Vega

Vega is the way you ask questions about your library data without building a query. You type the question in plain English; Vega writes the query, runs it, and shows you the answer.

What Vega is

A chat surface inside the staff app. You ask “What did the 4th grade check out most often in October,” Vega answers with a ranked breakdown of titles and their checkout counts. You ask “Do we own Holes by Louis Sachar,” Vega looks it up in the catalog and tells you. You ask “How do I set a loan policy,” Vega answers from this documentation site.

Most answers come back as a card you can act on — export it to a spreadsheet, save a search to an OPAC list, or check off items to weed or buy. See Working with results for what each card lets you do.

Getting to Vega

In the staff app, go to Library level navigation path: Vega in the sidebar. Vega is a library-centric tool. If it isn’t there, an account owner has turned Vega off for the whole account.

A running conversation keeps context across questions, so you can ask follow-ups like “now narrow that to books due before Halloween” after the first answer lands.

What to ask it

Examples, grouped by the kind of question. These are the shape of question Vega does well.

Circulation

Questions about loans, checkouts, and what to weed.

“What did 5th grade check out last month?”
“Show me books that haven’t been checked out in two years.”
“How many checkouts did we do during the week before spring break?”
“Which items are good candidates for weeding?”

Vega’s circulation answers aggregate loan-based metrics (counts, top titles, top topics, dormant items), and over a time series it can also count holds created and holds filled — so “were we taking more holds in October than September” is a fair question. What it won’t do is bucket anything by homeroom, or break circulation down by overdue status: there’s no “overdues by grade” answer from Vega, so use Active Loans for that.

By-grade answers cover only the grades your school serves. A student whose class year works out to a grade outside that range (a typo in the roster, say) isn’t counted in any grade.

Catalog

Questions about what’s in the catalog, what you already own, and what to acquire.

“What books do we have about climate change?”
“Do we own Holes by Louis Sachar?”
“Which Lexile range do we have the most books in?”
“Suggest some books I should buy on the topic of dragons.”

Using Lyra

Questions about how to do something in Lyra. Vega answers from this documentation site.

“How do I set a loan policy?”
“Where do I add a new staff user?”
“How do I create a grade-level override?”

What Vega will not do, and why

Vega never sees your patron records.

When you ask Vega a question, Lyra runs the query against your data and gives Vega the results to summarize. Before Vega sees those results, Lyra strips patron names, emails, barcodes, and identifiers.

Your own question is checked before it is sent. To ask about a particular patron or homeroom, type @ and pick them from the list: they appear as a chip, and Vega reads the chip as a stand-in, never the name. If you type a patron’s or homeroom’s name instead, Lyra holds the question and offers button: Mention Jonah with @ for each name it found, or button: Send as typed when the word only looks like a name. See Privacy & trust.

A question about a mentioned patron is a question about their loans or holds. “Does @Jonah have a hold ready?” counts that patron’s ready holds, and because you picked the patron, Vega gets the exact number and can answer yes or no. It still never learns who Jonah is. The card it renders opens All Holds with a button: Patron chip for them and Stage set to Ready. “Is @Jonah overdue?” and “What books does @Jonah have checked out?” open Active Loans the same way: Vega tells you how many, and the card shows which titles, since Vega never sees them. A mention stays the same chip for the rest of the conversation, so a follow-up can name the same patron.

A question about a title’s loans or holds works the same way, without a mention: “Is Wonder checked out?” or “Who is waiting for Wonder?” counts the loans or holds on the titles in your catalog that match, in one step, and Vega tells you which titles it counted. The card opens Active Loans or All Holds with an button: Item chip for those titles, where you see who has it or who is waiting.

The consequence: Vega can answer questions about your collection and circulation in aggregate (counts, top titles, by grade) but cannot identify or describe an individual student.

It can still answer questions about patrons, by counting them without seeing them. “Which 4th graders have a book more than two weeks overdue?” gets a count, and a card that opens the Patrons list with those filters, where you see who. Vega is given the number and the filters it asked with, never a name or a row, and a count of 1 to 4 reaches it only as “fewer than 5”. A count of zero it is told, since it points at no one. A count over patrons you @-mentioned is exact, since you chose them and Vega knows them only by a stand-in; a mentioned homeroom is a group, and keeps the floor. Counting patrons and loans needs a Standard or Admin role, the same line as seeing contact details and checkout history. When you ask for something Vega can’t surface but the staff app can, it tells you why and points you there.

For the full picture of what’s sent to the model and who can see your conversations, see Privacy & trust.

Things Vega will refuse

A few categories Vega is built to decline:

Identifying a specific student. “Who is overdue?” gets a count and a card that opens the list for you. Vega won’t return a name. Naming a student yourself doesn’t change the answer — but it does send that name to the model, so don’t.
Drafting communications. “Write an email to the parents of the kids who are overdue” is outside scope. Use the notices system or write the email yourself.
Acting on data by itself. Vega answers questions; it won’t run checkouts, edit records, place holds, delete anything, or change settings on its own. It can hand a result to you to act on — download it, or save a search to a list — but those are buttons you click, never something Vega does unprompted. See Working with results.
Topics outside the library. Asking Vega for help with your tax return, your lesson plan, or the weather will get a polite redirect.

If Vega declines, the response says why — and when the answer lives somewhere in the staff app, it points you to the page that has it instead of dead-ending.

If Vega can’t answer

Sometimes a question doesn’t get through — the connection drops, or Vega is briefly unavailable. Your question stays in the conversation, along with any part of the answer that had already arrived, and message: Vega couldn't answer. Try asking again. appears under it. You can ask again straight away.

A question that failed this way may not have been saved, and can be gone once you leave the page or reload — copy it first if you want to keep it.

When Vega runs into a problem partway through its own answer, it says what went wrong under the part it finished instead. That answer is saved with the rest of the conversation.

Rate an answer

Under each of Vega’s answers are two buttons, button: Helpful and button: Not helpful. Select one to rate the answer; select it again to take the rating back. A box opens where you can say what worked or what went wrong, and button: Send feedback sends it. The comment is optional, and button: Add comment or button: Edit comment beside the buttons opens it again later. Leave out student names.

Only you can rate the answers in your own conversations. A rating is stored with the answer it judged, so the team can see exactly which question, and which search, it was about.