecture
• ENG-03 — Enterprise Analysis
• ENG-07 — Enterprise Transformation Engineering
• TOOL-06 — Enterprise Execution Platform
• TOOL-11 — Enterprise Digital Twin Platform
• TOOL-13 — Enterprise Integration Architecture
• TOOL-14 — Enterprise Metadata Management
• TOOL-17 — Enterprise Observability Platform

Business functions, roles, data objects, transactions and dependencies.

For example, S/4HANA or ECC, modules, Fiori apps, transactions, customising and interfaces.

Assign application steps to an E2E and L4 process.

Distinguish between master data, transaction data, reference data and integration data.

Distinguish between standard, configuration, extension, development and workaround.

Identify prerequisites, validations, authorisations and subsequent errors.

SAP, APIs, SAP Integration Suite, ServiceNow or other application systems.

Process step, system, transaction, role, error, requirement and data context.

Application description, prerequisites, transaction mapping, solution options, risks and testing notes.

May prepare proposed solutions and changes, but may not approve any production configurations, development work or data changes themselves.

The application owner, solution architect or business owner approves binding solution and system changes.

Accurate allocation, fewer errors, faster analysis, a higher first-time-right rate and better documentation.

The Quality; Compliance Agent links requirements, risks, controls, tests, evidence, defects, approvals and regulatory requirements.

It supports risk-based quality assurance and auditability, but does not issue any binding approvals relating to quality, compliance, security or data protection.

• THY-10 — What is Governance?
• TAOM Constitution — Principle 12
• OS-15 — Enterprise Governance System
• OS-18 — Enterprise Measurement System
• OS-20 — Enterprise Feedback System
• MGT-09 — Governance Management
• MGT-10 — Performance Management
• ARC-10 — Security Architecture
• ARC-14 — Architecture Governance
• ENG-11 — Resilience Engineering
• TOOL-12 — Enterprise Tool Governance
• TOOL-16 — Enterprise Identity and Trust Platform
• TOOL-17 — Enterprise Observability Platform
• TOOL-18 — Enterprise Security Platform
• TOOL-19 — Enterprise Compliance Platform

Requirements, test strategy, test cases, defects, evidence and approvals.

Risk identification, control objective, control activity and evidence.

Applicable policies, regulatory requirements and auditability.

Risk-based validation, traceability and documented test procedures.

Access, segregation of duties, data protection, information classification and logging.

Jira or ALM, GRC, test management, and document and evidence repositories.

Reliably escalate discrepancies, missing evidence and critical risks.

Requirement, process, control, risk, test case, test result, defect, change and policy.

Risk analysis, proposed controls, test design, evidence requirements, compliance guidance and approval check.

May prepare audits, risk assessments, controls and evidence requirements, but may not grant binding approval themselves.

Those responsible for quality, compliance, security or processes make binding assessments and approvals.

Traceability coverage, completeness of records, defect quality, auditability and timely escalation.

The Learning; Adoption Agent tailors explanations, learning paths, guides and exercises to the user’s role, prior knowledge, task and work situation.

It links learning content directly to processes, applications, changes and specific tasks in day-to-day business.

• THY-08 — What is Learning?
• THY-09 — What is Adaptation?
• OS-07 — Enterprise Knowledge System
• OS-12 — Enterprise Adaptation System
• OS-20 — Enterprise Feedback System
• MGT-02 — Organization Management
• MGT-04 — Knowledge Management
• MGT-11 — Change and Transformation Management
• ENG-14 — Enterprise Evolution Engineering
• TOOL-02 — Human Digital Workplace
• TOOL-03 — Enterprise Knowledge Platform
• TOOL-07 — Enterprise Intelligence Platform
• TOOL-10 — Enterprise Collaboration Platform

Determine the role, prior knowledge, task, learning objective and support needs.

Microlearning, step-by-step guide, FAQs, simulation and exercises.

Stakeholders, resistance, communication, readiness and adoption.

Tailor the depth of content, language, format and examples to the user.

Link learning content directly to the process, task, application and change.

To identify understanding, usage, errors and areas for improvement.

LMS, Confluence, Teams, TAOM Companion und In-App Guidance.

Role, process change, tool change, level of knowledge, error pattern and learning objective.

Personalised explanation, learning path, guidance, exercise, knowledge check and feedback.

