Autonomy Index™

Benchmarks / Targets

What good looks like on the path to an AI-Native workflow.

Use these ranges to read your score honestly and to set realistic thresholds for the next 12 and 24 months. Targets are directional — the right pace depends on your industry, risk profile and team shape.

01 — Maturity bands

Overall score: what each band means

L0
00–20
manual

AI is barely present. Work is fully human-owned and undocumented.

L1
21–40
assisted

AI helps with isolated tasks. No stage is fully owned.

L2
41–60
enabled

AI embedded into parts of the workflow. Ownership is fragmented.

L3
61–80
autonomous

AI owns several meaningful stages with human oversight.

L4
81–100
native

AI owns major parts of the decision and execution chain.

02 — Category benchmarks

Targets by category

Each category contributes a different weight to the overall score. Workflow Ownership carries the most weight — moving it is the fastest path toward AI-Native.

Adoption

Weight 20%
00–30A few individuals using consumer tools, no tracking.
31–55One team using AI weekly. Usage informally reported.
56–75Daily usage across most of the team. Basic usage dashboard.
76–100Embedded into standard workflows. Tied to outcomes.
12-mo target
≥ 65
24-mo target
≥ 80

Unlock: Move from individual usage to team-wide rituals. Make AI use visible in standard reporting.

Capability Enablement

Weight 20%
00–30Basic chat features only. No integrations.
31–55Some capability turned on. One integration with a system of record.
56–75Core features enabled. Connected to several core systems.
76–100Full capability optimised across systems, data, comms and workflow.
12-mo target
≥ 60
24-mo target
≥ 80

Unlock: Audit what's switched off. Connect AI to your systems of record so it sees the full context.

Workflow Ownership

Weight 35%
00–30AI assists drafting. Humans own every stage.
31–55AI owns 1–2 narrow stages. Humans still drive the work.
56–75AI owns 4–6 stages end-to-end with human review.
76–100AI owns the majority of stages. Humans focus on exceptions and direction.
12-mo target
≥ 55
24-mo target
≥ 75

Unlock: Pick 2–3 stages to hand over fully this quarter. Define the human-in-the-loop checkpoint explicitly.

Strategic Impact

Weight 25%
00–30Faster individual tasks. No change in outcomes or operating model.
31–55Measurable productivity gains. Same team shape.
56–75Outcomes improving. Roles starting to shift up a level.
76–100Operating model has changed. Capacity, decisions and outputs are different in kind.
12-mo target
≥ 55
24-mo target
≥ 75

Unlock: Tie AI work to outcome metrics. Redesign roles around what humans uniquely contribute when AI owns execution.

03 — Stage thresholds

Suggested target by workflow stage

Not every stage needs to reach AI-Native. The goal is to know which stages to hand over fully, which to keep human-led, and where the next move should be.

01
Intake
Manual · Manual forms and inboxes
Enabled · AI summarises and tags inbound
Autonomous · AI triages and routes
AI-Native · AI accepts, validates and acts
Autonomous (60+)
02
Research
Manual · Human search and reading
Enabled · AI gathers sources on request
Autonomous · AI compiles briefs with citations
AI-Native · AI maintains live research base
Autonomous (60+)
03
Data gathering
Manual · Manual pulls and spreadsheets
Enabled · Templated AI extractions
Autonomous · AI orchestrates pulls across systems
AI-Native · Continuous AI data layer
Autonomous (60+)
04
Analysis
Manual · Analyst-led
Enabled · AI suggests cuts and patterns
Autonomous · AI runs analysis with human review
AI-Native · AI proposes and validates findings
Enabled → Autonomous (55+)
05
Prioritisation
Manual · Gut feel / spreadsheets
Enabled · AI scores options
Autonomous · AI ranks and recommends
AI-Native · AI decides within guardrails
Autonomous (60+)
06
Decision support
Manual · Meetings and memos
Enabled · AI prepares options
Autonomous · AI recommends with rationale
AI-Native · AI decides on routine calls
Enabled (50+) — keep humans on judgement
07
Content / output
Manual · Human-authored
Enabled · AI drafts, humans edit
Autonomous · AI produces, humans approve
AI-Native · AI produces and ships
Autonomous (65+)
08
Review
Manual · Peer review
Enabled · AI checklists
Autonomous · AI reviews against policy
AI-Native · AI gates quality automatically
Autonomous (60+)
09
Execution
Manual · Humans take action
Enabled · AI prepares next steps
Autonomous · AI executes routine actions
AI-Native · AI executes end-to-end
Autonomous (60+)
10
Follow-up
Manual · Manual reminders
Enabled · AI nudges
Autonomous · AI handles follow-up loops
AI-Native · AI closes the loop
Autonomous (65+)
11
Reporting
Manual · Hand-built decks
Enabled · AI generates reports
Autonomous · AI maintains live dashboards
AI-Native · AI narrates and recommends
Autonomous (65+)
12
Optimisation
Manual · Periodic human review
Enabled · AI suggests improvements
Autonomous · AI tests and iterates
AI-Native · AI continuously self-tunes
Enabled → Autonomous (55+)

How to use these targets

  • Score the workflow today. Mark the weakest category.
  • Set a 12-month target one band above current state — not two.
  • Pick 2–3 stages to move into Autonomous this quarter.
  • Re-score quarterly. Movement matters more than absolute level.

A note on AI-Native

AI-Native isn't the goal for every workflow. Stages with high judgement, regulatory exposure or relationship density should often stay in the Enabled-to-Autonomous range with a strong human-in-the-loop. The point is intentional ownership, not maximum automation.

Score your workflow →See the methodology