Internal tools built on numbers you can prove.

Turn a question into a tool: someone fills in a few inputs, the tool runs, and the next step happens.

Start freeBook a demo

Automation

1 trigger · 1 condition · 2 outcomes

Daily revenue checkActive

Every weekday at 07:00 UTC

Revenue moves more than 15%

Compared with the trailing 4 week mean

Anomaly detected
Post to SlackRan 2d ago

#revenue-alerts

Normal range
Log to run historyRan today

No notification sent

The tool nobody had time to build.

Between a chat answer and a full application sits the small tool your team keeps rebuilding by hand.

Start from a question

The question behind an answer becomes the question behind a tool. Add inputs for the parts that change, like a region, a plan or a date range.

A form, not a prompt

Give a colleague inputs instead of an empty chat box. They pick the values, the tool runs, and the result comes back the same shape every time.

Triggered by an event

Run on a schedule, on a threshold, or when someone asks in Slack. A tool is a step in a workflow, not a page someone remembers to open.

Shared with the team

Publish it to the workspace so anyone with the right role can run it, without handing your database password around to make that possible.

Connect a database

Pick the kind of database you have. You will need its address and a read only login.

PostgreSQL

PostgreSQL database

MySQL

MySQL database

ClickHouse

ClickHouse analytics

SQL Server

SQL Server database

Trino

Trino query engine

DuckDB

Local .duckdb file

Snowflake

Coming soon

BigQuery

Coming soon

Databricks

Coming soon

Connect PostgreSQL

Nothing is saved until the connection works. iDash opens it read only and reads the list of tables.

Name
Production Postgres
Host
db.meridian.internal
Port
5432
Database
commerce
User
idash_readonly
Password
••••••••••••
Credentials are encrypted per field before they are stored

Connection tested, read only

42 tables and 17 foreign keys read in 1.8s

Why you can automate on it

A tool inherits the same guarantee.

An app runs with nobody watching, and its output moves on. That is exactly where a plausible wrong number does the most damage.

  • One path, no shortcut

    A tool asks its question the same way chat does, with the same counting rules and the same checks. There is no faster, looser path for automation.

  • It stops instead of shipping

    When iDash cannot stand behind a number, the run stops with the reason attached rather than passing a guess to whatever happens next.

  • Runs as the person

    Every run works out roles, workspace and row access from whoever triggered it, so a shared tool never becomes a way around someone's permissions.

  • Custom code, no passwords

    Custom logic runs as Python in a throwaway workspace that gets the results and nothing else, with no internet access unless you allow it.

What iDash didVerified

The question

What was revenue by region last quarter?

  1. 1Used your definitionNet revenue, the way your finance team already defines it.
  2. 2Counted each order onceAn order with six items is still one order, so the total is not inflated.
  3. 3Grouped by regionFollowed the one clear link from an order to the customer's region.
  • Only one way to connect orders to regions, so no ambiguity
  • Each order counted once, whatever it contains
  • Read only connection, capped at thirty seconds

Ask again tomorrow and it runs the same way.

Where a tool goes next.

One tool is convenient. A shelf of them, all reading the same definitions, changes how the team works.

Agents you write

Point your own agent at the same governed numbers. It gets figures that have already been checked, instead of a database connection and a free hand.

Close the loop

A tool reads a number you can trust and hands it to the system that acts on it. iDash only ever reads, and your code does the writing.

A library over time

Tools build up on one set of definitions, so each new one is mostly configuration rather than a project that starts again from nothing.

Activity

Meridian Outfitters

  1. PR

    Priya Raman pinned a chart Net revenue by region

    Revenue analytics · Dashboard

    12m ago

  2. Scheduled run completed Weekly revenue digest

    6 tiles refreshed · 3.2s

    07:00
  3. DO

    Daniel Okafor saved a query New customers by channel

    Verified · semantic

    Yesterday

  4. Compiler refused a query Revenue by campaign

    Ambiguous join path on campaigns

    Yesterday
  5. MS

    Marco Silva connected a source Warehouse

    ClickHouse · 118 tables · read only

    Mon

Keep reading

Embedded analyticsPut answers inside your own productMetrics catalogOne definition per numberSemantic context layerThe context your agents run on

Turn a question into a tool.

Start with the report someone rebuilds every week, and give it inputs, a schedule and an owner.

Book a demo