This complete walkthrough shows how to configure the LoanOfficer.ai AI Assistant from User Settings. Set its name and phone number, control delayed replies, add important and trackable links, define booking availability and time off, build follow-up sequences, configure the optional production coach, teach the assistant about you and your products, test replies in Sandbox Mode, apply guardrails, and embed the assistant on your website.
Open User Settings and select the AI Assistant tab. The first section controls the assistant's name and optional phone number. A separate name helps the assistant distinguish itself while it schedules appointments and follows up with leads. Adding a phone number lets you text the assistant and receive updates on your own phone, while the platform's assistant-chat icon keeps those conversations available in the CRM. This is also where delayed reply is configured. Instead of delivering a full response immediately, the assistant can wait for a selected interval of up to three minutes. The delay changes pacing, not the substance of the response, and can make an active text conversation feel less automated.
Use Important Links to provide the destinations the assistant may need during a conversation. The loan-application URL is the essential starting point because it tells the assistant where to send someone who is ready to apply. You can also add a YouTube channel, social profiles, or other approved resources. A link can be converted into a smart tracking link from the same area. That option records how many times the specific link is selected, helping you distinguish a resource that was shared from one that was actually opened.
Scheduling tells the assistant when it may book time on your calendar. Select the appointment types and lengths you offer, define the days and hours you are available, and set the minimum notice required before a meeting can begin. A 12-hour minimum, for example, prevents a borrower from creating an unexpected same-day appointment. Time Off blocks dates that should not be offered. Add a vacation or another unavailable period, and the assistant will keep those dates out of its booking choices. These settings define availability; they do not stop the assistant from discussing appointments with a lead.
The scheduling area also provides website calendar options. You can publish the calendar inline, as a pop-up, or as a badge, then copy the supplied code into your site. A visitor chooses a meeting type, reviews the available times, selects an appointment, and enters their contact details. Test the embedded calendar after installation. Confirm the meeting types, time zone, minimum notice, availability, and contact fields all behave as expected on both desktop and mobile.
Follow-Up controls what happens after a lead stops replying. When a conversation goes quiet, the assistant can wait a day and create a contextual follow-up message. Additional steps can wait two days, three days, seven days, or another interval you choose. The assistant writes each message from the conversation context rather than requiring one fixed script for every lead. The re-enroll buffer adds separation before a contact can enter another follow-up sequence. Use it to keep a recently nurtured lead from restarting too quickly and receiving more messages than intended.
Coach is an optional mode inside the AI Assistant. When enabled, it uses the business context you provide to support production goals and challenges. Describe your business, experience, current priorities, obstacles, and available resources. You can also identify team members so the coach understands what can be delegated and who is available to help. The coach remembers the goals and communication preferences entered in this section. If you do not want coaching, leave it disabled; its setup does not affect borrower follow-up.
Training is where you explain how the assistant should represent you. Add direct rules, such as ending an interaction with a clear next step, sending a particular link after a defined response, or including required language when rates are discussed. Then complete the identity and company-bio sections with the experience and business details the assistant should know. Offerings is intended for products that need extra context. Standard mortgage knowledge is already available to the assistant, while specialized programs—such as a construction product, a product for a particular geographic condition, or financing for a defined profession—can be documented with the details that distinguish them.
Sandbox Mode lets you test the configured assistant before relying on it in a live conversation. Ask the questions borrowers are likely to ask, including product eligibility and current-rate questions. The walkthrough demonstrates that the assistant can answer a broad product question, ask a qualifying follow-up, and use the account's Rate Watch configuration when discussing rates. A useful test checks more than whether the answer is factually responsive. Confirm that the assistant follows your instructions, includes required disclosures, asks an appropriate next question, and directs the person to the correct approved link.
Guardrails bring the assistant's protections and configurable controls into one place. The walkthrough shows built-in handling for RESPA and TILA compliance, content moderation, and consumer opt-out requests such as STOP. Additional controls include rate disclosures and automatic escalation based on trigger words or related topics. Treat the screen as an operational review, not a one-time checkbox. Test the disclosures and escalation phrases you configure, and revisit them when your policies, products, or team responsibilities change.
The Widget section lets you place the configured assistant on your website. Choose the icon, logo, colors, and title, then copy the provided code into the intended page. Website visitors can open the widget and begin a conversation with the assistant using the same business context and rules you configured. Before publishing, test the widget on a phone and desktop. Confirm that it opens cleanly, identifies itself appropriately, handles a representative question, and offers the intended next step without covering important page controls.
Open User Settings and select the AI Assistant tab. Its identity, scheduling, follow-up, coaching, training, guardrails, sandbox, and widget controls are organized there.
Yes. Assigning the assistant a phone number lets you exchange text messages and receive updates on your cell phone in addition to using assistant chats inside the platform.
The walkthrough shows reply delays of up to three minutes. The setting changes when a response is sent, helping conversations feel more naturally paced.
Yes. Add the unavailable dates under Time Off. The assistant will not book appointments during those periods.
A configured follow-up sequence can wait for the interval you choose and then have the assistant write a contextual follow-up. You can add multiple steps and control re-enrollment timing.
No. Use Offerings to add the details of specialized or company-specific products that require context beyond standard mortgage-program knowledge.
Use Sandbox Mode to ask representative borrower questions and verify the answer, next question, disclosure, links, and escalation behavior.
Yes. Customize the icon, logo, colors, and title under Widget, then copy the supplied embed code into your website.
This walkthrough opens User Settings and the AI Assistant tab, then reviews each configuration area. It covers naming the assistant, assigning an optional phone number, using delayed replies of up to three minutes, adding an application URL and other important links, and turning links into smart tracking links. The scheduling section covers meeting lengths, available hours, minimum notice, time off, and calendar embeds. Follow-Up controls AI-written messages after a conversation goes quiet and includes a re-enrollment buffer. Coach is optional and uses business goals, challenges, resources, and team context. Training supplies behavior rules, disclosures, identity, company information, and specialized products. Sandbox Mode tests product and rate questions before live use. Guardrails include compliance handling, content moderation, opt-out honoring, disclosures, and escalation triggers. The final section customizes and embeds the AI Assistant website widget.