Telegram AI Automation
Project Brief: Development of an Autonomous Telegram Bot System for P2P/Traffic Arbitration
General Project Description
It is necessary to develop a system of 4 interconnected Telegram bots for automating client processing in the p2p direction and traffic arbitration. The system should include an AI module, a protection system against blocks, and an account auto-recovery mechanism.
---
MODULE 1: Main Bot (Ultimate Bot)
### **Functionality:**
1. **4-stage dialogue** with the user
- Stage 1: 3 initial messages (sent consecutively)
- Stage 2: Questions about p2p experience
- Stage 3: Confirmation of the usefulness of cooperation
- Stage 4: Referral to the administrator
2. **Multilingual support**
- Auto-detection of language (Ukrainian/Russian)
- Responses according to the user's language
3. **State system**
- Maintaining the state of the dialogue for each user
- Logic for transitions between stages
4. **Mailing**
- Working with the user database (users_base.txt)
- Configurable delays between messages
---
MODULE 2: Protector Bot (Arbitr Bot)
### **Functionality:**
1. **AI detection of questions/negativity**
- Integration with DeepSeek API
- Analyzing messages for the presence of questions, skepticism, aggression
2. **Protection system**
- Instant freezing of dangerous dialogues
- JSON database of suspended chats (suspended_chats.json)
3. **Admin notifications**
- Sending alerts about problematic users
- Categorization by types of problems (questions, aggression, refusal)
4. **Ignore system bots**
- Absolute ignoring of SpamBot, PremiumBot, and others
---
MODULE 3: AI Bot with DeepSeek
### **Functionality:**
1. **Autonomous communication**
- Integration with DeepSeek Chat API
- Contextual dialogue management
2. **Language detection**
- Algorithm for detecting Ukrainian/Russian language
- Responses according to the user's language
3. **Smart ignore of bots**
- Ignoring 20+ system bots of Telegram
- Logging ignored bots
4. **Mailing with balancing**
- Distributing users among several accounts
- Configurable mailing intervals
---
MODULE 4: Silent Unblock System
### **Functionality:**
1. **Automatic account recovery**
- Rotation among 6+ accounts
- Imitation of real user behavior
2. **Working with SpamBot**
- Automatic sending of `/start`
- Clicking on "Other"/"Never" buttons
3. **Cyclical operation**
- Checking and recovery every 10 minutes
- Detailed logging of the process
---
TECHNICAL REQUIREMENTS
### **Technology Stack:**
- Python 3.8+
- Telethon (for working with Telegram API)
- aiohttp (for asynchronous requests)
- DeepSeek API
- JSON for data storage
### **Architecture:**
1. **Class structure** (OOP)
2. **Asynchronous programming** (async/await)
3. **Configuration separation** (separate config.py)
4. **Modular architecture** (each bot is a separate module)
### **Configuration:**
- API_ID, API_HASH
- Phone numbers for accounts
- DeepSeek API key
- Delay settings
- Lists of ignored bots
---
## 📁 **PROJECT STRUCTURE**
```
project/
├── main_bot.py # Module 1
├── arbitrage_bot.py # Module 2
├── ai_bot.py # Module 3
├── silent_unblock.py # Module 4
├── config.py # Configuration
├── accounts/ # Account sessions
├── data/ # JSON data files
├── logs/ # Operation logs
└── users_base.txt # User database
```
---
ACCEPTANCE CRITERIA
### **Mandatory:**
1. **Stable operation** 24/7
2. **Handling all errors** (try/except)
3. **Detailed logging** to file and console
4. **Multilingual support** (Ukrainian/Russian)
5. **Asynchronous operation** of all modules
### **Testing:**
1. **Test on 10+ Telegram accounts**
2. **Test AI responses** (100+ messages)
3. **Test protection system** (questions, aggression)
4. **Test Silent Unblock** (recovery of blocked accounts)
### **Documentation:**
1. **Comments in code** (30% of the code should be commented)
2. **Installation instructions** (README.md)
3. **Configuration examples**
4. **Video instructions** for launching
---
TERMS
**Collaboration Terms:**
1. **NDA** (non-disclosure agreement)
2. **Stage payment** (30% advance)
3. **GitHub repository** with access
4. **Daily/weekly reports** on progress
---
QUESTIONS FOR DISCUSSION
1. **Do you have experience integrating with DeepSeek API?**
2. **What similar projects have you implemented?**
3. **How do you plan to test on multiple accounts?**
4. **Will there be support after project delivery?**
5. **How do you plan to protect the code from copying?**
---
PROJECT GOAL
Creation of a stable, scalable system for automating work with clients in Telegram with maximum protection against blocks and the ability to adapt to various niches (p2p, arbitration, lead generation).
**Ready to consider proposals with a detailed technical plan and realistic deadlines.**
CHECKLIST FOR EXECUTORS
**In your response, please indicate:**
- [ ] Fixed price or hourly payment
- [ ] Detailed development plan by weeks
- [ ] Examples of similar works
- [ ] Time calculation for each module
- [ ] Warranty support conditions
- [ ] Options for stage payments
---
**Looking forward to your proposals!**
*The project is serious, so I only consider adequate proposals with technical justification.*
-
Alexander, good afternoon.
