Latest
Infercline.
Agent Playbooks

How to use an AI sales assistant without annoying your customers

AI can research prospects, draft messages and keep records tidy. It can also flood inboxes and break marketing law. Where to use it, and where a person should stay in front.

The Infercline Desk · Friday, 6 November 2026 Sign in to save
assorted-color hanged clothes during daytime
Photo by Qin Fan on Unsplash

An AI sales assistant promises more conversations with less effort. It can research a prospect, draft a first email, suggest follow-ups, log every call and keep your customer records up to date. The risk is that the same tool makes it cheap to send far more messages than anyone wants to read. Customers notice, regulators notice, and email providers start sending your messages to spam. The aim is to use AI to make each contact better, not to make more of them.

The safest and most valuable uses are behind the scenes. An agent can read a prospect's website, recent news and your past dealings with them, and give the salesperson a short brief before a call. It can transcribe and summarise calls, update the customer record and remind the salesperson of what they promised. It can clean up duplicate records and fill in missing details. None of this touches the customer directly, and all of it saves time that salespeople usually spend on admin.

Drafting is the next step up. An agent can write a first version of an email or proposal that reflects what the prospect actually cares about. A person should read, edit and send it, at least until you trust the agent's judgement in your market. Generic AI-written messages are easy to spot, and they damage the very relationships sales depends on.

Fully automated outreach is where the problems start, and the rules differ sharply by edition. In the United States, the Federal Communications Commission ruled in February 2024 that the Telephone Consumer Protection Act's limits on artificial or prerecorded voices cover current AI technologies that generate human voices, so such calls need the called party's prior express consent. In the UK, regulation 22 of the Privacy and Electronic Communications Regulations bans unsolicited marketing emails to individuals unless they have consented, with a narrow soft opt-in for existing customers buying similar products, who must be offered a simple way to refuse in every message.

In Singapore, the PDPC's guidelines on the Do Not Call provisions say that before sending a marketing message to a Singapore number, a business must check the Do Not Call Registry unless it has clear and unambiguous consent in evidential form. In Australia, the Spam Act 2003 prohibits unsolicited commercial electronic messages and requires accurate sender information and a working unsubscribe facility. In South Africa, Nigeria, the UAE and Saudi Arabia, data protection laws govern how personal data may be used for marketing and require a lawful basis for processing it. An AI agent that sends at volume can break these rules thousands of times before anyone notices.

Customers' attitudes matter as much as the law. A Gartner survey of 5,728 customers, published in July 2024, found that 64 per cent would prefer companies did not use AI for customer service. Sales is not service, but the finding is a fair warning: many people are wary of being handled by software. Being open helps. If an AI assistant replies to an enquiry or books a meeting, say so.

Inbound enquiries are a better fit for direct AI contact than cold outreach. When someone has asked for information, an agent that answers quickly, books a call with the right salesperson and passes on the context is welcome. Speed of response often matters more than polish here.

Set limits in the tool itself. Cap the number of messages the agent can send per day and per prospect. Keep a suppression list of people who have opted out or complained, and make sure the agent checks it. Require a person to approve any message to a new contact. Log every message, so that if a complaint arrives you can see what was sent and why.

Why it matters to your business

Sales relationships are built on trust, and trust is slow to build and quick to lose. AI that helps your salespeople prepare better and follow up reliably can win business. AI that floods inboxes can damage your reputation, get your domain blocked by email providers and expose you to fines in several of the markets where you sell. The line between the two is mostly about volume, consent and who presses send.

What to do on Monday

  1. Start with behind-the-scenes uses: call summaries, record updates and prospect briefs. Measure the admin time saved.
  2. If you use AI to draft emails, require a salesperson to read and send each one for at least the first few months.
  3. For each country you sell into, confirm the consent rules for email, text and calls with your legal adviser before any automated outreach.
  4. Set a daily sending cap and a suppression list in the tool, and test that opted-out contacts are actually blocked.
  5. Tell prospects when they are dealing with an AI assistant, in a short line at the start of the message or conversation.

Sources