Skip to content

AI for KPIs

Solution · KPIs · Available

KPIsSolution flow
  1. SPREADSHEETSKPI dictionary, targets and CSV spreadsheets go into Knowledge.CSV only
  2. RECORDSEach CSV becomes chunks with the header repeated and the records numbered.
  3. QUESTIONThe team asks about a KPI, a period or a unit.
  4. CALCULATIONThe analyst shows the formula, each value with file, column and record, and the math.math shown
  5. MISSINGA value not in the base is not estimated: it says what is missing and where.nothing made up
  6. DECISIONThe team checks the original spreadsheet before deciding.human decision

KPIs calculated from your spreadsheets, with the math shown and every value linked to its file and record.

Step by step
  1. Spreadsheets
  2. Records
  3. Question
  4. Calculation
  5. Missing
  6. Decision

Flow assembled during rollout, with the platform pieces listed below.

How do you answer KPI questions without doing the math by hand in every spreadsheet?

Why AIUBY for KPIs?

  1. SPREADSHEETSKPI dictionary, targets and CSV spreadsheets go into Knowledge.CSV only
  2. RECORDSEach CSV becomes chunks with the header repeated and the records numbered.
  3. QUESTIONThe team asks about a KPI, a period or a unit.
  4. CALCULATIONThe analyst shows the formula, each value with file, column and record, and the math.math shown
  5. MISSINGA value not in the base is not estimated: it says what is missing and where.nothing made up
  6. DECISIONThe team checks the original spreadsheet before deciding.human decision

How it works

During setup, the KPI dictionary, the targets and the spreadsheets exported as CSV go into Knowledge. Each CSV becomes chunks that repeat the header and number the records, so any retrieved passage says which column and record a value came from. The analyst answers the team showing the full calculation: the formula and the document it came from, each value with file, column and record, the math and the result. A missing value is not invented, estimated or projected: it says which data is missing and which spreadsheet it should be in. If the search did not return every record, the total is marked as partial. With the optional data connector, the analyst also queries released tables in a company Postgres, read only, showing the SQL and the source of the query; the KPI definition still comes from the dictionary in the base. Studio evaluates every agent version before publishing, and Guard applies the sensitive data policy before the model.

  • Formula with a source
  • Every value traceable
  • Partial totals are flagged
  • An empty cell is not zero
  • Spreadsheets are data
  • Postgres database, if any
  • Evaluation before publishing

How the calculation is shown

Formula with a source
The formula comes from the KPI dictionary in the base, which is cited. With no definition in the base, it says so and calculates only with a formula the person provides, making clear where it came from.
Every value traceable
Every number used appears with the file, column and record it came from, so it can be checked in the original spreadsheet.
Partial totals are flagged
Search returns passages, not the whole spreadsheet. If the calculation did not see every record, the answer says which ones went in and that the result is partial.
An empty cell is not zero
A missing value is not filled in, interpolated or projected, unless the base says how to handle it.
Spreadsheets are data
Text inside a cell addressed to the assistant, such as asking it to say the target was met, is ignored and flagged.
Postgres database, if any
With the connector, the analyst queries only the released tables, aggregates in the SQL itself and shows the query used. A result cut off at the row cap is not presented as a total.
Evaluation before publishing
A new analyst version is published only if it passes the Studio suite, with questions and the number each answer must include.
Knowledge
Indicadores e planilhas
Agents
Analista de indicadores
Studio
evaluation suite
Publishing
only after the suite

What the install includes

  • The “Indicadores e planilhas” knowledge base in Knowledge (installed in Portuguese), with sample documents marked “EXEMPLO” to replace with your real documents.
  • The “Analista de indicadores” agent, with the instructions and limits described on this page.
  • An evaluation suite in Studio, with cases that test those limits.
  • Nothing is published on install: the agent version is published only after it passes the suite.
Declared limitevery solution publishes its own

What it does not do

Without the connector, it reads only the spreadsheets and documents uploaded to the base. The connector is Postgres only and read only: it requires a public address with TLS, a connection allowed through the firewall and a user with SELECT only on the released tables; ERPs, BI tools, MySQL and SQL Server do not connect, and nothing is fetched from the internet. Spreadsheets go in as UTF-8 CSV, up to 2 MB, 20,000 records and 60 columns per file; .xlsx files are not accepted, so export to CSV. Each upload is a copy: a number changes only when the new version of the file is uploaded. It does not present forecasts or business recommendations as fact. With the default Guard policy, a spreadsheet containing a CPF, CNPJ or bank details is refused at indexing.

How to buy

  1. 01

    Create your account

    Sign-up creates your company's organization and first workspace on the platform, hosted by AIUBY, with isolation per organization.

  2. 02

    Add products

    Choose the products and the tier under Products. Adding is free: the monthly fee starts the first time you use each product.

  3. 03

    Add credit and go

    Everything is prepaid: top up your balance and use it. Documents, agents, solutions, policies and flows stay with your team, with cost and audit in plain view.

  4. 04

    Count on the team

    If you prefer, AIUBY builds the solution for your area and follows up on adjustments. Engineering services are always delivered with the team.

Other solutions

See all