← back to all cases

Orakul: a multi-agent AI back office in Telegram for a service company

Orakul: a multi-agent AI back office in Telegram for a service company

The routine of a service company looks the same in every niche: scheduled client reports, audits and checks, answers to the same questions, tasks someone must not forget. People do all of it — across ten tabs and three services. I built Orakul for myself, as my own back office; today it is a platform that can be deployed for any company that earns by providing services.

One Telegram contact instead of a dozen services

The interface is a plain Telegram chat. Type, dictate a voice message, drop a photo or a document — an intent classifier decides which "employee" takes the job: every agent on the platform is a role with its own instructions, tools and permissions. Need a specific specialist — address them directly, like a colleague.

A  task in Telegram — a finished TXT/PDF report in the chat.

What the platform does instead of people

Proven on live projects

This is not a concept: Orakul is what runs the free LLM visibility audit on this site and the website-analysis widget on a client's WordPress site right now. Access is whitelist-only, API keys and tokens live in encrypted storage (Fernet), each user can bring their own LLM key, and agents, tasks and permissions are managed in a web dashboard — no programmer required.

The dashboard: agent roles, task schedules and the knowledge base are configured without code

I will deploy it for your company

The platform is installed on your server — your data, keys and conversations stay with you. I am ready to deploy the full server version turnkey: initial setup of agents around your processes, knowledge-base loading, tools and access wiring — plus ongoing technical support. If your company's core business is services, Orakul will take over the part of the work that repeats.

Maxim Safianov
Maxim Safianov

I work at the intersection of technical SEO and software engineering: helping sites stay visible for classic search and AI answers alike, and building the Python tooling that automates it.

No comments yet — be the first to share your thoughts.