How the Q2 2026 edition is built.
The AI Readiness Index scores organisations from the outside, using public evidence only. This page explains each measure, each step and each check, so you can see how each result is made.
From screening universe to published index
The gate a candidate has to clear
5,820 distinct entities
3,359 entities
1,000 organisations
23, of which 22 are scored
Candidates that are not published
A candidate below the gate has too little public evidence to score. A candidate that cleared the gate but is outside the published 1,000 was scored, and its public evidence was weaker. Neither is a verdict on the organisation. A low score can mean the work is invisible rather than missing.
Simple
What each number means
- RPS, Raw Public Signal
- A 0 to 100 company visibility score from five public signals. Visibility, not verified readiness. The index does not claim to see inside an organisation.
- Adjusted RPS
- RPS with a hiring-mix correction applied to the Capability Visibility signal only. A visibility indicator, still not readiness.
- CRP, Country Readiness Prior
- A macro and state readiness prior per market. It is never added to a company score.
- CERS, Country Enterprise Readiness Signal
- A triangulated market signal that shrinks raw company visibility by evidence coverage. It is market-level and never a company score averaged up.
- AER, Adjusted Enterprise Readiness
- Not published this release. The enterprise-adoption benchmark covers 12 of 23 markets and full coverage is not met.
The AI Readiness Index is deterministic and identity-filtered: search results are matched to the named organisation before scoring. Companies with identical RPS scores share a score band; within a band the tie-break is alphabetical by domain and does not distinguish readiness.
Strategy intent that has not shipped, and budget that has not produced visible output, are not measured. A high score does not confirm internal readiness.
The five signals
How a company is scored
RPS is a weighted 0 to 100 score over five public signals. The weights are a design choice, and we tested how much they matter. Under other reasonable weights, 76 to 84 of the top 100 stay in the top 100. Small differences near a tier boundary should not be read as meaningful.
- 25% Discoverability Whether search and AI systems can find and identify the organisation.
- 25% AI and Digital Evidence Public evidence of AI, data, analytics, automation and digital activity.
- 20% Platform and Data Footprint Public references to cloud, SaaS, ERP, CRM and platform maturity.
- 15% Freshness and Authority Recent third-party coverage: launches, partnerships, awards, news.
- 15% Capability Visibility Public hiring and role signals for AI, data, analytics and digital work.
Across the published 1,000, Discoverability has the highest average signal. Capability Visibility is the lowest, after a stricter hiring check removed false matches from job boards and job aggregator sites.
RPS does not distinguish a company that sells AI products from one that adopts AI internally. A consultancy with an AI practice and a bank deploying AI internally can receive similar RPS scores. Supplier and adopter evidence flags are shown on each company profile but are not used to adjust the published order. Read RPS as visibility, not as verified internal readiness.
Visibility tiers
Three tiers, by RPS percentile
The AI Readiness Index groups the published 1,000 into three visibility tiers by RPS percentile. Tiers name what AI systems can observe today. They are not a readiness verdict and they do not measure private internal capability.
| Tier | Share | RPS band | Organisations | Mean RPS |
|---|---|---|---|---|
| Pacesetter | Top 10 percent | 65 to 86 | 100 | 68.8 |
| Momentum | Next 30 percent | 57 to 65 | 300 | 60.6 |
| Emerging | Remaining 60 percent | 49 to 57 | 600 | 52.7 |
Tiers are defined by RPS percentile over all 1,000 published entities, not by a fixed score threshold. An Emerging-tier organisation may hold strong private capability that is simply not visible to outside-in scoring. Pacesetters carry strong, consistent public AI, data and hiring signal. Momentum organisations sit above the index mean with room to consolidate. Emerging organisations carry limited or inconsistent public signal. Many of these organisations are described by AI systems before they publish much about their own AI work.
Evidence confidence
The grade beside every score
Each company is assigned an evidence-confidence grade from the hosts behind its evidence. The grade states the quality of the public evidence, not the performance of the company.
- AStrong. The strongest public evidence. 113 of the published 1,000 hold grade A.
- BModerate. Good public evidence. 92 percent of published organisations carry grade A or B evidence.
- CThin.
- DWeak.
The grade comes from a simple rule. The rule uses the websites behind the evidence and the signal scores. It has not yet been tested against independent data.
