Blueprints

A Blueprint is a ready-made package of finished work for one job, such as an FP&A close or a SaaS revenue review. You install it into a workspace, and Querri builds its views, KPIs, monitors, dashboards and sites around your data.

Who can see this: people whose role includes the Blueprints permission. By default only the Admin role has it; an admin can turn it on for other roles. It also needs Library access. Installing into a workspace needs creator access to that workspace. A Blueprint that includes a dashboard, site, skill or project also needs your role to allow creating that kind of item.

Why use a Blueprint

Building a finance P&L or a retention analysis from nothing means deciding every definition yourself: which accounts count as revenue, how to translate currencies, what a churned customer is. A Blueprint arrives with those decisions already made, written down and tested.

Tested. Each Blueprint’s tables come with the SQL that builds them and a set of checks that prove the numbers add up. A Blueprint has to pass those checks on its own sample data before it’s published.
Versioned. Every Blueprint has a version number, like 1.2.0. A published version never changes.
Yours afterwards. Everything it builds is an ordinary item in your workspace: a view in your Library, a KPI, a dashboard. You can open, edit and share them like anything else.

Browse the gallery

Click Blueprints in the rail to open the gallery.

Search by name, workflow or question, or filter by Industry and Department. Clear filters shows everything again.
A card marked Live install can be installed into a workspace. Other cards are previews you can explore but not install yet.
Explore blueprint opens its detail page: what it answers, what it includes, and the data it needs.

On the detail page:

Explore with sample data shows the Blueprint’s views, dashboard and sites filled with a fictional company’s numbers, so you can see what you’d get before you connect anything.
Make it yours lists every piece of the Blueprint. Click a card to include it or leave it out. A piece that another piece needs stays on with it, and the card’s Requires list shows why.
Data you’ll need lists the data the pieces you’ve picked will read. If you leave out the pieces that need a source, you won’t be asked for it.

Install into a workspace

Choose the pieces you want under Make it yours.
Optional: on a piece, add Your build instructions to tell Querri how to fit it to your team before it’s built, such as “Group results by location and week.” See What the AI does for when these are used.
Pick the workspace under Choose a workspace….
Optional: turn on Start with example data. This adds the Blueprint’s example files to the workspace as sources, named as examples, so you can try it before you connect your own systems. If the workspace already has data that fits, the examples sit alongside it and this installation builds over the examples.
Click Install into workspace.

The installation runs in the background. You can leave the page and come back.

Everything is installed unpublished. A site it builds isn’t shared with anyone until you publish it.

How your data is matched to the Blueprint

Each Blueprint describes the data it needs as a set of inputs. An input lists the columns it expects, the type of each one, and the common names it goes by in different systems. For example, the FP&A Blueprint’s general ledger input accepts an entity column called subsidiary, company or company_code.

Querri compares the sources in your workspace with those inputs. A source fits an input when it has every column the input requires, under one of its expected names. You don’t map the columns by hand.

If you haven’t connected or uploaded the data yet, install anyway. The Connector agent helps you connect or upload what’s missing after you install, and the pieces that need it wait. When new data arrives that fits, Querri picks the installation up again and builds what was waiting. That’s what New data restarts unfinished artifacts automatically means on the installer.

Before it builds anything, Querri reads the Data Audit notes on your sources. Read them too: a Blueprint can only be as right as the data you give it.

What gets created

Depending on the Blueprint and the pieces you choose, an installation can create:

Item	What it is
A collection and its questions	The questions the Blueprint answers, in your Library
Views	The governed tables every other piece reads
KPIs	Confirmed metrics, measured from an installed view
Monitors	Watches that tell you when a number moves away from its usual level
Dashboards	Dashboards over the installed views
Skills	Instructions the AI follows when it answers questions in this area
Projects and sites	Analysis projects, and multi-page sites for a team to read

A KPI is installed as a confirmed metric with no AI involved. Its definition history records version 1 as coming from the Blueprint, so you can always see what you started from. A KPI measured by fiscal quarter or fiscal year shows Partially built until your organization sets a fiscal calendar, with the reason “Set it in Settings → Customize → Fiscal calendar, then retry.” Set it, then retry the installation.

A monitor installed by a Blueprint checks after its view refreshes, and has no email or other external delivery turned on. Choose who gets its alerts yourself, as on Monitors and alerts.

Exact install or AI-adapted

Most Blueprint views ship with reviewed SQL: the exact query that builds the table. How that view is installed depends on whether your data really fits.

