What an agentic organization looks like
Six dimensions, five levels each. Every workspace scores itself here, blending a quarterly steering-committee assessment with live signals from the work.
Leadership & Strategy
Whether AI is led as a business priority with clear problems, owners and targets, or left to scattered experiments.
- 1Curious. AI is discussed but nobody owns it. Pilots start from vendor demos, not business problems, and stall without a decision.
- 2Sponsored. A named sponsor and a list of problem statements exist. Leaders back the program but don't use Clones themselves.
- 3Committed. Steering meets monthly with real decisions. Year-1 targets are agreed and leaders use Executive Clones visibly every week.
- 4Directing. Leaders set autonomy appetite, fund rollouts from measured results and reallocate work and roles around what Clones now do.
- 5Native. Every planning cycle assumes Clones as part of capacity. Leaders talk about the organization as people plus Clones, and the strategy reflects it.
Governance & Trust
How safely Clones operate: who approves access and autonomy, how evidence is kept, and whether InfoSec is a partner or a gate.
- 1Unmanaged. AI tools are used without policy. InfoSec finds out after the fact, so its default answer is no.
- 2Defined. An AI Use Policy, Governance Charter and risk register exist. Connections are approved one by one, often slowly.
- 3Operating. InfoSec runs a weekly sync, approves connections against a plan and signs promotions on evidence. Audit logs are reviewed on schedule.
- 4Assured. Autonomy rises and falls on evidence with automatic demotion triggers. Audit logs map to the controls the organization reports on.
- 5Self-governing. The organization's own board runs approvals, promotions and evidence packs. Third-party AI sits under the same model, and incidents above agreed severity are zero.
People & Adoption
Whether people use Clones daily, trust them and help each other, and whether the fear of replacement has been answered.
- 1Wary. A few enthusiasts try AI on their own. Most people expect it to threaten their jobs and avoid it.
- 2Introduced. Training has happened and some teams use Clones, but usage drops after launch and questions go unanswered.
- 3Championed. Every live department has a champion, office hours run, and weekly active use is tracked and acted on.
- 4Habitual. Clones are part of how work gets done. People publish workflows, nominate Clone of the Month and coach each other.
- 5Owned. People design their own Clone work, roles have changed to use the time given back, and new hires learn Clones in their first week.
Process & Autonomy
How much of the work Clones actually carry, and how far up the autonomy ladder proven workflows have moved.
- 1Manual. Work runs on people, email and spreadsheets. AI helps individuals ad hoc, with nothing written down.
- 2Assisted. Clones draft and summarize in a few workflows. A person does everything else.
- 3Approved. Proven workflows run at Execute with approval. Clones act on schedules and triggers, and approvals happen in one click.
- 4Delegated. Proven workflows run within limits and escalate exceptions. Work passes between departments' Clones without re-keying.
- 5Orchestrated. Commander coordinates multi-Clone, multi-system work under monitoring. People set goals, own thresholds and handle the exceptions that matter.
Data & Systems
Whether Clones can reach clean, current, well-governed knowledge and systems to work from.
- 1Scattered. Knowledge lives in inboxes and people's heads. Systems are disconnected and data is rekeyed by hand.
- 2Connected. Core systems are connected read-only for a few Clones. Knowledge is loaded but often stale or duplicated.
- 3Curated. Knowledge sources have owners and are kept current. Connections follow an approved plan with least-privilege scopes.
- 4Integrated. CRM, ERP, HR, ticketing and the data warehouse are part of one Integration Architecture. Custom skills are versioned and pinned.
- 5Composable. The organization's own builders add skills and integrations through SkillForge with tests and change control, and data quality is monitored as a standing practice.
Value & Measurement
Whether the organization can prove what Clones changed, in numbers it trusts, and manage cost against value.
- 1Anecdotal. Value is described in stories. No baselines exist and nobody can say what AI costs or saves.
- 2Baselined. Baselines are captured for target workflows and usage is visible, but results aren't yet compared.
- 3Measured. Each live department reports actuals against signed baselines. Hours given back and cycle time are on a dashboard leaders read.
- 4Managed. Value and cost per task drive the roadmap. Low-value Clones are tuned or retired, and the next departments are chosen on measured returns.
- 5Compounding. Value is reported like any core business metric. The CoE reinvests gains into new workflows and the portfolio improves quarter over quarter.