Counting companies
Five counts for each market, and how they add up
A market page, the index table and the market dossier each report a company count, and the three differ. They are not estimates of one another. Each answers a different question, and the differences are exact rather than approximate.
| Count | Singapore | What it answers |
|---|---|---|
| Screened | 419 | Every candidate organisation assembled for the market. |
| Below the evidence gate | 154 | Too little public evidence to score. Not a claim that these organisations are unready. |
| Scored | 265 | Cleared the gate. The dossier's working set. |
| Domestic signal entities | 245 | Scored and headquartered in the market. This is the only count used in the market context score. |
| Listed rows | 257 | Scored, after groups scoring under several domains are collapsed to their strongest row. What a reader can count in a table. |
The arithmetic coverage
265 scored plus 154 below the gate equals 419 screened.
Ownership
245 domestic plus 20 subsidiaries and delivery centres equals 265 scored.
Display
265 scored minus 8 collapsed duplicate rows equals 257 listed. The collapsed rows are not dropped organisations: a group that scores under more than one domain is one organisation with several records, and it is shown once at its strongest record, so a table of 257 rows is a table of 257 organisations.
We check these three sums for all 22 scored markets every time we publish. If a market's numbers do not add up, the edition does not publish.
Companies and places
This methodology publishes an order for companies. Places are measured differently.
The index on this page measures organisations and publishes company tiers. AIR APAC also measures how AI systems represent places, and that instrument never ranks one place against another: a place is compared against its own prior measurement and against its published method. The two policies are different because the units are different. How places are measured states that method in full.
Audit appendix
What was checked, and what is still missing
1,000, from 5,820 candidates
100 percent
None
12
Known methodology deviations
| Item | What it is | Status |
|---|---|---|
| Coverage weighting | A temporary formula sets how much weight each market's company evidence carries in its context score. It is not yet tested. We will replace it when better coverage data exists. | Disclosed |
| Outsourcing correction | A rule based on company names flags outsourcing and offshore service firms. It is accurate when it fires but misses many firms, so the hiring correction relies mostly on other evidence. | Disclosed |
| CERS precision | Published as an integer plus a band spanning the k=75, k=150 and k=300 sensitivity results. Precise sensitivity values are retained in the audit JSON. | Disclosed |
| Evidence grade | A simple rule based on the websites behind the evidence and the signal scores. Not yet tested against independent data. | Disclosed |
Planned improvements are listed on the limitations page. Full source vintage, coverage and licence notes are on the sources page.
Technical
Why markets are not ranked against each other
Markets that we sourced more deeply also score higher on every signal. So we cannot tell whether a gap between two markets is real, or a result of how deeply we searched. For this reason we compare organisations within their own market and sector.
This closes the cross-market layer entirely rather than downgrading it to context. AIR APAC publishes no ranking, no dispersion comparison and no adoption-ratio comparison across markets. The estimand we do publish is a company's position among named firms in the same market and the same sector, which never crosses a market boundary and is unaffected by either defect.
Sourcing depth per market correlates with the country prior at r = 0.729, and each of the five public signals correlates with sourcing depth at between +0.43 and +0.65.
Market context, for the record
CRP v0.1 blends Oxford Government AI Readiness 2024 at 0.55 and the IMF AI Preparedness Index 2023 at 0.45. CERS triangulates the CRP prior, the sector-stratified public company signal and enterprise adoption with a coverage-based shrinkage, so it reads lower than a raw mean where evidence is thin. CERS is published as an integer plus a band to avoid false precision.
The hiring correction is evidence-based, using official-careers against third-party job-board counts, plus a market posting-bias factor. Each company is flagged as supplier or adopter, classed by entity type, and assigned its evidence-confidence grade. Rows are deduplicated only by exact normalised bare domain; different domains remain separate until reviewed.
Citation
How to cite this work
Source: AIR APAC AI Readiness Index, Q2 2026 results. Center for AI Readiness in Asia Pacific (AIR APAC). airapac.org/research. The underlying data is available as a CSV download.
The Index names organisations from what they publish about AI work. The order of the list is not a measure of quality. No organisation can pay to be included. Our method is public.
Cite RPS as a public-signal visibility score. Do not describe the published order as validated readiness, and do not average company scores into a market readiness claim. CERS is published per market as that market's own context; it is not a cross-market ranking, and no market-to-market comparison of CERS or mean RPS is published.
Read the limits next
Every known limit of this method is published on one page.