— STUDY · EPISODE 5

Can an AI agent find you?

Name: Agent discoverability of Romanian brands and agencies (ADO Score), 130 domains, September 2026
Published: 2026-09-10
License: https://creativecommons.org/licenses/by/4.0/

130 Romanian domains probed by machine — A2A Agent Cards, MCP discovery, llms.txt, schema.org, Wikidata. Zero Agent Cards, a mean ADO Score of 17.22/100, and the interoperability that does exist comes from Shopify and WordPress plugins, not from strategy. Original Websem study, raw data public under CC BY 4.0.

2026-09-10 · Dan Cristian Alexandrescu, Websem · ORCID

Until now, this series has measured one thing: when a person asks ChatGPT or Gemini "where do I get a good laptop", which brands come back in the answer. The person reads, the person chooses. But a second kind of customer is arriving. An AI agent receives an objective ("find a supplier of corporate jewelry for 200 gifts, under a given budget, with an invoice") and solves it on its own. It does not search Google and does not read marketing copy. It requests configuration files at standardised paths, checks structured data, cross-references sources and, if it cannot find what it needs in a format it can process, moves on to the next candidate. The first customer is persuaded; the second is verified.

This is the fifth episode in the series measuring how Romanian markets appear in front of artificial intelligence. The first three measured the human layer: jewelry, books, electronics. The fourth measured a technical artefact, llms.txt. This one measures the next layer, the agents', using the framework Exista.io published in February 2026 under the name AI Visibility Stack: whether an autonomous agent can find, evaluate and select a Romanian company without a human stepping in. The paper proposes a metric for this, the ADO Score, and makes a prediction: most companies will score near zero, not because they are weak, but because the necessary artefacts are not yet part of standard practice. We tested the prediction on Romania.

— IN SHORT

Zero Agent Cards. None of the 126 domains with an HTTP response serves /.well-known/agent-card.json — the dimension with the largest weight in the ADO Score, 30 points, is empty across the board.
The mean ADO Score is 17.22 out of 100, the maximum 35. 55 domains sit between 21 and 40, 61 between 1 and 20, 9 at exactly zero. Above 40 there is no one.
All the score comes from old SEO. Trust signals and sitemap freshness bring on average 12 of the 17 points; interoperability brings 0.3 of 15.
The interoperability that exists comes from the platform, not from strategy: of 8 domains with MCP discovery, 3 are Shopify stores, 4 run a WordPress plugin and only one built its own server.
Layer 1 transfers weakly to layer 2: the Spearman correlation between AI visibility (ep. 1–3) and ADO Score, across 41 brands, is 0.31. eMAG, the most visible brand in the series, scores 22.
Wikidata separates brands from agencies: 12 of 41 brands have an entity with an official website declared, against 3 of 85 agencies.
The closed door: 6 domains refuse a JavaScript-free request (403/503/406 or an anti-bot challenge), among them 4 of 13 electronics retailers the engines recommend and Amazon.de.

— THE NUMBERS

Across the 125 ranked domains (of 130 in the frame), measured on 2026-09-08, in a single run.
Dimension	Weight	Mean	Domains with points
Agent Card	30	0	0
Trust signals	25	8.19	102
Knowledge completeness	20	4.99	101
Interoperability	15	0.27	9
Freshness	10	3.76	89
ADO Score	100	17.22	116

45 of the 100 possible points — the two dimensions that explicitly concern agents, Agent Card and interoperability — produce on average 0.27 points. Every other point comes from three things a site has had for ten years for SEO reasons: structured markup, an entity in a knowledge graph and a dated sitemap.

— THE RANKING

Top 10 of 125 ranked domains. The rest, with every signal column, is in the CSV.
#	Domain	Group	ADO	Trust	Knowledge	Interop.	Freshness
1	IONA (iona.ro)	brand · bijuterii	35	13	10	4	8
2	TEILOR (teilor.ro)	brand · bijuterii	32.4	21	11.4	0	0
3	Litera (litera.ro)	brand · carte · editura	32.4	18	9.4	0	5
4	dafe.ro (dafe.ro)	agency · marketing-aeo	31	13	10	0	8
5	difrnt.ro (difrnt.ro)	agency · marketing-aeo	31	13	10	0	8
6	privatebrands.ro (privatebrands.ro)	agency · marketing-aeo	30.8	15	7.8	0	8
7	digitaliomarketing.ro (digitaliomarketing.ro)	agency · marketing-aeo	28	13	10	0	5
8	goai.ro (goai.ro)	agency · marketing-aeo	28	13	10	0	5
9	seocherry.ro (seocherry.ro)	agency · marketing-aeo	28	13	10	0	5
10	creativdigital.ro (creativdigital.ro)	agency · marketing-aeo	27.9	11	8.9	0	8