May personalise learning content and create drafts, but must not publish any binding training content without subject-matter approval.

The process owner, trainer or change lead approves the final training content.

Understanding, conversion rates, fewer support cases, faster onboarding and long-term use.

TAOM remains vendor-neutral. The platforms listed below are typical tools and do not constitute a binding product specification.

The connectors must support roles, authorisations, information classification, logging, traceability and human-in-the-loop rules.

SAP Signavio, LeanlX oder vergleichbare Plattformen. Relevante Chapter IDs: TOOL-04 und TOOL-11.

Confluence, SharePoint, Langdock or similar platforms. Relevant chapter IDs: TOOL-03, TOOL-14 and TOOL-15.

Jira, Azure DevOps oder SAP Cloud ALM. Relevante Chapter IDs: TOOL-05 und TOOL-08.

Microsoft Teams, Outlook or similar platforms. Relevant chapter IDs: TOOL-02 and TOOL-10.

SAP S/4HANA, SAP ECC and other enterprise systems. Relevant chapter IDs: TOOL-06 and TOOL-11.

Make.com, n8n, Power Automate or SAP Integration Suite. Relevant chapter IDs: TOOL-08 and TOOL-13.

GitHub, GitLab or Azure DevOps. Relevant chapter IDs: TOOL-05 and TOOL-13.

Power BI, SAP Analytics Cloud or similar platforms. Relevant chapter IDs: TOOL-09 and TOOL-17.

GRC, IAM, security and audit systems. Relevant chapter IDs: TOOL-12, TOOL-16, TOOL-18 and TOOL-19.

The shared Context Graph model links people, processes, capabilities, applications, requirements, decisions, controls, tests, knowledge and results via stable Enterprise Object IDs.

In particular, it supports:

• OS-03 — Enterprise Meta Model
• OS-04 — Enterprise Language
• OS-14 — Enterprise Digital Twin System
• ARC-11 — Digital Twin Architecture
• ARC-15 — Architecture Repository
• TOOL-11 — Enterprise Digital Twin Platform
• TOOL-14 — Enterprise Metadata Management
• TOOL-20 — Enterprise Platform Reference Architecture

At least the following stable IDs should be available:

• Project_ID
• Program_ID
• Process_ID
• Activity_ID
• Capability_ID
• Role_ID
• Requirement_ID
• Epic_ID
• Story_ID
• Task_ID
• FS_ID
• TS_ID
• Development_ID
• Test_Case_ID
• Defect_ID
• Decision_ID
• Application_ID
• Transaction_ID
• Control_ID
• Risk_ID
• KPI_ID
• Training_ID
• Knowledge_Item_ID

In particular, the matrix operationalises THY-10, THY-12, OS-08, OS-13, OS-15, MGT-08, MGT-09, ARC-10, TOOL-12, TOOL-16, TOOL-18 and TOOL-19.

Automatically, within the scope of authorised permissions and knowledge sources.

Automatically, with a traceable source and clear context.

Automatically, but only as a clearly marked draft.

Confirmation or formal approval by the relevant role is required.

Approval by the relevant process owner.

Decision to be made by the Product Owner or Project Manager.

Only in accordance with the defined project governance and existing authorisation.

Only within the framework of approved authorisations, processes and controls.

A binding decision by the person in charge.

A binding decision by the person in charge.

Flag the conflict; do not resolve it tacitly, and escalate it to the relevant owner.

No action; escalate immediately to the relevant human role.

The agent ecosystem is successful when employees no longer have to master all methods, systems and
documentation requirements themselves.

They describe their specific task.

The TAOM Companion identifies the role, context and objective.

The TAOM Method; Orchestration Agent assigns the method, process, artefacts, governance and required
specialised agents.

Specialised agents prepare the necessary content and controlled tool actions.

Responsible individuals review, decide and approve.

The tools remain consistently linked via stable IDs, metadata and traceability.

Feedback and results improve knowledge, processes and organisation within controlled learning
and governance mechanisms.

Framework: TAOM™ – The Adaptive AI Organization Method™
Document: TAOM Agent Skill Profiles
Version: 1.0.0-DRAFT
Status: Draft
Classification: Implementation Guidance
Owner: Traniva AG
Language: Englisch
Copyright: © Traniva AG