Short answer: in businesses where a deal takes three to twelve months, a lot of sales are lost without anyone saying no. The buyer goes quiet and nobody follows up at the right moment with something useful. AI is now good at the boring half of that job: noticing who has gone silent, reading what they asked last time, and drafting a relevant next message for a person to send. It shouldn't be left to negotiate, discount or handle an upset buyer.
Why long-cycle deals die quietly
Think of anything bought slowly: an apartment, a machine for a factory floor, a school admission, a software contract. The buyer enquires, gets a quote, then goes away to think, compare and ask their family or their boss. Weeks pass. The salesperson, meanwhile, has thirty newer enquiries.
The time simply isn't there. Salesforce's State of Sales report, published in February 2026, found the average seller spends only 40% of the week actually selling, based on a survey of 4,050 sales professionals in 22 countries. The rest goes on data entry, research and switching between tools. Follow-up on a lead that went cold in July is the first thing to slip.
Buyers have also changed how they buy. Gartner's survey of 645 B2B buyers, published in May 2026 found that 67% prefer a buying experience without a sales rep, and 45% used generative AI to research vendors. Yet 69% still turn to a sales rep to check what the AI told them. So buyers do most of the work alone, and then want a knowledgeable person at specific moments. Follow-up that lands at those moments wins. Follow-up on a fixed every-Monday timetable mostly annoys.
What AI can do in follow-up today
The useful version works like a very organised assistant sitting on top of your enquiry records. In practice it does four things.
It watches for silence. Every open enquiry has a last-contact date. The AI flags the ones that have gone quiet longer than your normal rhythm for that stage, so a hot quote that's been silent for nine days surfaces before a cold enquiry from March.
It reads the history before drafting. It pulls the buyer's own words from call notes, emails or WhatsApp: the budget they mentioned, the floor they preferred, the delivery date they were worried about. The draft refers to that, not to a template.
It brings something new. A good follow-up carries a reason to reply: a price change, a new unit in their range, a case where a similar buyer solved the same worry, a reminder that an offer closes on a date. The AI matches these updates to the buyers they're relevant to.
It does the logging. The message, the reply and the next step go back into the CRM without anyone typing them in.
An example of how this looks: a buyer asked about a three-bedroom flat in June, said the budget was tight, and stopped replying after the site visit. In September the developer revises payment plans. The system spots that this buyer's last objection was budget, drafts a short note about the new plan for their exact unit type, and puts it in the salesperson's queue with one line of context. The salesperson reads, tweaks a word, and sends. Two minutes instead of twenty, and it actually happens.
Where a person must stay in
Keep a human in front of anything that commits the business: price, discounts, delivery dates, contract terms. Same for any buyer who sounds unhappy, any message to a senior decision-maker, and the first message after a long silence on a large deal. The AI's job there is to have the draft and the context ready, not to press send.
Accuracy matters too. Buyers in that Gartner survey said they meet misleading information from AI and from sales reps at almost the same rate, 51% and 49%. An AI that quotes last quarter's price list hurts you as much as a careless rep. Give it only approved, current material to draw from.
The Indian rules to respect
WhatsApp costs money once the chat goes quiet. Per Meta's pricing page, the WhatsApp Business Platform has charged per delivered template message since 1 July 2025. Normal replies within 24 hours of the customer's last message are free, and so are utility templates in that window. A follow-up to someone silent for weeks needs a paid template, and India's marketing rate went up on 1 January 2026. Budget for it, and send fewer, better messages.
Calls and SMS have their own rules. TRAI's February 2025 amendment to the telecom commercial communication regulations tightened action against unsolicited calls and messages. Promotional calls now belong on the 140 number series and service calls on the 1600 series. An AI that dials leads from an ordinary mobile number is a compliance problem, however polite it sounds.
Consent and opt-outs need to be built in. Give every sequence a clear way to stop, and honour it at once. India's data protection rules make consent and purpose limits the norm, so record why you have each contact and what they agreed to.
How to start without a big project
Pick one stage where deals stall, usually "quote sent, no reply". Pull the last 90 days of those enquiries. Have the AI draft follow-ups for each, with a person approving every one for the first month. Track replies, meetings booked and how long each draft took to approve. If the drafts need heavy editing, fix the source material before adding volume.
The gain to look for is coverage: every quiet buyer gets a relevant message within days, where before many got nothing. If you want to see how this fits into a wider sales setup, our sales pipeline automation page shows the pieces we usually connect.
Arthat AI
Working on something similar at your company?
Arthat AI builds these automations for Indian businesses every week. If you want to talk through whether this fits your situation, we'll give you a straight answer — even if the answer is "don't build it."
