Which channels can we import history from?

Any channel that makes its history available, for example personal Telegram accounts, for the last 30, 90, or 365 days or the full history. Some channels only allow new messages under their own rules. The agent tells you upfront where history is available and shows an estimate of the volume before importing.

Will customers notice that we're importing history?

No. History imports silently: no notifications, no new leads, no bot replies. Nothing changes for your customers.

What happens to personal data in the chats?

Before analysis, the agent strips personal data and filters out ads and spam. The knowledge base gets facts about your product and how sales go, not customer phone numbers or addresses. Conversations marked Private are hidden from everyone and never analyzed.

What if the agent misunderstands how we work?

Everything stays a draft until you approve it. Where the agent is unsure or finds contradictions, it doesn't guess. It asks a manager. After approval, the knowledge base is tested on real customer questions the agent set aside in advance.

Can scripts come from calls instead of chats?

Yes. Calls are transcribed in CRMUS, and call reviews surface objections, unanswered questions, and mistakes. A Suggest Script button builds a draft, which your sales lead approves.

How much does this cost?

With an annual contract, chat history analysis is part of free implementation, up to the volume included in your plan. The agent shows a cost estimate before importing, so there are no surprises.

Sales Scripts and a Knowledge Base From Chat History

Know-how that walks out the door

Your best rep answers “why should I choose you?” in a way that makes people buy. A new hire answers the same question with three paragraphs and loses the deal. The difference isn’t talent. It’s experience that lives in one person’s head and one person’s chat history.

When that person leaves, you lose more than an employee. You lose the objection handling, the sense of what not to promise, and the instinct for which leads will buy and which are just browsing. New hires get told “look at how Anna used to answer,” if her conversations were even saved.

Writing all this down by hand takes time. A sales script for a single pipeline can take a sales lead or consultant several days. A knowledge base usually grows for weeks and never gets finished. So during implementation we do it differently: the agent reads what’s already written.

How it works: three services, one path

1. CONNECTUS collects the history

CONNECTUS brings customer conversations into one inbox. During implementation, the agent connects your team’s work accounts and imports their history: the last 30, 90, or 365 days, or everything. Before importing, it shows how many conversations there are and what the analysis will cost.

The import is silent. Customers get no notifications, no extra leads appear, and the bot doesn’t reply to anyone. If a channel doesn’t make old messages available, the agent says so and suggests other sources: exports, documents, call recordings.

2. The knowledge factory analyzes conversations

Next, the knowledge factory goes to work, step by step:

Channel inventory: what’s there, how many conversations, over what period.
Cost estimate: before anything runs, so you know the spend upfront.
Cleanup: strips customer personal data and filters out ads and spam.
Facts from every conversation: what was asked, what was answered, how it ended.
Draft documents: the facts are organized into ready-to-review sections.

The result is a set of drafts that usually take weeks to write.

For sales reps:

services and pricing: what you sell and for how much, in your customers’ words;
how we communicate: tone, forms of address, standard phrasing;
how a sale goes: how the conversation starts and which steps lead to payment;
objections: what customers say and which answers work;
common questions, with ready answers;
what not to promise: where reps have been burned before;
documents and processes: what to send the customer and when.

For support: five more documents of the same kind, built from customer support conversations.

3. MEMUS stores knowledge and answers with sources

Approved documents go into MEMUS, your company memory. MEMUS answers only when it can back the answer with a document, and it links to the source. If there’s no backing, it says “this isn’t in the knowledge base” instead of making something up.

Everyone uses it: new hires looking for an answer to a customer question, managers checking how things are done, and AI agents replying to customers in CONNECTUS.

Questions for you: where the agent doesn’t guess

Chat histories contain contradictions. One rep offered a 10% discount, another offered 15%. Someone promised delivery in two days, someone else in five. The agent doesn’t pick one at random. Each conflict becomes a question for a manager: “Chats show both 10% and 15% discounts for wholesale customers. Which is correct?”

Questions land on the Questions board in the CHATUS team messenger, along with a summary of where the question came from. You answer in one line, and your answer becomes part of the knowledge base. It doesn’t have to be you: a question can go to a specific expert, into a shared queue, or to a group discussion.

The same path keeps working after implementation. A customer asks something that isn’t in the knowledge base. The AI agent replies “let me check with a specialist” and sends the question to CHATUS. The specialist answers, the answer goes straight to the customer, and MEMUS gets a candidate entry. Once approved, it becomes a ready answer, and next time the agent handles it alone. Your knowledge base grows from everyday work, not from a separate “write the FAQ” project.

Scripts from calls

Chats are half of sales. Calls are the other half. In CRMUS, every call is transcribed and gets a summary, a deal stage, and a follow-up task. Call reviews show objections, unanswered questions, and rep mistakes, and managers get a report.

The Suggest Script button builds a draft from your best calls. Once approved, the script goes into the Sales knowledge base in MEMUS. From there, the AI sales agent and call scoring in SALEUS use it: every new call is checked against the script.

What it looks like in practice

It’s 11:40 p.m. on a Friday. A customer writes: “If I order 500 units, is shipping free?” The rep went home hours ago. In the past, that question would sit until Monday, and over the weekend the customer would find a faster supplier.

Now the AI agent finds the free-shipping rule in the knowledge base, the one the agent pulled from past chats and a manager approved. It answers the customer with a link to the terms and creates a Monday task for the rep to follow up and place the order. If the rule isn’t in the knowledge base, the agent says “let me check with a specialist,” and the question is waiting in CHATUS for the right person in the morning.

How we check that the knowledge base works

Before analysis, the agent sets aside some real customer questions from your history. Once the knowledge base is built, it asks those questions and checks how many it answers correctly. That shows where the knowledge is solid and where you need to add documents or answer questions.

Quality depends on how much history you have. If there’s not much, the agent will tell you and ask for what fills the gaps: price lists, procedures, meeting recordings.

What you get

Sales scripts and support documents built from your real experience, not a template.
A knowledge base that cites its sources and doesn’t make things up.
A list of contradictions you’ve resolved, instead of every rep doing it their own way.
A path for the knowledge base to grow on its own: every unanswered question becomes company knowledge.
New hires who answer customers like your best rep from day one.

This is part of implementation in 14 days and is included in free implementation with an annual contract. Want to see it on your own conversations? Request a demo.

Your chat history already knows how to sell. We write it down