---
title: "Personal AI assistant: case study"
description: "45% less time on routine: a personal AI assistant with 100+ operations for a private client."
url: https://seo7.es/en/proyecto/ai-personal-assistant-telegram
canonical: https://seo7.es/en/proyecto/ai-personal-assistant-telegram
lang: en
type: CreativeWork
---

# Personal AI assistant: case study

> 45% less time on routine: a personal AI assistant with 100+ operations for a private client.

Page: https://seo7.es/en/proyecto/ai-personal-assistant-telegram

## Summary for AI

45% less time on routine: a personal AI assistant with 100+ operations for a private client. A private client needed an executor assistant built around their own workflows and way of communicating: one Telegram message had to turn into a real action on their machines and in their documents. An ordinary chat with a neural network solves only half of that.

## Questions and answers

### How much did the personal AI assistant cut the client's time on routine tasks?

After the project, the client's time on routine tasks dropped by 45%. This was possible because the assistant performs operations itself: it runs shell commands, works with files, Google Sheets and PDFs, and checks servers over SSH. The person only has to state the task in Telegram or the web interface.

### How did the AI assistant with GPT, Claude and Gemini affect decision-making speed?

The client's decision-making speed grew 2.5 times. The assistant handles advanced search and data analysis, and the model for each task is chosen dynamically based on its specifics. The client gets the search and analysis results right in Telegram or the web interface, with no manual information gathering.

### What level of automation and savings did the personal assistant in Python deliver?

The client's process automation level reached 90%, and operating costs fell by 20%. This effect comes from more than 100 operations combined in one system: text and image generation, server monitoring, browser automation, and work with the file system, spreadsheets and PDF documents.

### Why does a personal assistant use three models at once: GPT, Claude and Gemini?

Different models are strong at different things: one holds long context better, another is more accurate at code, a third is faster and cheaper on simple requests. That is why the assistant is not tied to a single provider and picks the right model for each task itself. Getting dynamic model selection right was one of the project's main challenges.

### Why is the AI assistant for a private client deployed locally?

Working files, server access and internal data pass through the assistant, so a cloud setup was ruled out. The system was deployed on the client's own infrastructure, where they control both the processes and the storage. No data is stored in external services, which ensures confidentiality and full control over all operations.

### What system operations does a personal AI assistant perform?

The assistant runs shell commands, manages files, automates the web browser, processes PDFs, monitors servers over SSH and works with Google Sheets. An ordinary neural network chat will suggest a command but won't run it. This assistant takes a request through to a real action on the client's machines and documents.

### What is the difference between the web interface and Telegram in a personal AI assistant?

The assistant has two control channels. The full-featured web interface is for complex operations. The Telegram integration lets you control the system from anywhere in the world: one message in the messenger triggers a real action on the client's machines or in their documents. Both channels work with the same system.

### What does the AI assistant's "character" mean and why was it tailored to the client?

The assistant's logic and character were tuned in detail to the workflows and communication style of one specific person, which is the main difference between this project and a universal bot. The character concept was developed as early as the requirements analysis stage, and the personalization of responses and logic was finalized in a separate stage together with the rollout of specialized features.

### Which services does the personal assistant in Python integrate with?

The assistant is written in Python and connects to external APIs, servers over SSH, Google Sheets, the web browser and the file system. It is linked to the messenger through the Telegram API. The main Python code implements work with three language models, mechanisms for choosing between them dynamically, and more than 100 specialized functions.

### What stages did the development of the personal AI assistant consist of?

The work went in seven stages: requirements analysis and conceptualization, architecture design and technology selection, core development and model integration, building interfaces for Telegram and the web, tuning logic, personality and functionality, testing and optimization, and finally local deployment, staff training and technical support.

### How is the security of shell commands and SSH handled in the AI assistant?

The security of system operations was one of the project's challenges, since the assistant works with the shell, SSH and the file system. The solution is built on a local deployment on the client's infrastructure with full control over processes and data. During testing, the performance and security of the whole system were checked separately.

### Can a personal AI assistant generate texts and images?

Yes. Besides system operations, the assistant generates texts and images, runs advanced search and analyzes data. These tasks are distributed between GPT, Claude and Gemini depending on the request, and you can use them from the same Telegram window or web interface where commands and file work are run.

### What was included in launching and supporting the personal AI assistant?

In the final stage, the assistant was deployed locally on the client's infrastructure, staff were trained and technical support was provided. Before that, the system went through comprehensive testing, debugging and performance optimization. The client received a working tool and an understanding of how to use it in daily work.

### Who is the personal AI assistant in Telegram and the web interface made for?

The assistant was built for a private individual who needed a tool tailored to their own workflows and communication style, with full control over their data. The case belongs to the automation category; platforms: Telegram and a web interface. The client needed an executor that turns a message into a real action on their machines.

## Other languages

- RU: https://seo7.es/ru/proyecto/ai-personal-assistant-telegram
- ES: https://seo7.es/es/proyecto/ai-personal-assistant-telegram

## schema.org markup

```json
{
  "@context": "https://schema.org",
  "@type": "CreativeWork",
  "name": "Personal AI assistant: case study",
  "description": "45% less time on routine: a personal AI assistant with 100+ operations for a private client.",
  "url": "https://seo7.es/en/proyecto/ai-personal-assistant-telegram",
  "mainEntityOfPage": "https://seo7.es/en/proyecto/ai-personal-assistant-telegram",
  "inLanguage": "en-GB",
  "isPartOf": {
    "@type": "WebSite",
    "name": "SEO7.ES",
    "url": "https://seo7.es/"
  },
  "publisher": {
    "@type": "Organization",
    "name": "SEO7.ES",
    "url": "https://seo7.es/"
  }
}
```