— WHAT THE DATA SHOWS

Old signals dominate; agent signals are missing

The older and more "SEO" a signal is, the more adopted it is; the newer and more "agent" it is, the more absent. HTTPS and robots.txt are above 94%. JSON-LD is at 73.02%. llms.txt, born in 2024, is at 53.97%, but with a brutal asymmetry between those who sell it (65.88%) and those it is sold to (29.27%). Wikidata is at 11.9%. And the two files defined by the agent protocols, from 2025, are at 6.35% and 0%.

Signal	All (n=126)	Brands	Agencies
A2A Agent Card	0 (0%)	0 (0%)	0 (0%)
MCP discovery	8 (6.35%)	3 (7.32%)	5 (5.88%)
Wikidata	15 (11.9%)	12 (29.27%)	3 (3.53%)
JSON-LD Organization	92 (73.02%)	24 (58.54%)	68 (80%)
llms.txt	68 (53.97%)	12 (29.27%)	56 (65.88%)
Sitemap ≤ 30 days	65 (51.59%)	20 (48.78%)	45 (52.94%)
robots.txt	119 (94.44%)	39 (95.12%)	80 (94.12%)

Interoperability comes from the platform, not from strategy

The 8 domains serving MCP discovery are the only place in the study where agent protocols appear in production. Three are Shopify stores (the platform ships the protocol, the merchant decided nothing), four run a WordPress plugin, and a single domain, seo365.ro, built its own endpoint.

iona.ro — brand · WordPress plugin
seo365.ro — agency · own server
istyle.ro — brand · Shopify
limitless.ro — agency · WordPress plugin
growwwise.com — agency · WordPress plugin
moogu.ro — brand · Shopify
gregoire.ro — agency · Shopify
arobsgrup.ro — agency · WordPress plugin

The agencies selling AI visibility do slightly better

Brands (ep. 1–3): mean score 15.89, maximum 35. Agencies (ep. 4): mean score 17.87, maximum 31. The agencies' advantage comes exclusively from llms.txt (65.88% vs 29.27%) and JSON-LD (80% vs 58.54%). On the expensive signals — those requiring an off-site presence — the brands lead clearly: 12 brands have Wikidata against 3 agencies.

Being recommended by AI to people ≠ findable by an agent

The Spearman correlation between AI-answer visibility (ep. 1–3) and ADO Score, across 41 brands, is 0.31 — positive, weak, far from “inheritance”. eMAG, named in 94% of the electronics answers, scores 22, because it has no organisation JSON-LD and no file for agents. They are two separate subjects, with two separate to-do lists.

Ten domains cannot be read by an agent

Six domains refuse the JavaScript-free request, four do not answer at all. Seven of ten are electronics retailers. The consequence is mechanical: a blocked site cannot receive points for JSON-LD, content or freshness. The opposite case is just as instructive: libris.ro answers 200 with a generic JSON (“Forbidden!”) on any path — a naive agent would read it as an Agent Card. Verifying the schema, not the response code, is the first rule of layer 2.

snsys.ro — HTTP 406
pcgarage.ro — HTTP 403
flanco.ro — HTTP 403
quickmobile.ro — HTTP 503
amazon.de — HTTP 202
vexio.ro — HTTP 403
altex.ro — no response
mediagalaxy.ro — no response
start-seo.ro — no response
daredigital.ro — no response

— METHODOLOGY

Theoretical framework. AI Visibility Stack / Agent Discovery Optimization — Marco, G. (2026), Beyond AEO: The AI Visibility Stack and the Era of Agent Discovery Optimization, Exista.io Working Paper, doi:10.5281/zenodo.18728629. The weights are the paper’s; the machine-verifiable scoring criteria are Websem’s, published in full in scor-ado.md.

The sampling frame. 130 domains: 43 brands named by AI engines in episodes 1–3 (jewelry, books, electronics) + 87 marketing agencies and sites cited by the engines in episode 4. It is not a representative sample of the Romanian economy — it is the most favourable possible sample for the paper's hypothesis that LLM visibility is inherited by agent discoverability.