I have reviewed the technical specifications. Everything is clear.
I have extensive experience in developing Telegram bots and automation.
I have already developed a similar bot with ChatGPT integration, so there will be no issues with implementation.
Deadline: 4 days
Cost: 3200 hryvnias
Write to me, we will discuss the details.
-
"Agree, automation can significantly simplify p2p arbitration! I have developed similar systems for other clients, and your project interests me particularly with the possibility of integrating AI and adapting to multi-user tasks. Thanks to my experience working with Python and Telethon, I am ready to create a stable system with all the necessary features. I suggest discussing your requirements to present a detailed development plan. I would be happy to help you achieve success with this innovative system. Can we start a dialogue?"
-
484 2 0 Hello! I have carefully reviewed the technical specifications. This is a serious architectural task. I have deep expertise in Python + Telethon (development of user bots, spammers, auto-responders).
Your requirement for Module 4 (Silent Unblock) and Module 2 (DeepSeek Defender) indicates that you need not just automation, but a ban-resistant infrastructure.
My technical vision for implementation:
Architecture (Telethon + Asyncio): I will use asynchronicity for the simultaneous operation of all 4 modules in one event loop (or distribute them into microservices via Docker if a heavy load is anticipated).
AI (DeepSeek): I will configure the context window so that the bot remembers the essence of the dialogue but does not waste unnecessary tokens. For the "Defender Bot," I will write a strict System Prompt to ensure it accurately distinguishes a lead from a hater/competitor.
…
Silent Unblock: This is the most interesting part. I will implement this by intercepting messages from @SpamBot. The script will parse buttons (Inline/Reply), click "This is a mistake" -> "No, I did not send...", fully imitating a human.
Data security: Although the technical specifications mention JSON, for a high-load system (especially for Module 1 and dialogue states), I strongly recommend SQLite (asynchronous aiosqlite). It is more reliable, the file does not get "corrupted" during simultaneous writes, and it will not affect the complexity of deployment. However, if JSON is essential, I will implement it with file lock.
Answers to your questions:
DeepSeek API: Yes, I have worked with it. It is an OpenAI-compatible API, integrates easily, works quickly, and is cheaper than GPT-4.
Analogues: I developed a network of bots for processing leads in the crypto vertical (response -> warming up -> transferring to a manager).
Testing: I have my own farm with 20+ accounts (Session+JSON) for stress tests.
Support: Yes, I provide 1 month of free bug fixing.
Code protection: I use obfuscation through PyArmor or binding to the server's HWID (licensed server) if it is necessary to transfer the code to you while protecting it from leaks.
Work plan and cost:
Week 1: Architecture, Module 1 (Ultimate) + Database.
Week 2: Module 3 (AI Chat) + Module 2 (Defender/Filter). Integration of DeepSeek.
Week 3: Module 4 (Unblocker) + Logging + Tests on account rotation.
Week 4: Final tests, documentation, video instructions, handover.
Estimated cost: $1200 - $1800 (fixed for the entire project). Timeline: 3-4 weeks.
I am ready to sign an NDA. Do you already have DeepSeek API keys and a pool of accounts for testing?
-
658 3 0 Good day to you. The project is very interesting to me ))
I have extensive experience with bots of this type. I have ready-made developed bots for mailing, bots for analyzing crypto news with Open AI API (the logic of working with deepseek API is identical), bots for crypto transactions with AI analysis, and much more. By combining all my experience into one, it will result in your solution)
I work for a fixed price.
Week 1 - Architecture + framework. Modular system, architecture, logging, possibly setting up a database instead of json and txt. Since the project is complex, I will approach the structure very carefully.
Week 2 - Module 1 (Ultimate Bot). FSM (state machine), Auto-detection UA/RU, Lead transfer to admin, tags/statuses.
Week 3 - Module 2 (Arbitr Bot). AI classification of messages, freezing dialogue, ignoring system bots, etc.
Week 4 - Module 3 (AI-bot). DeepSeek Chat API, mailing, etc.
… Week 5 - Testing, hardening, deployment. Tests, config.py, deployment instructions, etc.
14 days of bug fixes after release (within the agreed terms of reference).
My option for milestone payments:
30% advance after architecture and the first module
40% after modules 2–3
30% after tests/deployment/docs
-
2287 25 1 Hello, Python developer with over 5 years of experience. I have worked with both aiogram for creating bots and with telethon for creating a bot farm. I have a project that you can view (https://freelancehunt.com/showcase/work/boto-ferma-telegram/1998029.html), where I integrated the ChatGPT API for adaptive message distribution, using Postgres as the database. I have significant experience with P2P, having worked in P2P on Binance for a short period with a merchant account. Currently, I am working on a project where I support other software with telethon, and everything is going great. I have extensive experience, the task is clear, and it will be completed with high quality. I am providing a fixed price for full execution and similarly for the timeline. I will conduct both manual and automated testing. After completion, I will support the project in case of bugs or errors and assist with installation and configuration. Regarding protection against copying, this will be a closed project, and access will only be available to you and me. I am interested in whether you want to integrate P2P functionality and which P2P you want to work with, crypto or currency. I would be happy to discuss the details and move forward with collaboration!
