Why do we need AI-powered Search?
As your course library grows, so does the amount of valuable knowledge buried inside it: policies, handbooks, process training, compliance material. But knowledge that can't be found when it's needed is knowledge that doesn't get used.
Until now, search in Learnster U only matched on course and resource titles, if a learner didn't know the exact words a course was named, their only option was to browse page by page, or ask someone else.
AI-powered search changes that. Learners can now ask a question in their own words, for example "how do I handle a customer refund?", and get a direct answer, summarized by AI with references to the exact course pages it's based on, along with the relevant excerpts from the course content they already have access to. They don't need to know which course the answer is in, or even that it exists.
This is what turns a course from something a learner completes once and moves on from, into a knowledge base they actually come back to, when they're handling a real situation, making a decision, or just need a reminder of how something works. Making this available for your learners is one of the most direct ways to make the content you've already built in Learnster genuinely useful, day to day.
👇 AI-powered Search is available through the search bar in Learnster U, once the user has typed a search query
How it works
AI-powered Search is available through the search bar in Learnster U.
Learners can type a question or describe what they're looking for in their own words. AI interprets the meaning behind the question and matches it to the excerpts of course content whose context fits best - so results stay relevant even when the exact search wording doesn't match
AI search results so called "Context matches" show short excerpts from the matching content, along with the course they belong to, so learners can find the information they are looking for directly from the search result modal.
Learners also get an AI Summary upon pressing Enter, a direct answer to their search with clickable references to the source pages.
Learners can also jump straight to the relevant spot in the course page from the search result or the source reference in the summary, to quickly find the context of search result.
Search results always respect existing course permissions - learners only ever see content from courses they have access to.
How AI-powered Search finds the right information
AI Search results, labeled as "Context matches" are powered by a technique called RAG (Retrieval-Augmented Generation). Here is a simplified version of how it works:
AI indexes your entire course library, splitting the content into smaller excerpts and analyzing each one to understand what it's actually about.
When a learner types a search query, AI interprets the meaning behind their question — not just the words they used — and compares it against those indexed excerpts to find the ones that match best.
The AI layer allows users to type search queries that are not purely keyword based, instead learners can describe what they are looking for or ask questions- AI will understand!
The best-matching excerpts are what show up as Context matches, each linked straight back to the right course and page.
Right now, indexing covers text-based course content only — course pages, descriptions, and syllabi. We're already looking into extending this to other content types, such as resources, video transcripts, and PDFs, so more of your content becomes searchable over time.
The Context matches are then used to generate an AI Summary, a synthesized answer with references back to the source pages (see AI Summary below).
This AI-powered indexing happens automatically and repeats every time your course content is updated, so new or edited content becomes searchable without any extra work on your end.
ℹ️ This is also why enabling AI-powered search for the first time can take a little while: AI needs to index your entire course library before Context matches start appearing, and the more text content your catalog contains, the longer that initial indexing takes — typically a few hours.
AI Summary: a direct answer, with sources
Context matches help learners find the right excerpt. AI Summary takes it one step further. Instead of reading through each excerpt, learners get a direct, synthesized answer to their search, with references that link to the exact course pages the answer is based on.
How it works
AI Summary is generated automatically when a learner submits a search, by pressing Enter.
The summary is written from the excerpts the search has already retrieved, meaning the same Context matches shown in the results.
If the learner asks a question, for example "how do I handle a customer refund?", AI answers the question directly.
If the learner searches for a term or topic, for example "refund policy", AI gives an overview summary of the relevant excerpts.
Source references are embedded in the text. Clicking a reference takes the learner straight to the course page the information comes from, so they can check the full context.
If nothing relevant is found, the summary tells the learner that the content doesn't contain any information related to their search. That way they know to rephrase their search, or that the answer may not exist in the course library yet.
AI Summary is based only on content the learner has access to. The same permission rules that apply to Context matches apply here.
⚠️ AI can make mistakes. Every summary is followed by the notice "AI can make mistakes. Double-check by visiting the content source." We recommend that learners use the embedded references to verify important information, especially for policies, compliance and safety-related content.
ℹ️ Because AI Summary is built from Context matches, it works best with full questions or descriptive phrases. Very short or single-word searches often don't give AI enough context to write a helpful summary. See the table below for what to expect.
Verbatim results, Context matches and AI summary: what to expect
When a learner opens search in Learnster U, they can type a question or a phrase in plain language. Search could return two types of results side by side:
Verbatim matches - traditional keyword matches on course and resource titles, just like before.
Context matches - excerpts that are matched by AI, these surface relevant excerpts from inside course content, even if the learner's words don't exactly match the text.
