All case studies

Case study · Technology · Data & operations

A B2B SaaS company

Millions of rows, answerable in Slack.

The answers were in the data; getting them meant a query, an analyst, and a wait. We built a knowledge graph connected to the company's systems of record, so any employee can ask a question in Slack and get an answer drawn from millions of rows, without writing SQL or filing a ticket.

Industry

B2B SaaS

Function

Companywide data access

Interface

Slack

Connected to

Systems of record

The problem

The company's data lived across systems of record with millions of rows, and only a handful of people could query it. Everyone else filed a request and waited, so most questions never got asked at all.

How many accounts on the enterprise plan haven't logged in this month?

When answering a question takes an analyst and a ticket, people stop asking. A knowledge graph over the real data changes that.

What we shipped

A knowledge graph wired into the systems of record, with Slack as the front door.

Entity graph

A knowledge graph

Models how the company's entities and records relate, so questions resolve across millions of rows instead of one table.

Systems of record

CRM

live

Billing

live

Product DB

live

Connected to the SOR

Reads live from the systems of record, so answers reflect the current state of the business, not a stale export.

# ask-data

JT

enterprise accounts inactive this month?

142 accounts · list attached

Ask in Slack

Employees ask in plain English where they already work, and get the answer back in the thread.

Outcomes

What changed after the rollout.

Getting an answer

a ticketseconds

Asked in Slack, answered from live data.

Data for everyone

Anyone can answer their own question in seconds, so decisions run on real numbers instead of gut feel or a week old report.

Routine lookups queued

0

Time back for real analysis

Analysts unblocked

Routine lookups stop landing in the data team's queue, freeing them for the analysis that actually needs an expert.

Your workflow could be next.

Thirty minutes. A straight conversation about your systems and where custom AI would pay back first.