We help businesses get cited when people ask AI.
Search is fragmenting across ChatGPT, Perplexity, and Google AI Overview. A small number of sources get pulled into the answer. The rest are invisible. We do the structural work that puts you in the cited set.
Score your own site in 30 seconds. No email, runs live.
Same engine our paid analysis runs on. Score, gap list, and inline card checkout — all in English.
What AEO actually is
AEO is the engineering side of being mentioned by AI assistants. Schema.org markup, machine-readable metadata, canonical entity definitions, consistent business identity across data sources, content that maps cleanly to how questions are asked.
It is roughly to AI assistants what SEO was to Google in 2005: a set of structural fixes that take a site from "the algorithm cannot tell what this is" to "the algorithm has enough signal to cite it confidently."
We are a small team, based in Stavanger, Norway. Most of our day-to-day is Norwegian SMB work. The methodology is language-agnostic, and we have shipped foundations for English-language SaaS clients on the same basis.
How we work
- Audit. We score the site (0–100), test what AI assistants currently say about you when asked, and write up the gaps.
- Foundation. Schema, content structure, machine-readable metadata, canonical entity, business profile, llms.txt, agent-card.
- Monthly monitoring. Visibility tracking, content updates, and quarterly review of what shifted in the AI assistants you care about.
The Sitert package
The measurement is free. This is the fix. Five buying-intent questions from your market, each run at least nine times over several days with byte-identical prompts — then the changes written out for every page that blocks citation (schema, answer blocks, titles, copy), paste-ready or as a pull request, plus a re-measurement after 30 days with the same questions and the same method. You get the before-and-after numbers, the names of who gets cited instead, and the raw data so you can re-run it yourself. If a high-priority item is not delivered as promised, we redo the work without a new invoice. Invoiced, 14-day terms.
Order the Sitert package — NOK 9,900 →AEO Lite
FAQPage JSON-LD for one page you choose (3–5 Q&A pairs). Ready to paste into <head>. Delivered by email within one business day. One-off, no follow-up.
Stripe charges in USD on card. Norwegian-registered company — NOK invoice available on request: email hei@synligdigital.no.
One case in public (related party — disclosed)
Nordic Lithium, a Norwegian battery materials retailer, was selected as Synlig's first internal AEO implementation. Score went from 67 (grade C) to 90 (grade A) over roughly two weeks of structured work. Related-party disclosure: Nordic Lithium and Synlig share common ownership (Håkon Åmdal). This is a documented methodology case, not an arm's-length customer engagement — full ownership disclosure on the case page.
The methodology is the same in any language. Our two published cases (Nordic Lithium and Pierstop) are both related-party implementations — chosen as methodology baselines because internal access let us run clean before/after audits without external coordination. The schema and structural work transfers cleanly to arm's-length customers; the first such case is in progress.
If you want to talk
The fastest way is email: hei@synligdigital.no. Mention what your business does and the URL. We reply within one working day, usually faster.
If you want your own AEO score first, don't email us — run it yourself at the top of this page. Same engine, in English, about 30 seconds, no email required: a 0–100 score and the structural gaps we found. It measures whether your pages are machine-readable. It does not tell you whether AI actually cites you — that needs its own measurement, and the changes are what you pay for.
About the tool you may have come from. LLM Smell Check is a small open-source thing we put together because we kept seeing AI-shaped marketing copy in the wild and wanted a way to score it. It runs entirely in your browser, nothing is uploaded. Code is on GitHub; patches welcome.