Services and pricing logic — in decision form
Not a marketing description — a decision tree: what determines which service fits, what changes the price, and what the honest boundaries are (who it's not for). This is what lets the agent qualify, not just describe.
Common questions and honest answers
Actual questions your team gets asked repeatedly, with the same honest answer a person would give — including "we don't do that" where true. An agent that's only fed marketing copy will eventually be asked something the copy doesn't cover.
Explicit knowledge boundaries
What the agent should say when it doesn't know — a defined fallback, not an improvised guess. See Common AI Lead Qualification Mistakes for what happens when this boundary isn't set clearly.
Keeping it current
Pricing and service changes need to reach the knowledge base, not just your website copy — a mismatch here is a common, avoidable source of an agent giving an outdated answer. See How to Test an AI Agent Before Launch for how this gets caught before it reaches a real lead.
Questions
How much documentation is actually needed?+
Less than people expect — a clear decision tree for services/pricing and a list of real recurring questions covers most of it.
Who prepares this — us or LATYNEX?+
A collaboration — you know your business specifics, we structure it into something the agent can actually use.
What if our services change often?+
The knowledge base needs an update process tied to that — covered during setup, not left as an afterthought.