The Framework
Four modes of human-agent collaboration
Every workflow in your organisation sits in one of four modes, whether you designed it that way or not. Mode 3 is where the step-change lives, Mode 4 is the frontier, and most organisations are stuck at Mode 2, the copilot ceiling. The diagnostic below scores where you actually are, dimension by dimension.
Mode 01
Human Solo
Humans do all the work, AI is absent or cosmetic. The org chart is the operating system.
Mode 02
Human-led, Agent-assisted
Humans decide, AI suggests or drafts. Copilot territory, a 10 to 15 percent gain and then the ceiling.
Mode 03
Agent-led, Human-governed
Agents execute end-to-end, humans set goals, handle exceptions, hold accountability. The step-change.
Mode 04
Fully Autonomous
Agents operate within guardrails. Humans design the system, not the steps. Already shipping.
Human control
Agent autonomy
The Diagnostic
Score your organisation, eight dimensions
Be ruthless, score where you are today, not where the strategy deck says you should be. The gap between your current mode and where you should be is where the value lives, and most organisations are surprised by the answer.
How are your core business processes designed relative to AI?
Mode 01
Processes are unchanged, AI is not part of any core workflow.
Mode 02
AI tools layered on top of existing processes, copilots and chatbots assist.
Mode 03
Some workflows redesigned end-to-end with AI agents executing core steps.
Mode 04
Core workflows designed AI-first, agents run autonomously, humans govern.
Who makes decisions, and how fast do they move through the organisation?
Mode 01
Decisions flow up the hierarchy, multiple approvals required for action.
Mode 02
AI provides recommendations, humans make all final decisions manually.
Mode 03
Agents make routine decisions autonomously, humans handle exceptions and strategy.
Mode 04
Decision rights mapped to agents and humans by type, real-time and embedded.
How are your teams composed and organised?
Mode 01
Traditional functional teams, org chart defines who works with whom.
Mode 02
Teams use AI tools individually, no structural change to team composition.
Mode 03
Some teams include AI agents as members, hybrid manager roles emerging.
Mode 04
Human-agent teams are the standard unit, agent orchestrators lead blended teams.
How do you govern AI activity and maintain accountability?
Mode 01
No AI governance framework, policies written for a pre-AI world.
Mode 02
Basic AI usage policies exist, periodic review, no real-time monitoring.
Mode 03
Governance embedded in workflows, automated monitoring, clear human escalation paths.
Mode 04
Real-time, data-driven governance, agents self-monitor within guardrails, continuous audit.
Is your data infrastructure ready to fuel agentic operations?
Mode 01
Fragmented data across siloed systems, manual data handling is the norm.
Mode 02
Some data integration, AI tools access limited datasets, quality inconsistent.
Mode 03
Harmonised data layer, agents access contextual data across systems, governance in place.
Mode 04
Unified, real-time data fabric, agents ingest, reason over, and act on cross-system data.
How are you developing people for a human-agent future?
Mode 01
Traditional roles and job descriptions, no AI skills development strategy.
Mode 02
AI literacy programmes launched, roles unchanged, skills bolted on.
Mode 03
New roles created, orchestrators, hybrid managers, career paths being redesigned.
Mode 04
Every role redefined for human-agent collaboration, continuous upskilling is operational.
Does your leadership model enable or block the shift to agentic operations?
Mode 01
Command-and-control leadership, AI seen as a threat or an IT project.
Mode 02
Leaders supportive but not active, AI delegated to technology teams.
Mode 03
Leaders actively redesigning their domains with AI, experimentation encouraged.
Mode 04
System-thinking leaders, ethical AI judgement embedded, continuous adaptation.
How do you measure the impact of human-AI collaboration?
Mode 01
Traditional KPIs only, no measurement of AI contribution or coordination cost.
Mode 02
AI usage tracked (adoption, queries), business impact not measured systematically.
Mode 03
Outcome metrics in place, time reclaimed, decision speed, coordination cost reduced.
Mode 04
Real-time value dashboards, human-agent performance measured as one system.