Loading
Loading
Bob answers from your own policy wordings, guides, and past answers, and shows the source on every reply. What it cannot answer with confidence, it hands to a licensed human. Per firm, and never shared across firms.
Bob does not answer from the open web. It reads your firm's own knowledge, answers from it, shows the source, and hands the rest to a human.
Add your policy wordings, product guides, past answers, and FAQs. Bob reads them so it can answer in your firm's own words, not from the open web.
Your documents, your firm
When a customer asks, Bob finds the right passage and answers in plain language, grounded in what your documents actually say. No guessing.
Grounded, not guessed
Every answer carries the document and section it came from, so a customer or an adviser can check it in seconds. Nothing arrives without its working.
Cited and checkable
If Bob cannot ground an answer, or the question is advice, it says so plainly and hands it to one of your licensed advisers. It never bluffs.
A human owns the rest
A general chatbot hands over an answer with nothing behind it. Bob does the opposite: it answers from your own wordings and attaches the document and section, so a customer or an adviser can verify it in seconds.
An example on your own site, in your brand. Bob discloses it is AI, answers from your documents with the source attached, and hands anything it is unsure of to a licensed adviser.
A general chatbot answers from the open web and cannot cite your firm. A shared AI model learns from a pool of everyone's data. Bob is trained on your firm alone, cites its sources, and keeps your book yours.
| A general chatbot | Bob | A shared AI model | |
|---|---|---|---|
| Answers from your firm's own documents | No | Yes | Pooled, not just yours |
| Shows the source on every answer | No | Yes | No |
| Per firm, never shared across firms | No | Yes | No, pooled across firms |
| Flags what it is not sure of | No | Yes | No |
| Hands unsure questions to a human | No | Yes | No |
| Health data never trains a shared model | No | Yes | No, that is how it learns |
| A licensed adviser owns any advice | No | Yes | No |
Real capture of your house style is on the roadmap and lands in the platform phase. When it does, Bob learns from your team's edits to match your voice: your greetings, your plain phrasing, how you sign off. It improves the wording, never the advice.
Every firm's Bob is trained on that firm's own documents and data, and siloed to that firm forever. Nothing is pooled, and nothing one firm's Bob learns can surface in another's. Sensitive and health data is handled in region and is never used to train shared models. Your book, your wordings, and your house style stay your competitive edge, not a shared model's.
From your firm's own documents, wordings, and the facts already on file for that customer. Every answer shows its source. Bob does not answer from the open web and does not guess. If it cannot ground an answer, it says so and hands the question to a person.