Looking up at glass office towers in a financial district

Case Study · Financial Services

Faster, consistent expense approvals, with people still in charge

Leading Canadian financial services organization · AI-Assisted Expense Review

Book a discovery call

At a glance

Client snapshot

Industry

Financial Services

Region

Canada

Program

AI-Assisted Expense Review

Environment

Dynamics 365

The challenge

Invoice and receipt review was manual, inconsistent and time-consuming. Approvers had to validate every claim against compliance documents themselves before they could approve it.

80%

Faster approvals

AI Analysis and Recommendation

Each claim arrives with its data extracted and a recommendation written.

Auditable, with people in charge

Every recommendation is logged with its reasons, and the approver still decides.

The story

01 · Where they started

Every receipt checked by hand

Expense claims flowed through a Microsoft Dynamics 365 workflow, but the review itself was manual. Approvers read each invoice and receipt and checked it against compliance documents on their own.

Approver checking a paper receipt against a thick policy binder, with a pile of receipts on the desk

02 · What was at stake

Inconsistent decisions and compliance risk

Reviews took time, and different approvers could reach different decisions on similar claims. In a regulated financial services organization, that inconsistency is a compliance risk.

Two approvers reach opposite decisions on the same receipt while a compliance officer looks on

03 · What Taiga AI did

An AI review step inside the existing workflow

Taiga AI implemented an AI-assisted expense evaluation pipeline that intercepts the Dynamics 365 workflow, extracts structured and unstructured data from each claim, evaluates it with Copilot Studio grounded on the policies in SharePoint, and writes back a recommendation before routing the claim to the approver. Every decision is logged in Dataverse for audit.

Receipt data extracted, checked by an AI agent against policy, and passed as a recommendation to the approver, with every step logged

04 · What changed

AI does the checking, people make the call

Expense review is now faster, consistent, auditable and AI-augmented. It reduces manual effort while keeping human authority over every approval. The team is tracking improvements in approval accuracy, pipeline duration, retry count, agent decision reasons and compliance risk indicators.