Each result shows:
The course name the excerpt belongs to
A short snippet of the matching text, with the option to expand and read more
A link that jumps the learner directly to that part of the page — no need to browse through the whole course to find it
These two result types work differently, and each is better suited to a different kind of search:
Verbatim results look for an exact match between the learner's search words and course or resource titles. They're fast and work well even for very short searches — for example, searching just "GDPR" reliably surfaces a course titled "GDPR Policy." AI-based Context matches, on the other hand, need enough words to understand what the learner means, so it might not return relevant results for single-word or very short searches.
Context matches look for meaning rather than exact wording, searching inside course content (description, syllabus, page content) rather than just titles. This is what makes natural-language questions like "how do I handle a customer refund?" work, even if no course title contains those exact words.
ℹ️ Context matches are best when a learner is describing what they're looking for or asking a full question, rather than typing a single keyword.
Here's roughly what to expect depending on how a learner searches:
A learner searches... | Verbatim results | Context matches (AI-powered Search) | AI summary |
A single word or short phrase (e.g., "GDPR") | Likely to appear, if it matches a title | May not give accurate matches, might be too short for AI to interpret meaning | Often too general to be helpful, since AI lacks the context to know what the learner is after |
An exact course or resource name | Appears reliably | May also appear if the context is presented in course description, syllabus or page content | Gives an overview of matching excerpts |
A full question (e.g., "how do I handle a customer refund?") | Unlikely. Result appear if a keyword match is present | Most accurate — AI interprets the meaning and surfaces matching excerpts | Most helpful. AI answers the question directly, with references to the source pages |
A topic described in different words than the course uses (e.g., "money back" instead of "refund") | Won't match unless those exact words appear in a title | Most accurate — AI interprets the meaning and surfaces matching excerpts | Gives a summary of the relevant excerpts, even though the wording differs |
Something the course content doesn't cover | No results | No relevant matches | States that the content doesn't contain information related to the search |
Together, the three give learners a good chance of finding what they need, however they choose to phrase it. Verbatim results catch quick, exact lookups. Context matches surface relevant content that a keyword search would never find. AI Summary turns those matches into a direct answer, with references so learners can check the source.
Search results respect permissions
Learners only ever see search results from content they're allowed to access:
Courses they're enrolled in or assigned to → full content excerpts are shown.
Courses visible in their catalog but not yet enrolled in → only the course description and syllabus are shown as an excerpt in the "Context matches" search results and in the summary. The learner is prompted to enroll to unlock the full content.
Courses outside a learner's access are never surfaced, regardless of query.
Enabling AI-powered search in your platform (for Super admins)
AI-powered search and summary is available to Core-plan customers and is off by default.
Super Admins can enable it in following steps:
Go to Studio → Settings → General → Search in Learnster U
Turn on the AI-powered search toggle. This will also enable AI summary.
ℹ️ Important information:
After enabling AI-powered Search setting, a delay of few hours is expected before Context matches start appearing while your content is indexed for the first time. The AI-powered system needs to index your entire course library before Context matches start appearing, and the more text content your catalog contains, the longer that initial indexing takes, typically a few hours.
If you turn off AI-powered search, search in Learnster U reverts to the previous verbatim search
Tips for content creators: making your content easier to find
A few content habits make a real difference in how well your courses surface in AI-powered search:
Write clear, descriptive course descriptions and syllabi. This matters more than it might seem: for courses that are visible in a learner's catalog but that they haven't enrolled in yet, only the course description and syllabus are searchable — not the full page content. A vague or empty description means that course effectively won't surface in search for those learners, even if the content inside it is exactly what they need.
Use clear, descriptive course titles. Verbatim results match on titles, so a title that reflects how learners would actually describe the topic — rather than an internal or abbreviated name — is more likely to surface in a quick keyword search.
Write page content in full, natural sentences where possible. Context matches work by understanding the meaning of your content, so pages written as complete thoughts tend to match learner questions more reliably than sparse bullet fragments alone.
Known limitations
AI-powered Search covers text-based course content only: course pages, descriptions, and syllabi. We're already looking into extending this to other content types, such as resources, video transcripts, and PDFs, so more of your content becomes searchable over time.
Very short or single-word searches are better served by Verbatim results; Context matches and Summary work best with fuller questions or phrases.
Verbatim and Context matches are displayed in separate lists, but we are looking into merging the two lists for a better user experience.
User permissions are applied to content searchable in the Context matches. We are looking into making this more flexible so that you can choose more freely if a course should be available in the Context match search results regardless of permissions.
AI Summary can make mistakes. Learners should verify important information using the embedded references to the source pages.
What's next
We're continuing to improve AI-powered Search based on feedback from our users. Coming soon: AI-powered Search and AI Summary in the Learnster U app, so learners can get answers on the go.
Need more help?
You can reach out to our support via chat directly from Learnster!