Customer story
Great Day Improvements is a multi-brand home improvement group whose staff span installation, sales, manufacturing and corporate functions across the United States. The problem it brought was not a novel one and that is rather the point: two departments — human resources and the contact centre — each wanted a conversational assistant over their own material, and the company did not want two systems, two data paths and two answers to the question of what an assistant is allowed to say.
What was built
None of these is technically remarkable. Together they are the difference between an assistant two departments can share and two assistants that diverge within a quarter.
HR questions and contact-centre questions draw on different material and have different rules about what may be said, but they run on the same deployment, the same guardrail configuration and the same audit trail. Adding the third department later is a configuration exercise rather than another procurement.
Rather than committing the whole system to one provider, the model can be chosen per query. That matters commercially — routine lookups do not need the most expensive model — and it matters practically, because provider availability and pricing both moved during the build.
Answers are generated from the organisation's own documents rather than from what a model happens to know about home improvement, and they carry the source back. For an HR policy question that is not a nicety; it is the difference between an answer an employee can act on and one they have to go and verify.
The whole system runs in the customer's own Azure environment, against their own model endpoints, under their existing security posture. No new cloud relationship, no new data-processing agreement, and no separate perimeter for the security team to reason about.
The characteristics that shaped the build more than any feature did.
Inside the customer's existing Azure environment
Authenticated — staff sign in; nothing is open to the internet
Human resources and the contact centre, on one platform
Selectable per query rather than fixed at build time
The company's own material, preloaded and retrieved with citations
Integration with the existing Databricks estate, and single sign-on
This engagement is in delivery rather than finished, and the page is written in that tense on purpose. There is no outcome metric here because there is not yet an honest one to publish — a first phase covering both departments is in place, and a second phase extending into the company's data platform with single sign-on is planned. A vendor page claiming a result at this stage would be claiming something nobody has measured.
More stories
What each one built and where it runs.
Trading house · M&A
Retail & commercial banking
M&A advisory
Universal banking · State bank
Pension fund administration
Fintech · Digital payments
Medical education · Pharma & biotech
Higher education · Student services
Telecom · Investor relations
Software · HR operations
Consumer tech · Internal HR
Story 1 of 11: Itochu
Talk to a NeuralSeek expert about how it fits your stack, where your data has to live, and the governance your auditors already expect.