No, and it should not. The system prompt explicitly forbids quoting current rates, APR, or estimated payments because those are loan-officer-licensed activities under state and federal rules. The bot can describe how VA loans price in general terms (no PMI, funding fee schedule, comparison to conventional), but every rate question is routed to a licensed LO with NMLS attached. Compliance officers can review the exact prompt language before launch.

It explains what a COE is, the three ways to request one (eBenefits, lender-side request, mail Form 26-1880), and the documents the VA typically needs based on service category. It does not pull the actual COE. When a veteran is ready to start, the bot captures branch, dates of service, and discharge status, then routes to your LO who can request the COE through the lender portal.

Yes. List your loan officers with their NMLS IDs on a team page or custom post type and SleekAI reads them through the data-source wizard. When the bot books a call, it can attach the LO's name and NMLS number to the calendar invite. The NMLS number also surfaces in the bot's signoff messages so the disclosure stays consistent with the team page.

Yes. Run a separate bot on /irrrl or any URL pattern via display conditions, with a dedicated system prompt that knows IRRRL eligibility (existing VA loan, occupancy certification, net tangible benefit). The purchase bot stays focused on first-use entitlement and COE requests. Multibot keeps the prompts tight and the conversations clean.

It explains that veterans receiving VA compensation for a service-connected disability are generally exempt from the funding fee, and that exemption is verified via the COE before closing. It does not confirm exemption for any specific veteran. The LO verifies status through the lender portal during underwriting. The bot frames it this way every time, which is exactly the consistency a manual front desk struggles to maintain.

Claude Sonnet and GPT-4o both follow strict 'no rate quotes, no specific advice' system prompts reliably. Smaller models drift into advice-shaped statements under pressure, which is the failure mode you cannot afford on a regulated lending site. Bring your own key for OpenAI, Anthropic, Google, or OpenRouter and pick the model that matches your risk tolerance and budget.

Yes. Every transcript stores in WordPress with timestamp, page URL, model used, and token count. CSV export is available for examination requests, and the data never leaves your hosting except to call your own API key. Many brokers archive transcripts on a recurring schedule via webhook into their compliance system; the SleekAI conversation log is treated like any other written communication record.

The bot answers around the clock, which matches the reality that active-duty borrowers and their spouses browse at odd hours. It explains POA usage, occupancy intent (which has VA-specific rules for active-duty buyers), and how documents can be signed remotely through your closing platform. It does not promise a specific closing timeline, since that depends on appraisal queues and underwriting; that estimate comes from the LO.