Exact install, no AI. Querri starts from the assumption that your data doesn’t quite match. It installs the view exactly as written only when it can prove the fit:

every table the SQL reads maps to exactly one of your sources, or to a view the Blueprint already built;
every column it reads exists, with the expected type, and isn’t empty;
every value the SQL looks for (such as Budget or Forecast) actually appears in your data;
every join between tables actually matches;
no open Data Audit note affects what it reads, except notes the Blueprint declares as expected. For example, ledger amounts are meant to be signed, so a note that flags negative amounts on a general ledger doesn’t block the FP&A Blueprint. Once the view is ready, Querri marks each such note acknowledged, with the reason (“Expected for gl-detail by the FP&A close and plan blueprint”). A note you already dismissed or acknowledged keeps your decision;
the SQL runs, returns rows, keeps one row per key, and produces the planned columns.

Familiar column names or a known connector aren’t enough on their own. If any check can’t be run, it counts as a failure.

When the fit holds, the reviewed SQL becomes the view’s definition and no AI writes or changes anything. The view’s status reads “Built from the blueprint’s reference SQL and validated in the View.” Your build instructions for that view aren’t used, because the view is installed as written.

AI-adapted. When the fit doesn’t hold, which is the common case, Querri’s AI starts from the reviewed SQL plus a list of the exact mismatches it found (a column called something else, a value spelled differently) and edits the SQL to fit your data. Your build instructions are used here. The view’s status reads “Data audited, cleaned, and validated in the View.”

Either way, the finished view must have the planned columns and pass the Blueprint’s checks below. Building a view uses credits whichever way it’s built. AI work, like adapting a view or building a dashboard or site, uses more. See Credits.

Checks on every install

Shape checks can’t see a flipped sign or a double-counted account, so most views also ship with reconciliation checks: tests that compare the view’s numbers with the data it came from, or with the views it depends on. For example, the FP&A Blueprint checks that each entity’s postings net to zero every month, and that the P&L’s revenue equals the ledger’s revenue with the sign flipped once.

A check passes when it finds nothing wrong. Each one runs on your installed data, not the sample.

When a check fails:

On an exact install, the Blueprint’s own SQL is wrong for your data. The view shows Needs attention with the reason, ending “The next attempt adapts the blueprint’s SQL to this data.” The installer then tries again on its own (each piece gets two attempts per run), and that second attempt takes the AI-adapted path, giving the AI the failing check and the rows that broke it.
On an AI-adapted view, Querri makes one repair attempt with the failing checks. If the view still fails, it’s kept but marked Partially built, with the check that failed and how many rows broke it, so you can open it and see.

A check that can’t run counts as failed, not passed.

Follow the installation

The installer shows each piece with its status, such as “3 of 9 ready”:

Status	Meaning
Planned, Queued	Waiting its turn
Building	Being built now
Ready	Built and checked. Click to open it.
Partially built	Built, but something’s missing or a check failed. The reason is shown.
Waiting on prerequisite	Needs another piece, or data, that isn’t ready yet
Needs attention	The build failed. The reason is shown.
Waiting for credits	Your organization is out of credits for this build
Needs access	You don’t have access to something it needs
Not included	You left this piece out

Pieces are built in order, so a dashboard waits for the views it reads. A piece that’s ready is never rebuilt by a later attempt.

Watch the installer work opens Installer activity: a live log of what the installer and its agents are doing, piece by piece, including earlier attempts. Only the person who started an installation can see it and its activity.

What to do next lists anything that needs you: data to upload or connect, an item to open and fix, or a link to billing if credits ran out.

Retry

Retry installation (or Retry all under What to do next) tries every unfinished piece again. Use it after you fix the data or add credits. Each piece gets a limited number of attempts per retry, and pieces that are already ready are left as they are.

If you edited an item the installer built and it still had work to do, the installer won’t overwrite your changes. It marks the item for review instead, with “The View has saved work that needs review. Open it to finish without overwriting your changes.”

Versions

Every Blueprint version is fixed once it’s published. An installation is tied to the exact version you installed, so the definitions it built always match what that version says.

A newer version in the gallery doesn’t change anything you’ve already installed.
Items you edit after installing stay as you left them. The installer never rewrites your changes to match the Blueprint.
To use a newer version, install it. You decide when.

Querri can also turn work in your workspace (views, dashboards, sites, KPIs, monitors and more) into a Blueprint file that another organization can import and install. The file carries the data as input descriptions and generated sample rows, never your real rows unless you choose to include a small sample.

This is available to selected partners on request. Contact your Querri account team if you’d like it turned on.