Skip to main content

​
What it does

The Library is your organization’s context: the shared understanding Pluvo works from every time it answers. It holds the definitions your figures are calculated from, and the instructions that shape how Pluvo approaches a question. Approval is what makes a definition count. Only approved definitions are used to answer questions, so anything suggested, rejected, disabled, or broken is never treated as authoritative. That is what keeps results consistent and defensible.

​
What’s in it

Open Library from the sidebar and you land on the Context Library, organized into tabs:
  • Needs review, suggestions and broken items waiting on a decision
  • Metrics, the named measures that ground Pluvo’s answers
  • Tables, governed sets of rows that the Assistant and your metrics can read from
  • Skills, standing instructions people switch on for themselves
  • Rules, organization-wide instructions applied to every answer, visible to admins
If something needs your attention the Library opens on Needs review. If you are all caught up, it opens on Metrics. To add something new, choose Add context and pick Table, Metric, Skill, or Rule. What you see depends on your role, so Rule appears only for admins.

​
Reviewing what Pluvo has found

As Pluvo works with your data, it proposes metrics and queries it thinks are worth keeping, and it flags metrics that have stopped matching your data. All of it lands in Needs review, with a count so you can see how much is waiting. Search the list, or filter by status, source, and data source to narrow it down. Then make a call on each item, right from its row:
  • Approve puts it to work
  • Reject marks it wrong, so Pluvo will not suggest it again
  • Disable takes it out of use without saying it was wrong
Rejecting asks for an optional reason. It is worth writing one, since it saves the next person wondering why.
Broken metrics are the ones to watch. A metric is flagged broken when its numbers stop reconciling with your data, or when another metric it depends on was taken out of use. Either way it is excluded from answers until someone looks at it, which is how you find out something quietly stopped being trustworthy.

​
Metrics

The Metrics tab lists every metric with its status, where it came from, how often it has been used, and when it last changed. You can search, filter by status and source, and sort by name, usage, or last updated.

​
Reading a metric

Open a metric and everything about it is on one page. A strip across the top shows its status, source, how often it has been used, when it was updated, and its currency, followed by its description and any other names people use for it. Below that, in order:
  • Calculation, the formula in plain language, with the exact query available under Under the hood if you want the precise version
  • Source, the connection and tables it draws on
  • Preview, the live numbers, which you can break down, along with how the metric checks out against the expected total for your data
  • Dependencies, the other metrics built on this one, so you know what else moves if you change it
  • Usage, where it has actually been used in answers
  • History, what has happened to it over time

​
Creating a metric

Metrics are built in conversation with the Assistant.
1

Start a new metric

Choose Add context, then Metric. A new Assistant chat opens, and the Assistant explains how metrics work and asks what you want to measure, one question at a time. You can also just ask for a metric in any chat.
2

Shape the draft

As you answer, the draft appears in a metric workbench beside the conversation. Adjust it there directly, or keep refining it in the chat. Choose what you are measuring, the column it comes from, and how to add it up, combine terms with add, subtract, multiply, and divide, then narrow it down with filters and choose how it can be broken out.
3

Check the preview

A live preview shows the number your formula produces, so you can confirm it before saving.
4

Name it and save

Give it a clear name and description, add the other names people use for it so the Assistant finds it, then choose Save metric. Open in Library takes you to it.
If the Assistant suggests a change after you have edited the draft yourself, you choose: Use assistant version or Keep my edits. To change a saved metric, open it and choose Edit. Formulas follow standard order of operations, so multiplication and division happen before addition and subtraction. To group terms differently, click the operator between two terms and add an opening or closing bracket. You can also build a metric out of other metrics. When you are working from a Table, Saved metric lets you reference other metrics built on that same Table.
A formula needs at least one measure or saved metric. Fixed numbers on their own cannot define a metric. And if what you are building duplicates an existing metric, Pluvo tells you and offers to open that one instead.

​
Tables

A Table is a reusable set of rows your organization has agreed on, defined once and used as the foundation for metrics and answers. Rather than every metric working out which rows count, you define the shape once, check it, and build on it. Each Table carries a plain-language contract: what a row represents, how dates are handled, which columns are safe to group or filter by, and what each column means.

​
Getting a Table into your Library

Tables are built in conversation with the Assistant too, and there are two ways in. Start from a result. When the Assistant shows you a set of rows worth keeping, choose Save Table in the data panel, give it a name and a short description, and choose Review Table. The complete result is captured, not just the rows on screen. Start from a description. Choose Add context, then Table, or ask for one in any chat. The Assistant asks what rows you need, one question at a time. Either way, the draft appears in a Table workbench beside the conversation:
1

Review the draft

Look through the columns, what each row represents, and the query behind it. To change anything, ask the assistant.
2

Check it against live data

Choose Check and preview to run it. You get real rows back, plus confirmation that the structure holds up and every row is uniquely identified.
3

Save and publish

Once the checks pass, save it, then Publish it to make it available to everyone. Each save after that is a new revision.

