No. SleekAI's system prompt is configured to describe your firm's services and explain ERISA terms at a general-education level without recommending a specific fund, glide path, or target-date series. Lineup decisions are made by your investment committee or, under 3(38), by your firm under fiduciary authority, neither of which is something a chatbot should weigh in on. Advice-shaped questions get routed to a credentialed advisor with the conversation summary attached and visible on every log.

Yes. Add your fiduciary-scope explainer as a page or PDF, and SleekAI reads it through your WordPress structure. The bot describes 3(21) as a co-fiduciary advisory role where the committee retains lineup decisions, and 3(38) as a full-discretion investment manager role that shifts lineup liability. It uses your own service language rather than a generic ERISA summary, which matters because firms describe these services differently in practice.

Yes, at the level of describing what your firm covers during a recordkeeper search and transition: RFP, benchmarking, fee analysis, conversion timeline, blackout period management, and participant communication. It does not recommend a specific recordkeeper for a specific plan, that is a fit question that depends on plan size, payroll, and pricing, and it is one your team handles in the benchmarking engagement.

Yes. Upload ADV Part 2A, Part 2B, and Form CRS as PDFs or pages, and SleekAI links them automatically when sponsors ask about fees, conflicts, or how your firm operates. Form 5500 references can be added the same way for educational purposes. Many 401(k) advisors configure the bot to surface ADV proactively at the end of any fee discussion, mirroring the disclosure flow of a careful intake coordinator.

Logs live in your WordPress database with model name, token usage, and page URL captured per session. Most 401(k) advisors export them to their CRM or to a compliance archive like Smarsh or Global Relay on a recurring schedule via webhook so chat lives in the same archive as email and sponsor correspondence. You control retention, admin visibility, and whether summaries get emailed to the advisor after each session.

Yes. SleekAI supports multibot deployment with 28 plus display conditions: post type, taxonomy term, user role, logged-in state, URL pattern. A small-plan bot can run on the small-plan landing pages, a separate bot on the mid-market pages, and a different bot on the large-plan pages. Each bot has its own system prompt, knowledge sources, and disclosures, so the framing matches the plan-size segment the sponsor is researching.

It can be configured to, or to decline politely and route participants to the recordkeeper's own help center. Most advisor sites prefer the second approach because participant questions, hardship withdrawals, loan rules, beneficiary changes, are handled by the recordkeeper and routing them through the advisor adds latency. The system prompt and display conditions make the routing explicit so participants do not end up in a sponsor-shaped conversation.

Claude Sonnet and GPT-4o both follow the no-specific-recommendation prompt reliably and produce accurate ERISA language, which matters because the cost of a wrong answer in this category is a sponsor walking away with bad information about fiduciary scope. Smaller or older models tend to conflate 3(21) and 3(38) under pressure. SleekAI is bring-your-own-key for OpenAI, Anthropic, Google, or OpenRouter.