Not when configured correctly. The system prompt explicitly forbids quoting rates, suggesting products, or estimating borrowing amounts. Every advice-shaped question is routed to a regulated adviser. We strongly recommend having your compliance officer review the system prompt before launch and sample transcripts weekly for the first month — the prompt itself is the compliance perimeter, so it deserves the same scrutiny as any other piece of consumer-facing collateral.

If your panel is published on the site, SleekAI reads it. If it's a private supplier list, drop it into OpenAI Files for vector retrieval. The bot will be transparent about what 'whole-of-market' means in your case — most brokers cover 80-95 lenders, which is functionally the active market though not literally every lender. That nuance is worth getting right; it's a TCF (treating customers fairly) issue, not just a marketing one.

Same approach: shares general information about what each product is — life cover, critical illness, income protection — then refers visitors to your protection adviser for any recommendation. If your firm holds permissions for non-investment insurance contracts, the bot can capture an interest in protection alongside the mortgage and route both to the right specialist on the team.

Yes. Every transcript lives in WordPress with a timestamp, page URL, and visitor IP for review by your compliance team. CSV export is available for SUP 15 evidence trails. We recommend establishing a written policy on retention period — six years is the FCA standard for advice records — and configuring WordPress backups accordingly. The data never leaves your hosting except to call your own OpenAI key, which simplifies the data-protection picture considerably.

SleekAI is configured to escalate any signs of distress, debt difficulty, bereavement, or other vulnerability indicators straight to a human adviser, with the conversation flagged for sensitive handling. The bot does not attempt to advise vulnerable customers itself, in line with the FCA's Vulnerable Customers Guidance. The escalation prompt should be reviewed alongside your firm's existing vulnerable-customer policy so the two reinforce each other.

Multibot lets you run a residential bot on /residential and a commercial bot on /commercial, each with its own tone and source pages. Commercial mortgages and bridging fall under different regulatory regimes — commercial lending is largely unregulated except for regulated bridging — so a separate prompt and knowledge base helps keep the perimeters clear. Each bot can route to a different specialist team.

The bot recognises adverse-credit signals — defaults, CCJs, IVAs, recent bankruptcies — and routes them straight to a specialist adviser without attempting general guidance. Specialist lending is where mainstream broker advice carries the highest risk of being inadvertently misleading, so the threshold for human escalation is set deliberately low. Visitors get a warm 'this is a great question for one of our specialist advisers' rather than a half-answer.

Yes. Display conditions can show SleekAI when no one is at the desk and your live chat during staffed hours, or run them on different pages — SleekAI on the FAQ pages, live chat on /contact. The two don't share threads, but the SleekAI transcript stays available in WordPress for any adviser picking up the lead later, which preserves the full conversation history for compliance purposes.