How eSkill’s AI-powered search works — and how your content and privacy are protected.
Linea Search is the search technology behind the eSkill question library. When you type a search, it finds the most relevant questions from eSkill’s library of approximately 60,000 professionally developed questions and from your own authored content — matching not just the words you typed, but what you meant. This FAQ explains how the technology works, what happens to your content, and how we protect your privacy.
Linea Search runs every query two ways at once. A keyword search matches the literal words you typed, giving extra weight to fields like subject and topic names, so precise terminology still wins where it should. In parallel, an AI-powered semantic search understands what your query means, so it can surface relevant questions even when they don’t share your exact vocabulary — a search for “spreadsheet formulas” can find an Excel question that never uses the word “spreadsheet.” The two result lists are then merged so that questions both approaches agree on rise to the top, while a strong match from either one still gets its shot.
No content is generated. The AI used in Linea Search is an embedding model — a model trained to represent the meaning of a piece of text as a set of numbers, so that texts about similar things end up numerically close together. Finding relevant questions becomes a matter of finding the questions whose meaning sits closest to your query’s meaning. The AI never writes, edits, or alters questions; it only helps find and ranks the questions that already exist in the library. This embedding model is a private model and runs entirely within the eSkill platform.
The search index is updated continuously, not on a nightly schedule. When a question is added, edited, or removed — including your own authored questions — the change is reflected in search shortly after it is made. Incomplete or retired questions are kept out of results, so you won’t land on broken content.
Some subjects, like the Microsoft Office family, accumulate many legacy versions over the years. To keep results useful, older versions are demoted in generic searches — a search for “excel” returns current content by default. The moment you name a version (for example “excel 2010”), the older versions come right back. You get the current version by default and the specific one on demand.
Yes. Searches can be filtered by language, question type, and difficulty, and the filter options you see reflect what exists in the content available to you — not a static list.
Only you. Every search is scoped to your account: results combine eSkill’s shared library, your own authored questions, and content from your reseller’s master account where applicable — the same content you can already see when browsing the library by hand. One customer’s private content never surfaces to another customer.
When you author a question, it is added to the search index along with its subject, topic, and answers, and an AI-computed representation of its meaning is created, so the question can be found by semantic search from the moment it lands. This processing happens entirely within eSkill’s own application environment and exists for one purpose: making your own content findable by your own team.
No. The AI model behind Linea Search is a private model developed by eSkill. It is fully contained within our application environment and does not communicate with external AI models or third-party AI services. Your content never leaves the eSkill platform to be processed by an outside AI provider.
No. Your questions are passed through eSkill’s already-trained private embedding model to compute their search representation — the model reads your content to index it, it does not learn from it. Because the model is developed and operated by eSkill inside our own environment, your content is never used to train eSkill’s or anyone else’s AI models.
Yes. AI processing of your authored content is opt-in — your questions are only indexed for semantic search if you choose to enable it. If you don't opt in, your questions are simply excluded from indexing and won't appear in search results, though they remain fully usable in your assessments.
No. Linea Search operates only on question content — question text, answers, subjects, and topics. It never touches candidate information, responses, or scores.
Search keeps working. If the AI semantic service is slow or unavailable, Linea Search automatically falls back to keyword search for that request — you still get results, matched on the words you typed. Search is designed so that no single component can take it down.
Every search request is authenticated, and results are strictly scoped to the requesting account. The service is also built with safeguards against slow or degraded dependencies, so a problem in one backend component can’t delay your request.
Have a question we didn’t cover? Contact your eSkill account representative or support team — we’re glad to walk through how Linea Search works in as much detail as you need.
