Objections and "not now": handled, not templated

A canned FAQ response to a real objection is the fastest way to lose a warm prospect. The AI reads what actually came back, in the language it was written in, works out what kind of reply it is, and responds to that. Most of the work is in that sorting: a yes, an unclear answer, an objection and a stop each get different treatment, and confusing any two of them is how automation embarrasses the company it writes for.

First, what kind of reply is it?

Every reply is read for what the person intends, its sentiment and whether there is any intent to meet before anything is written back. Test Lab shows that reading on each simulated reply: “we already have a supplier for this” comes back as an objection, neutral in tone, with low meeting intent. Because a simulation runs the same AI with the same rules as the live system, what you see there is the call it makes on a real reply. That classification decides which path the conversation takes next.

Only a clear yes counts as a yes

Automation books a meeting only on an unambiguous positive reply: then a real slot goes into a real calendar, the invitation is sent and the CRM is updated. When an answer is unclear, the AI asks one neutral clarifying question first. It clarifies before it pushes.

A “not now” gets an answer

A polite “thanks, not right now” is an objection, and the Persuasion Agent answers the reason behind it with the argument that fits. It reads why the answer was no, picks that argument and keeps the conversation going until a meeting is booked. An existing system becomes the reason to meet, because your product sits on top of what they already run: told that a plant already runs SAP, the answer for a shop-floor product is that it sits under SAP and feeds it live machine data, followed by two concrete times. “We have a supplier” gets a question about which one, or only the differences the agent can stand behind.

No invented claims

Persuading is not inventing. The agent names only differences it can stand behind, from the product knowledge you loaded: each product's description, value proposition, capabilities and demo guidance. Before any reply leaves, fabricated meeting commitments, unresolved placeholders and prohibited links are stripped from the draft, and if both the reasoning model and the fallback model fail to produce a clean reply, the reply is held. A generic template is never sent in its place.

Real questions get real, specific answers

The AI answers the actual question asked, whether it is about pricing, scope or “how is this different from X”, drawing on the product catalogue. It does not send a generic brochure link. Asked for more details, it can send your presentation and propose a time in the same reply. The answer stays with the product the conversation is about: every send records which product and persona it was for, and the AI never switches offers mid-thread.

A “stop” gets silence

An opt-out is not an objection, and the agent never argues with one. “Stop” is permanent in every language, on every channel: a refusal never gets a clever closing message, and the thread simply goes quiet. It is also the rule Test Lab lets you confirm before a single real contact has the chance to say it.

Which arguments actually work

Every argument the agent uses is measured in Persuasion Performance, so you can see which one produces meetings. In Test Lab the objections your salespeople hear every week can each be tried in a minute, before one of them reaches a real inbox.

More about this product: LinkedIn Sales Navigator AI

← Back to the blog

Objections, questions, and the polite "not now": handled, not templated

First, what kind of reply is it?

Only a clear yes counts as a yes

A “not now” gets an answer

No invented claims

Real questions get real, specific answers

A “stop” gets silence

Which arguments actually work

See real objection-handling threads

Further reading

Meet the AI sales team that never sleeps, never blasts, and never forgets a follow-up

The safety engine that protects the account it works from

The role fence: why the AI would rather skip than guess

LinkedIn has no outreach API. Here is what we do instead.

Talk to a human 🙂