Node AI.03

AI data assistant

Ask your business data a question in plain English and get the answer with a chart.

Start a project Take the AI readiness scan

An AI data assistant that answers questions about your sales, operations and finance data in plain English, with charts and the source of every number, for your whole team.

The problem

Simple questions such as "what were sales last month by region?" still need someone to open a spreadsheet or wait for the analyst. Decisions get made on last week’s numbers because asking is slow.

A general chatbot cannot see your data, and when it guesses it invents numbers. A data assistant has to query your real tables, use your metric definitions and show how it reached each answer.

What we build

A chat assistant on the web, Slack or Microsoft Teams that answers questions from your databases and warehouse
Answers as tables and automatically generated charts, with the query or source behind every number
Your business vocabulary: product names, regions, metric definitions and fiscal calendar
Permission-aware access, so people only see the data they are allowed to see
Suggested follow-up questions and saved answers for questions asked every week

How it works

A question is matched to your metric definitions, turned into a safe read-only query, run against your data, checked, and returned as an answer with a chart and its source.

Step 1Map your dataIdentify the tables, metrics and terms people actually ask about.
Step 2Define the metricsAgree definitions so the assistant’s answers match your existing reports.
Step 3Build the assistantQuery generation with read-only access, validation and charting.
Step 4Test with real questionsCollect questions from managers and analysts and measure accuracy before launch.
Step 5Roll outOpen it to the team, with usage and accuracy monitoring.

Deliverables

Data assistant on the agreed channels
Metric definitions and business glossary
Read-only data connection with permissions
Accuracy test set and results
Usage and accuracy monitoring

Typical stack

Python
SQL
Snowflake
LLM and agent frameworks
Google Cloud

We choose tools to fit your environment; this is where we usually start.

Ways to work with us

Build and HandoverA defined system, delivered, documented and handed to your team.Engagement Forward Deployed EngineersOur engineers inside your team until it runs in production.Engagement Managed AI OpsOngoing monitoring, evaluation and improvement of your AI systems.Engagement

Questions

Can it change our data?

No. It uses read-only access, and every query is validated before it runs.

How do we know the numbers are right?

Every answer shows its source and query, and before launch we test it on a set of real questions with known answers.

Which data sources does it work with?

Databases and warehouses such as Snowflake, PostgreSQL and SQL Server, and spreadsheets once they are loaded into a database. We check your setup in the first week.

How is this different from a dashboard?

Dashboards answer the questions you planned for. The assistant answers the follow-up questions, and the useful ones can become dashboard tiles or scheduled reports.

Adjacent nodes

Part of AI & Agents.

Executive and analytics dashboardsDashboards people open every morning, because the numbers are right.DATA.03 Automated reporting and alertsThe numbers that matter, in your inbox before you ask.DATA.04 AI assistants and chatbotsAssistants that answer from your documents and data, and say when they do not know.AI.02 AI-ready data foundationsClean, modelled, governed data, so AI systems give answers you can use.DATA.02

Talk to us about aI data assistant for your business.

Send a short brief. We reply by email, usually with questions before we propose anything.