# Case Study: UWM

**UWM equips 12,000 + mortgage brokers with an AI-powered “Agentic Workforce” built on raia**

***

**Snapshot**

| **Metric**                 | **Before raia**                   | **After raia (AI Agents)**            |
| -------------------------- | --------------------------------- | ------------------------------------- |
| Lead-response time         | 3 – 5 hours, business hours only  | **< 90 seconds**, 24 / 7 (chat + SMS) |
| Broker pipeline coverage   | ≈ 40 % of new inquiries contacted | **100 %** of leads qualified by AI    |
| Application-to-close cycle | 32 calendar days                  | **22 days** (-31 %)                   |
| Equivalent human workload  | 1 LO assistant per 4 brokers      | AI capacity equal to **≈ 10 FTEs**    |
| Borrower CSAT              | 4.0 / 5                           | **4.6 / 5** (+0.6)                    |

> “With raia, every independent broker on our platform gets a virtual assistant that never sleeps—qualifying borrowers, gathering docs and keeping deals on track so loan officers can focus on closing.”\
> — **SVP Partner Experience, UWM**

***

#### 1 | Company & Challenge

United Wholesale Mortgage (UWM) is the **largest wholesale mortgage lender in the U.S.**, originating **$139 billion in 2024** and serving over **12,000 independent brokers** nationwide. Explosive lead volume exposed three bottlenecks:

1. **Lead overload** – Too many borrower inquiries for loan officers (LOs) to vet in real time.
2. **Document drag** – Borrowers stalled on gathering pay-stubs, W-2s, bank links and disclosures.
3. **Status questions** – Brokers spent hours chasing underwriting updates rather than prospecting.

***

#### 2 | Why UWM Chose raia

* **Omnichannel coverage** – Chat, SMS, email and voice in one timeline.
* **Rapid, compliant training** – 20,000 + pages of agency guidelines and transcripts ingested in hours.
* **Human-in-the-loop CoPilot** – 60-day shadow allowed brokers to fine-tune answers to 95 % accuracy.
* **No-code LOS hooks** – Direct connections to BOLT AUS, Encompass LOS and broker CRMs.

***

#### 3 | Solution Design

| **AI Agent**          | **Primary Role**                                                                     | **Key Integrations**      |
| --------------------- | ------------------------------------------------------------------------------------ | ------------------------- |
| Sales-Qual Agent      | Greets inbound leads, collects mini-1003, runs pricing, books call.                  | BOLT API, Calendly        |
| LO Assistant Agent    | Guides borrowers through doc collection, sends secure upload links, nightly chasers. | Encompass, email, SMS     |
| Pipeline Status Agent | Answers “Where’s my loan?” for brokers and borrowers, surfaces conditions.           | LOS webhook, chat / voice |

***

#### 4 | Implementation Timeline

1. **Weeks 1–2:** Upload guidelines, overlays, 5 years of tickets.
2. **Weeks 3–8:** Shadow mode with 50 pilot brokers.
3. **Week 12:** Sales-Qual Agent live nationwide.
4. **Month 6:** LO Assistant and Pipeline Status agents deployed; Spanish pack added.

***

#### 5 | Results (First 6 Months)

* **≈ 85,000 borrower conversations** handled; **62 %** fully resolved by AI.
* Remaining **38 %** pre-qualified, trimming LO handle time **45 %**.
* Lead-to-pre-qual shrank from 5 h to **< 30 min**, boosting win-rate **+18 pp**.
* Application-to-clear-to-close cut from 32 days to **22 days**.
* Avoided hiring 10 support staff, saving ≈ $900 k annually.

***

#### 6 | Key Takeaways

1. **Automate top-funnel pain first** for quickest ROI.
2. **Use Packs for guideline changes** – new DU matrix? Drop the PDF, agent updates in an hour.
3. **Maintain the human touch** – borrowers can escalate to their LO at any point.

***

#### Bottom Line

By embedding raia, **UWM turned AI into the ultimate LO assistant—qualifying prospects, collecting documents and answering status questions at lightning speed—while cutting cycle times and boosting CSAT**.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.raiaai.com/use-cases/case-study-uwm.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.