AI agent with integration into Telegram (stage 1, non-standard logic)

Hello!
I can implement it. I will set up correct dialogue processing, context preservation, and predictable behavior. I have experience working with OpenAI, Telegram bots, and prompt engineering.

How I will implement it: I will move it to a separate configuration, optimize the structure (breaking it into blocks/roles), control the length of the context and dialogue history to avoid exceeding limits and losing logic. I will also anticipate error handling and stable interaction with the API.

Deadline: approximately 2-4 days.

Please let me know if it is necessary to limit the number of requests/users or to create a basic admin panel for usage control?

Show original

Translate

Good day!
1. OpenAI API - I work regularly: chat completions, role management, generation parameters. I understand the difference between "connecting ChatGPT" and building a managed agent.
2. Telegram bots - I developed using aiogram 3, with multi-step dialogues, state storage, and integration of external APIs.
3. Working with long prompts
I store the system prompt in a separate file outside the code - it is loaded at startup and passed to the system field with each request, regardless of the dialogue history. The history is maintained separately in SQLite with a sliding window - the agent does not lose instructions even in long chats.
The architecture is modular: memory, agent logic, handlers - separate modules. In the future, it is easy to connect new tools or change components without rewriting the entire bot.
4. Deadline - 3-4 days after receiving the prompt and access.

Show original

Translate

Andrey, I have reviewed the description — the task is clear and technically precise.

My stack: aiogram 3.x (Python) + OpenAI API — daily toolkit. I will implement the following: I will set up a bot on webhook (aiogram 3.x + FastAPI), connect OpenAI through the official SDK, configure the transmission of a 12k-character system prompt in the `system` role of each request without truncation. The conversational history will be accumulated in the `messages` array with a context depth limit — so that during long sessions we do not exceed the model's token limit. I will save the context through Redis or in-memory with TTL — we will clarify at the start. Regarding stability: API error handling (rate limit, timeout, retry with backoff), logging, deployment on Railway with auto-restart.

I will separately go through the agent's logic: I will check that the constraints and behavior from the prompt work predictably, without "breaks" in instructions.

One question: is the system prompt already finalized, or will it be adjusted during the process? This affects the number of testing iterations.

Show original

Translate

Andrey, good afternoon. There is a case in the portfolio - a recruiting bot for Telegram with AI integration and questionnaire processing. I am writing in Python using the aiogram 3 library. I fix the system prompt at 12k characters in the system message with each request, and I store the context through a sliding window - the agent does not fall out of character. Deadline: 3 days. Budget: 3000. Write to me.

Show original

Translate

I have experience in developing AI models and improving them through prompts, and I have also worked with the OpenAI and Gemini APIs. I have been working with Telegram for a long time as well. I am ready to complete the task quickly and efficiently; after discussing the details, I will suggest something that might interest you regarding how to improve all of this! Write to me.

Show original

Translate

Greetings. I understood the task: a Telegram bot is needed that not only proxies requests but also strictly maintains a role based on a long 12k prompt, without deviating from the given context and logic.
I will respond point by point:
1. Experience with OpenAI API: I work closely with the API, including fine-tuning parameters (temperature, top_p) and the architecture of sending requests to keep the agent within system constraints.
2. Experience with bots: I have developed complex logic for Telegram bots in Python. For example, I created a bot for automating arbitration with optimized signal processing and a response time of less than 1 second. The bot operates seamlessly thanks to stable performance on a VPS. I have also implemented complex real-time alert systems in Telegram for risk managers.
3. Working with long prompts: 12,000 characters is about 3,000-4,000 tokens. The system prompt will be sent with the role of system with each API call to ensure the AI does not deviate from instructions. Basic context preservation will be implemented through a sliding memory window (storing the last N question-answer pairs). This will prevent context window overflow (Context Window Limit) and save you money on tokens while maintaining the thread of dialogue.
4. Deadlines: 2-3 days to implement a stable version with context caching. Which specific OpenAI model do you plan to use for testing (gpt-4o-mini, gpt-4o)? I suggest discussing the details in private messages.

Show original

Translate

Good day!
I have reviewed your project — the task is clear, and I resonate with the approach of building a controlled AI agent rather than just a "wrapper" over ChatGPT.
1. Experience with OpenAI API:
Yes, I have experience integrating the OpenAI API, including working with system prompts, processing dialogue context, and controlling model behavior.
2. Experience with Telegram bots:
I have developed Telegram bots of varying complexity:
- bots with subscriptions and payments
- automation of business processes
- bots with role and state logic
… - integrations with APIs and handling a large flow of users
3. Working with long system prompts (~12k):
I implement this through:
- clear separation of system/user/assistant messages
- optimization of prompt structure (if necessary — decomposition into blocks)
- control of context length (token management)
- maintaining a short dialogue history (memory) for stability
- additional checks to ensure the agent does not "deviate" from logic
If needed, I can suggest improvements in prompt engineering for more predictable behavior.
4. Timelines:
⏱ 3–5 days for MVP (a working bot with stable logic)
I will also ensure:
- stable operation without crashes
- clean and scalable code
- the ability to easily expand the project in subsequent stages
I approach such tasks not just as a performer, but as a developer who thinks about architecture and the future development of the product.
I am ready to discuss the details and start working.