The probe. 8 September 2026, a single pass, ~12 HTTP requests per domain, no JavaScript, with probe_ado.py (standard Python 3, published). A file counts only if it parses as JSON and contains the protocol's mandatory fields — a 200 response does not mean “served”.

Reproducibility. Every figure on this page comes from the raw data, computed with analiza_ado.py, which programmatically checks that the sum of the dimensions equals the score on every row. The scripts and the raw JSON are published below, under CC BY 4.0.

— THE DATA, DEPOSITED PUBLICLY

The dataset is deposited on Zenodo under DOI 10.5281/zenodo.22664431, licensed CC BY 4.0. The DOI always resolves to the latest version of the data.

Alexandrescu, D. C. (2026). Agent discoverability of Romanian brands and agencies (ADO Score). Websem. https://doi.org/10.5281/zenodo.22664431

DOI 10.5281/zenodo.22664431 →Zenodo →GitHub →Kaggle →Hugging Face →

Download the raw data (CSV, CC BY 4.0) ↓The B2A service — get your brand ready for agents →Free AEO audit →

— LIMITATIONS

A single pass, on a single day (8 September 2026). There is no volatility section.
The score's operationalisation is Websem's. The Exista.io paper gives the dimensions and weights, not the criteria — published in full in scor-ado.md.
Binary signals, no quality: a 200-byte llms.txt and a 60 KB one are “served” alike.
The sampling frame is not representative: 43 brands from three verticals and 87 marketing sites, not “Romanian companies” in general.
Ten domains without a full read — their scores are, by construction, incomplete.
Declared conflict of interest: websem.ro is in the frame and is reported, but excluded from the ranking and the means.

— FREQUENTLY ASKED QUESTIONS

What is the ADO Score?

A 0–100 metric proposed by Exista.io (Marco, 2026) for a company's discoverability by autonomous AI agents, from five dimensions: Agent Card (30), trust signals (25), knowledge completeness (20), interoperability (15), freshness (10). This study keeps the dimensions and weights and adds published, machine-verifiable scoring criteria.

Is this the official Exista.io score?

No. It is Websem's operationalisation of their framework, with its own transparent criteria. The paper does not publish scoring criteria, so any ADO Score computed by anyone else is, likewise, an operationalisation.

Why does no domain have an Agent Card?

Because the A2A protocol is a year and a half old, and the file is not generated by any popular website platform, unlike llms.txt or MCP discovery.

What is "MCP discovery" and why does it matter that it comes from Shopify?

/.well-known/oauth-protected-resource is the file through which an MCP client learns where to authorise. Shopify publishes it for every store; the merchant decided nothing. It shows that agentic interoperability, such as it exists, is an effect of the platform, not of strategy.

How were the 130 domains chosen?

We did not choose them. They are the brands named by the AI engines in episodes 1–3 and the domains cited as sources in episode 4, deduplicated. A single domain appears in two episodes (emag.ro).

Does a low score mean the company is weak?

No. It means the public artefacts an agent could verify do not exist. The largest companies in the study, eMAG or Amazon, are not at the top.

Why is websem.ro in the study?

Because the AI engines cited it in episode 4, and this study's sampling frame is everything the engines cited. We probed it with the same script and report it with its score (28.0), but it enters neither the ranking nor the means.

How can any figure in the study be verified?

With probe_ado.py and analiza_ado.py, published in the repository. The first runs the probe on domenii.csv, the second computes the scores from the raw JSON and checks programmatically that the sum of the dimensions equals the score, on every row.

