AI assistant for Instagram and Telegram
It answers the routine messages itself, qualifies enquiries and hands a manager only what needs a person.
- Buildfrom $1,700
- Timeline2–4 weeks
- Supportfrom $570/mo
Who this suits: Accounts and chats where the same questions arrive every day and a missed message is a lost order.
What is in the first version
- direct, comments, Telegram
- knowledge base
- lead qualification
- reply log
The figures above are the package this sits on: AI agent or process automation. Anything past its edges the calculator adds before the work starts, not after.
How the path goes
- 01
A message arrives
The assistant matches the message against what it is allowed to answer: your prices, your terms, your materials.
- 02
It answers or asks a follow-up
A routine question closes in your tone; an enquiry gets the two or three questions that decide who takes it.
- 03
A person gets the rest
The thread reaches a manager with the contact, the answers and the question isolated, in chat or your CRM.
What decides whether this works or annoys people
It knows when to shut up
It fails not on a wrong answer but on a confident one to a complaint, so silence and handover is the fallback.
Tone is a setting
Voice, greeting, the customer’s name and the form of a refusal are configured and reviewable, so nothing drifts when a model updates.
Platform limits are real
We build inside Instagram and Telegram rules rather than around them, because an account is easier to lose than to get back.
What to measure once it is live
- share answered without a human
- time to first reply
- manager handovers per day
- threads that became orders
Why this comes out faster
The shape of this one is known: the states, the edge cases and the things that usually go wrong have been decided before. Nothing here is a template, and the saving is not in your half of the work. It goes into your process, your content and the systems this has to talk to, which is the part nobody can have solved in advance.
A likely stack for this
Picked against the task when we scope it, not decided in advance. This is the shelf it usually comes off.
- Server and agentPythonFastAPIPostgreSQLRedisTelegram Bot API
- Models and knowledgeOpenAIAnthropicpgvector
- Running itDockern8nGrafana
Price it yourself, right here
Five steps, and you can see the number without leaving a contact. The estimate accounts for the kind of work, what you already have and what it will need inside.
These bills do not come from us
- AI model calls, at the model’s published rate
- Hosting and the server it runs on
- Platform business accounts and their connections
- SMS and telephony for manager alerts
Asked before the first call
- Will customers know it is not a person?
- That is your call and a setting. We recommend saying so once at the start: the reply is faster anyway, and nobody feels tricked when a person takes over.
- Where do its answers come from?
- From your own materials: the price list, the delivery terms, the questions that already repeat, the wording your managers use. We collect that once, and afterwards you edit it without us.
- Can it do WhatsApp as well?
- The answering logic is shared, so yes. But each channel is its own connection and its own platform rules, so WhatsApp is a line in the estimate rather than a free extra.