Agents
TAOM™ AI Agent Skill Profiles
Joint, specialised support roles for Human + AI project support in day-to-day operations. The profiles serve as technical ‘skill CVs’ for the design, development, training, governance and operation of the TAOM agent ecosystem.
Agents are shared, specialised support roles – not competing digital colleagues. They all operate with explicit authorisations, access shared knowledge sources and follow binding human-in-the-loop rules. The TAOM Companion acts as the common gateway; the Method & Orchestration Agent assigns enquiries to the appropriate methods, artefacts, responsibilities and specialist agents.
Shared basic profile for each agent
Identity & Mandate
Unique ID, name, version, owner, business sponsor; clear purpose, defined responsibilities, defined scope.
TAOM integration & user base
Referenced concepts and canonical chapters; defined roles, projects and organisational units.
Knowledge & Tools
Approved sources with owner, validity and version; documented connectors with clear read/write permissions.
Autonomy & Human-in-the-Loop
Permitted actions, confidence thresholds, escalation triggers and explicit human approval points.
Traceability & Security
Comprehensive logging (source, context, rationale, action, result); least privilege, data protection, classification, client segregation.
Quality, Operations & Learning
Test cases, hallucination checks, consistency checks; monitoring, incident handling, versioning, rollback; learning only from verified feedback.
Companion & Orchestration
🚪 TAOM Companion
Shared user interface and personalised entry point to the agent ecosystem – the intuitive interaction layer between humans and TAOM. Understands natural language and intent, recognises roles, expertise and language preferences, conducts structured dialogues with summaries and confirmations, and passes Project_ID, Process_ID and Role_ID to the orchestration. Cannot authorise critical tool changes or override roles, permissions or business decisions.
🧭 Method & Orchestration Agent
The methodological and technical coordination hub: maintains project methodology, TAOM structure, artefact logic, roles, dependencies, governance and traceability. Proficient in TAOM constitution, theory, OS, management, architecture, engineering and tooling, as well as PMI, IPMA, PRINCE2, Scrum, SAFe and SAP Activate; utilises the Project Context Graph with stable Enterprise Object IDs. Coordinates, analyses and creates drafts – critical releases and production changes require human approval.
The specialist agents in detail
🔀 Process Agent
Skills: Proficient in end-to-end (E2E) processes, BPML, BPMN 2.0, SIPOC and value streams; models events, tasks, gateways, lanes, message flows, data objects and controls; understands the L0–L4 hierarchy and process landscapes; navigates Signavio structures, dictionaries, attributes and roles; translates unstructured statements into clear, testable process information and checks for completeness, clarity, consistency and compliance with modelling rules.
Scope of autonomy: Creates models and changes as drafts. Does not publish binding process versions independently.
📚 Knowledge Agent
Capabilities: Collects, structures, evaluates, consolidates and links knowledge from authorised sources; operates Confluence, SharePoint and Langdock; performs semantic searches using metadata and context filters; synthesises multiple sources and flags contradictions and outdated content; documents processes, decisions, how-to guides, FAQs and specifications, including owners, validity, review dates, classification and approval status.
Limits on autonomy: Consolidates content and creates drafts. Does not independently approve any normative or binding statements.
✅ Delivery Agent
Skills: Manages projects, programmes, products and services using agile, predictive and hybrid models; is proficient in epics, features, user stories, acceptance criteria, tasks, sprints and backlogs, as well as work packages, milestones, plans, statuses, risks and dependencies; is familiar with SAP Activate, Fit-to-Standard, WRICEF, Customising, Development and Testing; is proficient in Jira/Azure DevOps issue types, workflows, fields, links, boards and releases.
Limits to autonomy: Prepares draft delivery artefacts. Does not set priorities, scope or binding release commitments independently.
🤝 Collaboration Agent
Skills: Analyses meetings, chats, emails and communication content; identifies intentions, decisions, tasks, risks, blockers and open issues; supports the agenda, minutes, decision log and action items; uses Microsoft Teams, Outlook and calendars; assigns responsibilities and forwards content to Jira, Confluence, the decision register and process repositories.
Limits on autonomy: Does not publish unconfirmed statements as decisions, tasks or external communications.
📦 Application Agent
Capabilities: Establishes business and technical links between processes, roles, systems, transactions, data objects and integrations; understands S/4HANA or ECC modules, Fiori apps, transactions, Customising and interfaces; assigns application steps to end-to-end (E2E) and L4 processes; distinguishes between master, transaction, reference and integration data, as well as between standard, configuration, extension, development and workaround; integrates SAP, APIs, SAP Integration Suite and ServiceNow.
Scope of autonomy: Prepares solution options and changes. Does not authorise productive configuration, development or data changes independently.
🛡️ Quality & Compliance Agent
Skills: Links requirements, risks, controls, tests, evidence, defects, approvals and regulatory requirements; structures test strategies, test cases, defects and evidence; identifies risks, defines control objectives and collects evidence; interprets guidelines and audit requirements; addresses access, segregation of duties, data protection, classification and logging; operates Jira/ALM, GRC and test management.
Limits on autonomy: Prepares reviews, risks, controls and evidence. Does not issue binding approvals independently.
🎓 Learning & Adoption Agent
Skills: Tailors explanations, learning paths, guides and exercises to the role, prior knowledge, task and work situation; designs microlearning, step-by-step guides, FAQs, simulations and exercises; understands change management dynamics and adoption mechanisms; embeds learning directly into processes, tasks, applications and changes; uses LMS, Confluence, Teams, TAOM Companion and in-app guidance.
Limits on autonomy: Personalises content and creates drafts. Does not publish binding training materials without business approval.
Via connected, controlled enterprise tools
A Context Graph connects people, processes, capabilities, applications, requirements, decisions, controls, tests, knowledge and results via stable Enterprise Object IDs (Project_ID, Process_ID, Capability_ID, Role_ID, Requirement_ID …).
Binding Human-in-the-Loop Matrix
- Search for & summarise information: automatically, within authorised permissions and sources.
- Explain relationships & sources: automatically, with traceable sources and visible context.
- Create a draft (process, document, story, test): automatically, but clearly marked as a draft.
- Make significant changes to existing content: confirmation or formal approval by the responsible role.
- Publishing a process model: approval by the Process Owner required.
- Changing scope or priority: decision by the Product Owner or project management.
- Creating critical Jira objects: only in accordance with defined project governance and existing authorisation.
- Modify SAP production data: only within approved rights, processes and controls.
- Quality or compliance approval: binding decision by the responsible person.
- Security or data protection decision: binding decision by the responsible person.
- Conflict between sources: flag it, do not resolve it silently, escalate it to the person responsible.
- Critical exception or low confidence: no action; escalate immediately to the responsible role.
Target state
Employees remain at the heart of the hybrid organisation: They describe their task; the Companion identifies the role, context and objective; the Method & Orchestration Agent assigns methodology, processes, artefacts, governance and subject-matter agents; specialised agents prepare content and controlled tool actions – and the responsible individuals review, decide and approve. Tools remain connected via stable IDs, metadata and traceability; feedback and results improve knowledge, processes and the organisation within controlled learning and governance mechanisms.
People create value. AI agents augment people. TAOM preserves method, context, governance and traceability.