B2B manufacturer of industrial refrigeration equipment · 24/7 sales
A two-channel AI sales agent with memory and a knowledge base covering hundreds of models.
The AI takes a customer from the first touch on a marketplace to a commercial proposal. One agent works on a classifieds marketplace, the other in Telegram. Both share the same conversation memory for each customer. They know the full product line, handle objections, and follow up steadily until the deal closes.
Picking a model out of hundreds takes a sales rep 5–15 minutes, and overnight marketplace inquiries get lost
Two AI sales agents on a classifieds marketplace and Telegram with shared customer memory and a product knowledge base take the customer to a quote
Model selection: 5–15 minutes → instant; an overnight inquiry answered within a minute
By the numbers
The problem
What keeps a B2B seller from turning inquiries into deals.
- B2B refrigeration inquiries are messy: a customer writes "need an 8 m³ cold room, on a fixed budget", and the sales rep has to pick a model from hundreds on the spot and estimate installation costs.
- A classifieds lead goes cold within minutes, and night-time inquiries are lost entirely because the sales reps are asleep.
- A long sales cycle needs consistent follow-ups after 3, 7 and 14 days. Sales reps forget them, and the customer goes to a competitor.
- When a sales rep leaves, the knowledge leaves too. New hires take a long time to catch up on the specs of hundreds of models.
- Sales reps can't remember every spec and price, so they guess or make mistakes. Customers later catch the inconsistencies.
How it works
Five steps from the first message to a commercial proposal.
-
The classifieds agent replies first
The goal is to get a phone number fast. It handles one typical objection, stays out of technical detail, and doesn't scare the customer off with complex questions.
-
The conversation moves to Telegram
Once the customer shares a phone number, the system moves the conversation to Telegram automatically. The AI writes first: "we talked on the marketplace, let's continue here", so the customer feels no break.
-
The Telegram agent sees the shared context
Shared memory holds a summary of the classifieds chat: what the customer asked, what was answered, which objections came up. The AI doesn't ask the same questions twice.
-
Qualification and closing
First, "qualifier" mode collects business details and specs. Then "closer" mode handles objections and pushes toward a commercial proposal.
-
A live sales rep can take over
A sales rep can take over the conversation at any moment with one command and hand it back to the AI when they go offline. Full control.
Features
What the solution includes.
Live knowledge base
The agent's knowledge updates from company documentation in Google Drive every 15 minutes. Change the price list, and the AI knows the new prices within a quarter of an hour.
Handling split messages
If a customer sends three short messages in a row, the AI waits 20 seconds and merges them into one context. No choppy replies to every word.
Human-like behavior
A human name, formal address, pauses before replying. In most cases the customer doesn't realize they're talking to AI.
No developer needed for setup
The list of sales reps for notifications, greeting texts, working hours: the head of sales changes all of it in the admin panel.
Automatic deal progression
Once qualification is complete, the deal moves to the right CRM pipeline stage automatically. The sales rep gets a ready-made deal card.
Memory across channels
The customer never has to repeat themselves: classifieds and Telegram chats live in one memory. The AI refers back to earlier messages naturally.
Follow-ups only during business hours
Nobody gets a reminder at 3 a.m. The follow-up window is set for your audience and time zones.
Takeover commands
A sales rep types /connect to take the conversation. Types /disconnect to hand it back to the AI. A clean switch that doesn't confuse the customer.
Economics
Before and after in one table.
| What changes | Before | After |
|---|---|---|
| Response to a night-time classifieds inquiry | in the morning, when the customer is already with a competitor | within a minute |
| Matching a model to a request | 5–15 minutes of a sales rep's time | instantly |
| Sales knowledge | in the sales rep's head | in the company's live documentation |
| Follow-up after a commercial proposal | the sales rep forgot | the AI sends a reminder on the right day |
| Night and weekend inquiries | lost entirely | handled the same as during the day |
24/7 response on marketplaces adds 15–30% to "inquiry → phone number" conversion, based on hospitality and B2B marketplace benchmarks. It frees up expensive sales engineers for complex deals.
Who it's for
Where the solution works.
Anywhere the product has a long sales cycle, configurable options, and an active presence in messengers and on marketplaces.
What sets it apart
Why this isn't an "auto-reply bot".
Two linked channels
Not a single bot in a single messenger, but a classifieds→Telegram link with shared memory. The customer goes through the whole journey with no breaks and no repeated questions.
Knowledge tied to your documents
Price lists, rules and technical specs live in your Google Docs. You edit them, and the AI picks up the changes within 15 minutes. No need to call a developer.
Full control for the client
Texts, follow-up rules, working hours, the list of sales reps: all of it is edited in Google Docs and the admin panel. No black boxes and no dependence on the contractor.
We'll connect an AI sales agent to your classifieds account in 1–2 weeks.
Free audit: in a 30-minute review, we'll show how many night and weekend leads you're already losing and how many deals you can win back from your current inquiry flow.
Quiz · 6 questions · 1 minute
Or right away: Telegram · @one_lab_support → · hello@1-lab.dev