-
1309 7 0 Good day, I have experience working with Python as well as Telegram bots, I can complete everything quickly and efficiently, write to discuss the details.
-
3926 15 0 Hello.
I have reviewed the technical assignment. The project looks complex and atypical for Telegram bots — it involves a system of several asynchronous modules, AI logic, and anti-ban mechanisms.
______________________
My relevant experience:
— Python — over 3 years in production, total experience — 10+ years
— Worked with Telethon on real projects (channel monitoring, parsing, asynchronous data processing)
— Experience working with multiple Telegram accounts and proxies (isolation, rotation, parallel operation from one node)
… — Integration of LLM via API (official services, providers like OpenRouter, local models via Ollama)
— Prompt engineering, sentiment analysis, context and role control
— Designing stable systems with logging, error handling, and retry logic
______________________
Approach to implementation:
— Each bot is a separate module with clear responsibility
— FSM / state machine for Ultimate Bot
— AI detection (DeepSeek) is separated into its own service
— Automatic freezing of risky dialogues + notification to the admin
— Silent Unblock is implemented not only through SpamBot but also through:
— adaptive delays
— user behavior simulation
— load control on accounts
— Configuration and logging are separated, with a focus on scalability
______________________
Estimation and work format:
Preliminary estimate: ~150–220 hours
Format: hourly payment
Rate in the range of $25/hour, depending on the final scope and complexity.
Stage-wise payment and work under NDA are possible
GitHub repository + regular updates on progress
I am ready to discuss details, clarify risks, and propose the final architecture before starting.
The project looks interesting and long-term — provided the correct technical approach.
I would be happy to receive feedback.
-
586 1 0 Hello!
The project is very clear to me: I have experience working with Telethon and developing account farms for arbitration, so I will implement the architecture based on sessions and asynchronous queues, taking into account all Telegram limits (FloodWait) and protection against spam filters. I have worked with the DeepSeek API for intent classification, so I will set up prompt engineering in such a way that the bot not only responds but also "closes" the lead according to your funnel, instantly cutting off fraud and negativity. For testing, I have my own infrastructure that allows me to check the stability of Silent Unblock and load balancing on 20+ accounts simultaneously. I am ready to provide a detailed weekly plan broken down by modules, fix the "turnkey" cost, and ensure full support after the release, including video instructions and logging of critical errors.
-
296 Good day. I am interested in your project. I have relevant experience. I can answer your questions in private messages.
-
10130 117 0 Hello.
I develop bots for Telegram using NodeJS. I am ready to take on the project. Write to me, we will discuss.
Current freelance projects in the category AI & Machine Learning
Help to elevate the business in AI issuance (GEO/LLMO project)I'm looking for a specialist who can help organize GEO/LLMO for a beauty salon. Geography - Kyiv and the region. AI & Machine Learning, Search Engine Optimization (SEO) ∙ 2 hours 45 minutes back ∙ 9 proposals |
Automatic daily data export from Google Ads to Excel for analyzing the effectiveness of advertising campaigns.
23 USD
Technical task https://www.loom.com/share/d1ffc60537394ba98785684b1e84e2c7 Write your proposals for the technical task and implementation Goal: automatic daily data export from Google Ads to Excel for analyzing the effectiveness of advertising campaigns. 1. Data source 1.1.… AI & Machine Learning, Contextual Advertising ∙ 3 hours 20 minutes back ∙ 12 proposals |
Automation of creating advertising creatives.There is a solution in the form of Higgsfield AI + nodes that generate quite close to the idea based on the prompt, paired with almost frame-by-frame editing, everything works well, except for one thing, which is the censorship on the use of celebrity faces and such. An… AI & Machine Learning, Bot Development ∙ 3 hours 38 minutes back ∙ 7 proposals |
Creation of automatic video shorts
45 USD
What exactly do I need I need to create a light AI agent or content factory 1. This is idea creation Reference https://youtube.com/shorts/omgVX00yaD8?is=1Gz_TUUlftZnkxQv 2. This is text creation 3. This is background creation + text 3. Animating the background while keeping the… AI & Machine Learning, Data Parsing ∙ 13 hours 17 minutes back ∙ 12 proposals |
Automation of product publication: Google Sheets → Prom.ua via Make.comMake.com scenario: Telegram channel → Cloudinary (photo) → OpenAI (product card generation) → Google Sheets (draft). Moderation and automatic publication on Prom.ua need to be completed. What is ready: • Make scenario with all modules • Google Sheets with product data… AI & Machine Learning ∙ 15 hours 29 minutes back ∙ 14 proposals |