— THE DATA, ROW BY ROW

All 130 domains in the frame, sorted by score. “—” means an unranked domain (no HTTP response, or websem.ro).
#	Domain	Group	ADO	Wikidata	JSON-LD	llms.txt	MCP	Status
1	iona.ro	brand	35	—	✓	✓	✓	200
2	teilor.ro	brand	32.4	✓	✓	✓	—	200
3	litera.ro	brand	32.4	✓	✓	—	—	200
4	dafe.ro	agency	31	—	✓	✓	—	200
5	difrnt.ro	agency	31	—	✓	✓	—	200
6	privatebrands.ro	agency	30.8	—	✓	✓	—	200
7	digitaliomarketing.ro	agency	28	—	✓	✓	—	200
8	goai.ro	agency	28	—	✓	✓	—	200
9	seocherry.ro	agency	28	—	✓	✓	—	200
—	websem.ro	agency	28	—	✓	✓	—	200
10	creativdigital.ro	agency	27.9	—	✓	✓	—	200
11	digitalrocket.ro	agency	27.9	—	✓	✓	—	200
12	dmaster.ro	agency	27.9	—	✓	✓	—	200
13	mahucode.ro	agency	27.9	—	✓	✓	—	200
14	natix.ro	agency	27.9	—	✓	✓	—	200
15	seo365.ro	agency	27.8	—	✓	✓	✓	200
16	okian.ro	brand	27.7	—	✓	✓	—	200
17	istyle.ro	brand	26.7	—	✓	✓	✓	200
18	editura-art.ro	brand	26.4	✓	✓	—	—	200
19	apexdigital.ro	agency	26	—	✓	✓	—	200
20	automatizez.ro	agency	25.9	—	✓	✓	—	200
21	code24.ro	agency	25.7	—	✓	✓	—	200
22	spacemedia.ro	agency	25.7	—	✓	✓	—	200
23	limitless.ro	agency	25.6	—	✓	✓	✓	200
24	growwwise.com	agency	25.4	—	✓	✓	✓	200
25	sabrini.ro	brand	24.2	✓	✓	—	—	200
26	edituracorint.ro	brand	23.8	—	✓	—	—	200
27	digitalcraft.ro	agency	23.4	—	✓	✓	—	200
28	roweb.ro	agency	23.4	—	✓	✓	—	200
29	smarters.ro	agency	23	—	✓	—	—	200
30	carturesti.ro	brand	23	✓	—	✓	—	200
31	stildiamonds.ro	brand	22.9	—	✓	✓	—	200
32	chatbotsite.ro	agency	22.7	—	✓	✓	—	200
33	emag.ro	brand	22	✓	—	—	—	200
34	digitalizare.ai	agency	21.9	—	✓	✓	—	200
35	dwf.ro	agency	21.9	—	✓	—	—	200
36	malvensky.com	brand	21.8	—	✓	—	—	200
37	evomag.ro	brand	21.7	—	✓	✓	—	200
38	dancovision.ro	agency	21.6	—	✓	✓	—	200
39	f64.ro	brand	21.6	—	✓	✓	—	200
40	seoagency.ro	agency	21.6	—	✓	—	—	200
41	targetweb.ro	agency	21.6	—	✓	✓	—	200
42	metaflow.ro	agency	21.6	—	✓	✓	—	200
43	seozilla.ro	agency	21.6	—	✓	✓	—	200
44	seo-ai.ro	agency	21.4	—	✓	✓	—	200
45	arcsoft.ro	agency	20.7	—	✓	—	—	200
46	aysa.ro	agency	20.7	—	✓	✓	—	200
47	flowmatic.ro	agency	20.7	—	✓	—	—	200
48	sanbi.ai	agency	20.7	—	✓	✓	—	200
49	aiautomatizari.ro	agency	20.7	—	✓	✓	—	200
50	canopy.ro	agency	20.4	—	✓	—	—	200
51	fgd.ro	agency	20.4	—	✓	✓	—	200
52	marketinglab.ro	agency	20.4	—	✓	✓	—	200
53	brandscan.ro	agency	20.4	—	✓	✓	—	200
54	outglow.ro	agency	20.4	—	✓	✓	—	200
55	moogu.ro	brand	20.3	—	✓	✓	✓	200
56	snsys.ro	agency	20	✓	—	—	—	406
57	cosmicweb.ro	agency	19.9	—	✓	—	—	200
58	mindloop.ro	agency	19.9	—	✓	—	—	200
59	nexaiagency.ro	agency	19.9	—	✓	✓	—	200
60	targulcartii.ro	brand	19.9	—	✓	—	—	200
61	ewdigital.ro	agency	19.9	—	✓	✓	—	200
62	vivinet.ro	agency	19.9	—	✓	✓	—	200
63	mda-digital.com	agency	19.7	—	✓	✓	—	200
64	nextchapter.ro	agency	19.7	—	✓	✓	—	200
65	geoflux.ai	agency	19.6	—	✓	✓	—	200
66	themarkers.ro	agency	19.3	—	✓	✓	—	200
