Create a GPT assistant for texts
It is necessary to create a GPT assistant that:
Takes as input an example of correspondence (template) between a client and a sales manager.
Extracts information specifically about our product from project files (legend, product description, etc.).
Based on the template generates new correspondence — in the same style, tone, and format as in the example, but with our project and our product.
I am unable to come up with a good prompt. I need someone with experience who has already done such tasks and understands how to work with this!
-
2 days75 USD2 days75 USD
Good evening!
I have an assistant written in Python that works directly through GPT using an API key, meaning you enter questions in the program that are then automatically sent via the API and after that return a response from the AI, and it is also voiced but not very well. If you are interested, please reply.
-
1 day150 USD
25 1 day150 USDI will do high-quality work as a professional in my field.
-
2 days75 USD
834 8 0 2 days75 USDI can help in creating this project. You do not need a prompt in this case, unless you have very few products. You need to use the Vector Knowledge Base text-embedding-3-small, text-embedding-3-large from Open AI API and provide answers to the client based on it. I can implement this.
Please write to me for clarification of additional details because it is unclear where the messages should actually be sent: to WhatsApp, Telegram, your website, or to a Telegram bot. Should it be a desktop program in which the text data is generated? I can implement everything, but I need to understand what you need.
I am currently writing the price and deadlines as a reference because I do not understand what the final result you need is.
-
1 day75 USD
452 1 day75 USDHello! I have experience working with GPT models and creating chat assistants, particularly with generating texts in a specified style and tone. I understand the importance of considering information from the legend and product description to create relevant and natural correspondence. I am ready to perform this task with high quality, set up prompts, and adapt the result to your project. Feel free to reach out — I will be happy to help!
-
3 days75 USD
1347 36 0 3 days75 USDGood day, I can create such an application for you. I have experience integrating with the OpenAI API and can format the application as needed. I can start right now. I hope for cooperation!
Linkedin: https://www.linkedin.com/in/maximdziuba/
-
1 day75 USD
267 1 day75 USDSYSTEM
You are Sales Dialogue Synthesizer for a specific product.
Your job: from the user’s inputs, (A) extract product facts only from provided files; (B) infer the style rules from a sample client–manager chat; (C) synthesize a new dialogue in the same style, tone, format, and length as the sample, but using the extracted product facts.
Constraints:
Do not use outside knowledge. If a fact is missing, keep a neutral phrasing or mark [MISSING].
Never invent prices, specs, or guarantees; only rephrase/compose from extracted facts.
…
Maintain the same speakers’ labels, emoji usage, politeness level, message count rhythm, and formatting (bullets, bold, links) as in the sample.
Keep language the same as in the sample.
No chain-of-thought in the output; return only the final dialogue.
USER (шаблон запроса к модели)
You will receive three inputs:
PRODUCT_FILES (verbatim text):
{{PRODUCT_FILES_PLAIN_TEXT}}
TEMPLATE_DIALOGUE (the style to mimic):
{{TEMPLATE_DIALOGUE_TEXT}}
SCENARIO_BRIEF (who the client is and what they ask now):
{{SCENARIO_BRIEF_TEXT}}
Your tasks:
Step 1 — Extract product facts to JSON (do not output explanations):
Return a JSON object named product_brief with fields:
{
"name": "", "value_prop": [], "key_features": [],
"target_users": [], "constraints": [], "pricing": "",
"channels": [], "faq": []
}
Fill arrays with concise bullet strings. If unknown, use [] or "".
Step 2 — Derive style rules from TEMPLATE_DIALOGUE (do not output explanations):
Return a JSON object named style_rules with fields:
{
"tone": "", "formality": "", "emoji_usage": "",
"avg_msg_length": "", "formatting": [],
"opening_pattern": "", "closing_pattern": "",
"manager_persona": "", "client_persona": "",
"message_structure_notes": ""
}
Step 3 — Synthesize the new dialogue
Using only product_brief and style_rules, generate a new client–manager conversation in the same style and format as the template, tailored to SCENARIO_BRIEF.
Requirements:
Keep the same speaker labels and overall rhythm (number of turns ±1).
Preserve formatting patterns (bullets, bold, links) and emoji frequency.
Ensure every claim about the product is supported by product_brief.
If a needed fact is missing, write a neutral line or [MISSING] instead of guessing.
Final output structure (strict):
=== product_brief ===
{...JSON...}
=== style_rules ===
{...JSON...}
=== dialogue ===
-
1 day75 USD
516 10 3 1 day75 USDHello.
I am ready to complete your project.
I will be very happy to collaborate.
-
1 day111 USD
631 5 0 1 day111 USDGood afternoon
I am ready to complete the task, feel free to contact me
--------------------------------------
-
3 days75 USD
3090 36 0 3 days75 USDGood evening, Mark! I am ready to develop an effective prompt for generating correspondence according to your specifications :)
Current freelance projects in the category AI Content Creation
Improvement of the catalog for the websiteFor the catalog website, improvements need to be made: (the work can be divided into stages or taken by qualification part) 1. In the admin panel, the ability to upload SQL files for updating prices (in the archive it is price_balluff_2026_for_site.sql). (currently they are… AI Content Creation, Web Programming ∙ 3 hours 50 minutes back ∙ 25 proposals |
Integration of text into ready-made lyric video
22 USD
What is needed: It is necessary to integrate the song lyrics into a ready lyric video. Main task: The appearance of the text should be creative, interesting, and non-template. Most importantly, the text must be synchronized with the vocals. An example of text formatting:… AI Content Creation, Transcribing ∙ 4 hours 47 minutes back ∙ 2 proposals |