Sales agents
AI should know what it can offer — and what it can't.
Standard pricing an agent may send. Up to 5% it may offer. Between 5 and 10% a manager approves. Above 10% the Commercial Director does. Contractual terms and service levels are a person's decision only, on every plan and in every configuration.
A ladder, with names on each rung
Not “requires approval” — approval by whom. Derived from your own people and responsibility records, so it stays right when somebody changes role.
Two checks, not one
Does this agent hold this capability, and is this specific offer inside its authority? A system with only the first will happily offer 40% off.
Some rungs have nobody on them
Non-standard contractual terms cannot be granted to an agent by anybody, at any price. If that could be unlocked under commercial pressure it was never a limit.
Exceeding it produces a request, not a failure
The agent prepares the case, attaches the competitor quote, and asks. Escalation is the useful behaviour; silence and improvisation are the two bad alternatives.
AI sales agents
The rest of this area.
Consistent follow-up, with authority that stops where it should.
Agent profile
What it may say, offer and commit to — on one page.
Human handoff
Everything discussed transfers with the conversation.
Approvals
A discount request with the reasoning attached.
Communication policy
Hours, disclosure, attempts, opt-outs, consent — enforced.
Commercial promises
“I'll send that tomorrow”, captured and tracked.
Make sure every opportunity gets a next action.
Let AI handle the repetitive engagement, keep follow-up moving, and bring your team in when a customer needs a real conversation.
Agents identify themselves on every call · Nothing commits without you · [email protected]