# Case Study: Constellation Software

**Constellation Software equips its global Finance organization with an AI “Agentic Workforce” powered by raia**

***

**Snapshot**

| **Metric**                                  | **Before raia**                             | **After raia (AI Finance Agent)**                      |
| ------------------------------------------- | ------------------------------------------- | ------------------------------------------------------ |
| Time to locate policy / procedure           | 3–6 hours, email chains & SharePoint search | **< 90 seconds** via chat / email / SMS — 24 / 7       |
| Questions answered without escalation       | 0 %                                         | **≈ 70 %** fully resolved by AI                        |
| Finance team “wait time” for clarifications | Overnight (time-zone lag)                   | Real-time in **5 languages**                           |
| Document-update latency                     | Monthly PDFs                                | Live vector store refresh within **2 hours** of upload |
| Equivalent human workload offset            | —                                           | **≈ 2 FTEs** worth of capacity                         |

> “With raia, every finance colleague—no matter the country or time-zone—gets instant, policy-accurate answers. It’s like adding a senior analyst to every local office, without adding head-count.”\
> \&#xNAN;**– VP Global Finance Operations, Constellation Software**

***

#### 1 | Company & Challenge

**Constellation Software Inc. (CSI)** is a global enterprise software group with **64,000 employees in 100 + countries**. CSI’s decentralized model empowers each business unit, but produces:

1. **Diverse local policies** for expenses, allocations, revenue recognition and inter-company billing.
2. **Constant change** as IFRS / GAAP updates and corporate integrations roll in.
3. **Time-zone friction**—finance staff in Asia-Pac waited overnight for Toronto HQ clarification.

***

#### 2 | Why Constellation chose raia

* **Bulk ingestion & continuous sync** – hundreds of finance manuals, SOX controls and policy updates are drag-and-dropped into Launch Pad; auto-versioning keeps the vector store fresh.
* **Multilingual out-of-the-box** – a single agent answers in English, Portuguese, Spanish, French and German, matching CSI’s top regions.
* **Granular permissions** – SSO & role-based visibility ensure sensitive M\&A ledgers are visible only to authorized controllers.
* **CoPilot feedback loop** – during a 60-day pilot, controllers thumbs-up/down answers, pushing precision above 96 % before company-wide rollout.

***

#### 3 | Solution Design

| **AI Agent**             | **Primary Role**                                                                                                    | **Key Integrations**                                     |
| ------------------------ | ------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------- |
| Finance Policy Concierge | Answers queries on chart of accounts, expense rules, revenue recog, allocation logic; cites exact policy paragraph. | Vector store of live policies; chat / email / SMS skills |
| Journal-Validator Bot    | Checks draft journal lines against policies, flags GL or entity mis-codes.                                          | ERP webhook (Oracle, SAP)                                |
| Allocation-Explain Bot   | Breaks down cost-allocation logic for BU CFOs, producing Excel-ready explanations.                                  | Data warehouse API; email attachment                     |

***

#### 4 | Implementation Timeline

1. **Weeks 1-2** – Curated 1,200 PDFs / DOCs → Markdown; bulk uploaded.
2. **Weeks 3-6** – Pilot in CoPilot with 75 controllers across 6 time-zones; feedback captured.
3. **Week 8** – Company-wide live chat and Teams bot released.
4. **Month 4** – Journal-Validator & Allocation-Explain agents deployed.

***

#### 5 | Results (first 6 months)

* **≈ 48,000 finance conversations** handled; 70 % resolved without human escalation.
* Remaining 30 % arrive to analysts pre-diagnosed, cutting handle-time **45 %**.
* **Policy-lookup time** collapsed from hours to seconds, supporting quarter-close crunch.
* Live update workflow means policy changes propagate to answers **within 2 hours** of approval.
* Avoided hiring 8 additional support analysts—saving ≈ US $1 million annually.

***

#### 6 | Key Takeaways

1. **Treat policies as data, not PDFs**—vectorizing documents unlocked instant retrieval & citation.
2. **Start with pilot power-users**—controllers’ feedback sharpened accuracy fast.
3. **Continuous sync is critical**—automated doc-watch ensures answers never drift from the source of truth.

***

**Bottom line:** Constellation Software turned raia’s Agentic Workforce into a 24 / 7 financial policy concierge—eliminating time-zone delays, boosting compliance confidence and freeing finance talent to focus on analysis, not document hunting.