​
Using a Table

Open a Table to see its status, readiness, and description, the questions it is meant to answer, a live preview of its rows, and its current and published versions. Edit reopens the conversation it was created in. If a Table takes inputs, such as a date range, you can try different values in the preview. Choose Save defaults to make your values the new defaults for everyone. Metrics already built on the Table keep the version they were built against. Once published, a Table becomes a source you can build metrics on, and the Assistant can query it directly when answering, filtering, totaling, grouping, and ranking its rows rather than working the logic out again.

​
When the source data changes

Pluvo re-checks a Table each time its data source syncs. When new data has arrived, the Table shows New source data is available. and updates itself. If an update fails, choose Try again. If a change you make cannot be saved because the Table no longer fits its data, Fix in Assistant opens the Assistant to repair it with you.
Define a Table when several metrics share the same underlying rows. It stops each one re-deriving the same logic, and a fix in one place fixes everything built on it.

​
Skills

A skill is a standing instruction for Pluvo: how a metric is calculated, what counts and what does not, a pitfall to steer around, or context about your organization worth keeping in mind. Instead of explaining the same thing in every conversation, you write it down once. A skill applies to you only once you switch it on, and even then Pluvo brings it in when the question calls for it, rather than every time.

​
Writing a skill

1

Start a new skill

Choose Add context, then Skill. You get a conversation on the left and the skill on the right.
2

Describe what Pluvo should know

Tell the assistant what you want to teach it. It asks a question or two and fills in a draft, which you can Accept or Reject. You can also type straight into the editor, and choose Stop if a draft goes the wrong way.
3

Name it and activate it

Give it a clear name and instructions, up to 10,000 characters with Markdown supported, then choose Activate. Drafts save as you go but are not used until they are active.
A good skill states the instruction, the pitfall, and the boundary. For example: revenue means recognized revenue net of refunds and never bookings, and intercompany accounts are excluded from all rollups.

​
Switching skills on, and sharing them

The Skills tab lists every skill you can see, with its status, whether it is Organization or Personal, and an Applied to me switch. A skill only affects your answers once you switch it on, so a skill someone else wrote waits until you choose to use it. A skill you write, or one the assistant saves for you when you ask it to remember something, is personal to begin with. Use Share to make it available to the whole organization, or to specific people.

​
Changing a skill

Open a skill to read it in full. From Edit you can change it directly on the page, or choose Edit with AI to refine it in conversation. The actions menu lets you activate or deactivate it, share it, or delete it.

​
Rules

A rule is a standing instruction the assistant applies to every response, for everyone in your organization, with nothing to switch on. Rules are for the things that must never be got wrong: what revenue means, which accounts are excluded, which currency figures are reported in. Rules are managed by organization admins, so the Rules tab appears only for admins. If you are not an admin, the rules your admins have set still apply to your answers.

​
Writing a rule

1

Start a new rule

Choose Add context, then Rule.
2

Describe what Pluvo should always do

Tell the assistant the instruction you want enforced. It drafts the wording for you to Accept or Reject, or you can type it yourself.
3

Activate it

Give it a name and instructions, then choose Activate. From that moment it applies to every answer across your organization.
Open a rule to read it, edit it directly or with AI, activate or deactivate it, or delete it. Deleting is permanent, and the assistant stops applying it immediately.
Rules apply to everyone with no opt-out, so keep them few and deliberate. A long list of competing instructions dilutes the quality of every answer your team gets.

​
Skills or rules?

Use a rule for something that must hold every time, and a skill for guidance that helps in context or that only part of the team needs.
Your organization’s brand kit, the colours, fonts, and logo for artifacts, can live here as a skill or a rule. The Assistant applies it to new artifacts automatically. See Artifacts.

​
Removing a metric or table

Metrics and Tables can be deleted from their page or from the list. Deleting is permanent, so Pluvo asks you to confirm first. Deleting a Table that live sheets use lists those spreadsheets first. Their live sheets keep their current values but can no longer refresh. If a data source is disconnected, the confirmation lists the saved Tables and Metrics built on it. Those Tables and Metrics are marked Source deleted and cannot be restored.
A metric that other metrics are built on cannot be deleted until those are removed. Pluvo names the dependent metrics so you know what is in the way, and the Dependencies section shows you the same picture before you try.

​
Where your context comes from

Metrics reach your Library five ways: from Pluvo’s catalog of common definitions, discovered in your data, promoted from a saved query, saved from a conversation, or authored by your team in the builder. Every metric shows its source, so you always know where a definition came from. When the Assistant uses any of this context in an answer, it says so, and links back to the metrics, tables, skills, and rules it drew on. Clicking a metric figure in an answer opens the metric explorer, where you can see its governance and break the number down.