Show original

Translate

Good day!
I have 3+ years of experience in developing Telegram bots, AI automations, and integrations with the OpenAI API, including projects with non-standard logic, system prompts, and controlled agent behavior.
I can implement a Telegram bot with OpenAI integration, correctly embed a large system prompt, set up message reception and sending, basic dialogue context preservation, and stable operation without crashes and failures.
I suggest discussing the details, as well as the cost and timelines in private messages.

Show original

Translate

Hello.

I am developing bots for Telegram using NodeJS. I am ready to take on the project. Write to me, and we will discuss.

Show original

Translate

Hello!
1. Yes, I have worked with the OpenAI API — integrations with agents, call analysis, data categorization through GPT.
2. I have created Telegram bots — I will show examples in private upon request.
3. I implement a long system prompt through the system role in each request, the dialogue context — through a sliding window to avoid exceeding the token limit. The prompt remains unchanged regardless of the dialogue volume.
4. Deadlines: 3-7 days after receiving the prompt.

Additionally, I suggest adding error logging and fallback responses — this is the stability you are requesting. Free of charge.
I am waiting in private.

Show original

Translate

Hello. I can create this AI agent. I have some developments. If you are interested - write to me, we will discuss.

Show original

Translate

has experience working with everything listed above.
long prompt is not a problem.
2-3 days
budget for MVP

Show original

Translate

Good evening. Let's go point by point:
1. Yes, I have extensive experience with the integration of various AIs, including Open AI.
2. I have created different bots and various projects. Nothing demonstrates its ability to "think" except for the development of non-standard projects. I have made copies of bots, neural network bots, bots that consist of a collection of neural networks with specific settings.
3. Most likely, it is best to create such prompts through a "rule" in a specific neural network. When connecting via an API key (and not only), you can set a rule that the neural network will follow.
4. UP TO 3 days. However, usually, clients who respond quickly receive the work on the day of the order.

Show original

Translate

Hello! I can implement a Telegram bot with OpenAI API integration and stable operation of the AI agent according to your system prompt. I have experience in creating Telegram bots using Python/aiogram, connecting APIs, processing messages, maintaining context, and managing dialogue logic. I have also worked with bots where it was important not just to respond, but to adhere to a given scenario and constraints. I will implement a long system prompt separately as the main instruction for the agent, keeping it stable in requests to OpenAI, and I will manage the dialogue context through a minimal message history so that the bot does not "get confused." I am ready to carefully analyze your logic and propose solutions so that the bot operates not as a regular ChatGPT, but specifically as a controlled AI agent.

Show original

Translate

Hello. Interesting project. At least the prompt for 12k caught my interest. I will write in PHP, and a Telegram bot and AI integration. Feel free to contact me.

Show original

Translate

Yes, I have experience working with OpenAI API and Telegram bots.

In my portfolio, there is a project - a bot for a salon, which operates on a similar principle (dialogue processing, communication logic, response automation), so I understand how to properly build the architecture of such a solution.

Regarding your project:

How I will implement work with a long system prompt:

I will extract the system prompt into a separate structure (to manage and update it conveniently)
… I will use it as the base context for each dialogue
I will add length control for the context (to avoid overflow)
If necessary, I will break the logic into parts for more stable operation

Dialogue context:

I will implement basic storage (for example, through SQLite or JSON)
with a history limit for stability

Technologically:

Telegram Bot API + OpenAI API
processing incoming/outgoing messages
error and crash protection

Show original

Translate

Let's go, I'll do it by tomorrow evening, but this is exactly the MVP to implement the idea, not an instantly scalable product, etc. Keep this in mind, I'll make it for you in Python with an OpenAPI schema that will be built upon push, hosting on Railway for $10 a month. If you need it, write to me now because I don't have much free time.

Show original

Translate

"Here I just formed the response through a similar bot of mine)) - besides this text"

1. I have experience integrating the OpenAI API in several projects, including the implementation of complex management logic and processing large system prompts with context consideration.
2. I have developed Telegram bots using the Bot API – from simple chatbots to complex CRM integrations and automation. For example, I implemented bots for customer support and data parsing from chat.
3. I handle long prompts by storing and dynamically forming the context of messages. To optimize, I use a combination of the system prompt with the user request, as well as minimal but sufficient context from the dialogue history. If necessary, I break the prompt into logical parts and embed them as several parameters in the request to OpenAI.
4. The preliminary deadline for completion is 7-10 days, depending on the clarification of details regarding logic and the need for stability testing.

I base my work on experience in PHP, Laravel, and Docker for stable and scalable bot deployment. I am ready to offer additional ideas for improving the interaction between the bot and the user and maintaining context in the future.

Show original

Translate

Hello! I am interested in your project. I have extensive experience in:

📊 Data processing: working with databases, structuring and analyzing information, automating the processing of large volumes of data, import/export and validation;
🤖 Automation and emulation of user actions, development of bots of varying complexity;
⚡️ Asynchronous and multithreaded parsing: collecting and processing data with performance optimization;
🔍 OCR and text search: recognition and structuring of information;
🖼 Media processing: working with images and multimedia;
🖥 Software development, desktop applications, system services and utilities;
📱 Mobile development: native and cross-platform applications;
… 🌐 Working with APIs and third-party services: integration, automation, and data exchange;
🗣 Translation and text processing: automation of translation, working with language models and text analytics;
🤖 AI/LLM solutions: integration and use of artificial intelligence, working with language models and automating intelligent processes.

I will complete the work quickly and efficiently. Contact me to discuss the details and deadlines of the project!

Show original

Translate