67	gregoire.ro	agency	19.2	—	✓	✓	✓	200
68	aifactory.ro	agency	18.8	—	✓	✓	—	200
69	pcgarage.ro	brand	18	✓	—	—	—	403
70	cellini.ro	brand	17.4	—	✓	—	—	200
71	groway.ro	agency	17.4	—	✓	✓	—	200
72	nion.ro	agency	17.4	—	✓	✓	—	200
73	nemira.ro	brand	17	✓	—	—	—	200
—	altex.ro	brand	17	✓	—	—	—	—
74	libris.ro	brand	16.6	—	✓	—	—	200
75	optimizareai.online	agency	16.6	—	✓	✓	—	200
76	marketiu.ro	agency	16.4	—	✓	✓	—	200
77	agentiiseo.ro	agency	16.3	—	✓	—	—	200
78	bookzone.ro	brand	16.3	—	✓	—	—	200
79	klain.ro	agency	16.3	—	✓	✓	—	200
80	marketos.ro	agency	16.3	—	✓	✓	—	200
81	edituratrei.ro	brand	16	✓	—	—	—	200
82	flanco.ro	brand	16	✓	—	—	—	403
83	bluedotfusion.ro	agency	15.8	—	✓	✓	—	200
84	humanitas.ro	brand	15	✓	—	✓	—	200
85	aiengineoptim.ro	agency	14.3	—	✓	—	—	200
86	apolloboostmedia.ro	agency	14.3	—	✓	—	—	200
87	librarul.ro	brand	14.2	—	✓	✓	—	200
—	mediagalaxy.ro	brand	14	✓	—	—	—	—
88	aifrontdesk.ro	agency	13.4	—	✓	—	—	200
89	blogdigital.ro	agency	13.2	—	✓	—	—	200
90	adsem.ro	agency	13.2	—	✓	—	—	200
91	quickmobile.ro	brand	13	✓	—	—	—	503
92	curteaveche.ro	brand	12.4	—	✓	—	—	200
93	editura-arthur.ro	brand	12.4	—	✓	—	—	200
94	books-express.ro	brand	11.3	—	✓	—	—	200
95	itgalaxy.ro	brand	11.3	—	✓	—	—	200
96	polirom.ro	brand	11.3	—	✓	—	—	200
97	sabion.ro	brand	11.3	—	✓	—	—	200
98	upswing.ro	agency	11.3	—	✓	—	—	200
99	gomag.ro	agency	11	✓	—	—	—	200
100	iqads.ro	agency	11	✓	—	—	—	200
101	arobsgrup.ro	agency	9	—	—	✓	✓	200
102	agentii-seo.ro	agency	8.2	—	✓	—	—	200
103	agentiadeai.ro	agency	8	—	—	✓	—	200
104	automatizaricuai.ro	agency	6	—	✓	✓	—	200
105	agentiegeo.ro	agency	5	—	—	✓	—	200
106	coriolan.ro	brand	5	—	—	—	—	200
107	edituraaramis.ro	brand	5	—	—	—	—	200
108	elefant.ro	brand	5	—	—	—	—	200
109	iagency.ro	agency	5	—	—	—	—	200
110	raobooks.com	brand	5	—	—	—	—	200
111	sostenia.ro	agency	5	—	—	—	—	200
112	ai4business.ro	agency	3	—	—	—	—	200
113	aientity.ro	agency	3	—	—	—	—	200
114	instapress.ro	agency	3	—	—	—	—	200
115	pionmedia.ro	agency	3	—	—	—	—	200
116	sphinx-it.ro	agency	3	—	—	✓	—	200
117	amazon.de	brand	0	—	—	—	—	202
118	autoads.ro	agency	0	—	—	—	—	200
119	cel.ro	brand	0	—	—	—	—	200
120	donamarketing.ro	agency	0	—	—	—	—	200
121	ksd.ro	agency	0	—	—	—	—	200
122	kultho.ro	brand	0	—	—	—	—	200
123	librariaromana.ro	brand	0	—	—	—	—	200
124	remotemarketing.ro	agency	0	—	—	—	—	200
125	vexio.ro	brand	0	—	—	✓	—	403
—	start-seo.ro	agency	0	—	—	—	—	—
—	daredigital.ro	agency	0	—	—	—	—	—

— AUTHOR

Dan Cristian AlexandrescuFounder & General Manager, Websem

Dan Cristian Alexandrescu is the founder of Websem and researches how brands surface inside generative engines. He publishes AEO/GEO studies on Romanian markets, built on data collected across ten large language models and archived on Zenodo and Kaggle.

Author page →ORCID 0009-0007-1994-6700 ↗