Possible
What could reasonably be done?
Built from current capabilities, business tasks, firm context, and practical constraints.
Methodology
How AIBI keeps what’s possible separate from what’s observed.
AIBI first maps what today’s AI can reasonably do, then reviews credible evidence of what industries report using. Keeping those questions separate prevents adoption data from changing the assessment of what is technically and practically possible.
Actual
What is credibly observed?
Built from external adoption evidence, normalization, evidence grades, and uncertainty.
AI in practice
The levels describe implementation depth—not the media AI uses.
- Standard AI covers mainstream language, document, voice or image assistance that a person starts, checks and controls.
- Integrated AI connects approved business data and systems for bounded multi-step workflows with defined access and ownership.
- Advanced AI covers continuous, multimodal or agentic systems that can plan and prepare actions across tools, with approval gates and accountable people.
- Voice, vision or multimodal generation is not automatically Advanced. Depth depends on integration, autonomy, risk and operational control.
- Country guidance describes the practical implementation path for the selected industry; it does not change the underlying global capability assessment.
Capability horizon · 2026.10
A dated view of what businesses can use now.
Effective .
- The horizon covers reasoning, language, documents, voice, vision, image and video creation, data and code, computer use, connected agents and selected physical AI.
- Product names are evidence of capability, not the taxonomy. A capability remains in the model even as vendors and models change.
- The same 9 common business functions are considered for every country and industry: a personal assistant, marketing, sales, service, administration, people, operations, technology and risk.
- Industry-specific opportunities are added separately, so specialist work does not crowd out the functions nearly every business performs.
- Country guidance then describes practical constraints such as privacy, infrastructure, language, cost and sector rules; it does not pretend the underlying technology is different by country.
- This snapshot is reviewed and timestamped because capability, price, reliability and availability change.
- 01
Current AI capability
Stable capability identities receive dated assessments of maturity, affordability, reliability, integration difficulty, oversight, and risk.
- 02
Business tasks
Capabilities are mapped to concrete work rather than broad claims about replacing jobs or industries.
- 03
Practicality
Transparent weights and safety caps convert capability-task mappings into a reproducible task score.
- 04
Industry baseline / Possible
Task scores roll into business functions and then a representative industry archetype.
- 05
Actual adoption evidence
Government surveys, credible studies, and other observations are normalized without entering the Possible calculation.
- 06
AI Utilization Gap
A gap is calculated only when Possible and Actual measure compatible populations, periods, constructs, and scales.
- 07
Evidence confidence
Grades and confidence communicate directness, source quality, uncertainty, and limitations without false precision.
- 08
Versioning over time
Capabilities, evidence, methods, and scores are timestamped so the index can change transparently as AI changes.
Non-negotiable rules
Missing is a finding, not a zero.
- Missing adoption evidence remains missing unless a separately labeled model is introduced.
- Modeled estimates and observed values are never presented as the same thing.
- Evidence quality may vary by country, industry, population, and measured concept.
- Every score must be reproducible from dated inputs and a declared score version.
- Scores change as capabilities, costs, reliability, and evidence change.
Illustrative snapshot
How the hypothetical adoption gap works.
- The Standard AI destination is fixed at 100%: businesses in the industry using at least one applicable Standard AI opportunity. It is not 100% automation or task coverage.
- Actual is one reported country-industry rate, or the closest disclosed broad-sector proxy when no narrower rate is available.
- The displayed headroom is 100% minus that reported Actual, expressed in percentage points. It is not the scientific task-utilization Gap Score.
- Rates with different questions or periods are not combined, averaged, or silently converted.
- If no suitable current reported rate exists, Actual and the hypothetical gap remain empty.
- Every result is a dated snapshot and can change as newer or more specific evidence is found.