AI-agent system for marketing and sales of an educational project

Budget: 27000 UAH Deadline: 28 days

Hello. The described 6 blocks optimally fit into a microservices architecture with orchestration on Python. For content generation, transcription, and lead routing, I will use autonomous AI agents based on the OpenClaw architecture, connecting with the Meta API and CRM via REST, and I will derive hypothesis analytics through PostgreSQL. I have practical experience in developing AI lead generation systems for the B2B sector in Germany, as well as complex backend systems with multi-accounting and fast data routing without losing context. Which CRM do you plan to use, and are there already ready-made templates in Canva/Figma for setting up automatic API assembly?

Budget: 10000 UAH Deadline: 30 days

Hello! This is my profile — I build AI automation for marketing and leads. Vision based on your specifications:

Architecture: central orchestrator (n8n or Python) with task queues and a mandatory "human approval" gate before publication — a cross-cutting requirement, I will implement from day one. Data (leads, hypotheses, content statuses) — in one database (Postgres).

Agents by blocks:
1. Generation — LLM agents pull the project knowledge base (programs, transcripts) and provide texts for each channel + options for hypotheses.
2. Production — assembly of PDF lead magnets (Canva/Figma), visuals for Meta with resizes, landing pages on Framer CMS.
3. Video→blog — transcription (Whisper) → cutting into articles → publication in Framer CMS.
4. Meta — uploading texts/creatives to the dashboard via API + tagging by hypotheses.
5. CRM/routing — segmentation of leads in the TG bot + transfer and auto-tagging in CRM.
6. Analytics — hypothesis dashboard (costs/results) + periodic AI analysis "what to scale / what to stop".

I work in stages: Stage 1 — orchestrator + generation + review (core); then production and video→blog; next Meta+CRM integrations; finally analytics. I fix the cost by stages; the estimate for Stage 1 — $800-1200, the full system scales from there.

Questions regarding the specifications:
— what CRM is used?
— blog on Framer: does the system create the section itself or is this a separate stage (your open question) — I would separate it out, it's more reliable.
— in what format are the course materials — text, video, audio?

Portfolio: https://god-devils.netlify.app/ Expand

Budget: 27000 UAH Deadline: 30 days

I can design and implement an AI service system for the entire marketing pipeline: content generation, transcription, PDF preparation and creatives, Framer, Meta Ads, Telegram, CRM, lead routing, and hypothesis analytics. In each critical process, I will include a separate review and approval stage by a person.

I would start with a process map and MVP: content sources → generation → approval → publication → lead → CRM → results by hypothesis. After that, the system can be expanded with separate modules without restructuring the entire architecture. Budget and timelines will be discussed in personal correspondence.

Where the most time and money are currently lost: content gets stuck in preparation and approval, leads are lost between Telegram and CRM, or the team does not see which hypothesis actually generates sales?

Budget: 2000 UAH Deadline: 1 day

Hello!
I am Oleg, project manager at Business Atlas (partners of n8n). Our team has extensive experience in developing multi-agent AI systems for marketing (50+ projects).
We will build an autonomous ecosystem on self-hosted n8n with a mandatory approval block (Human-in-the-loop) via a Telegram bot before publishing or launching ads.
Briefly about the architecture and stages:
1. Architecture (5 AI agents)
Transcriber Agent: Whisper API + Claude 3.5 Sonnet for processing videos into articles and posts on Framer CMS.
Creator Agent: Text generation for landing pages, email chains, Telegram, Meta Ads, and lead magnets (PDF).
Media Buyer Agent: Publishing in Meta Ads and Framer CMS + sending previews in the Telegram bot for approval with Accept or Edit buttons.
Router Agent: Segmentation and auto-tagging of leads in the Telegram bot and CRM (KeyCRM, HubSpot, etc.).
Strategist Agent: Data collection in Looker Studio / Google Sheets and AI analysis of ROMI and hypotheses.
2. Relevant experience
YouTube Automation: Transcription -> script generation -> illustration creation.
Multi-agent system: 4 AI agents for news analysis, post writing, and visual creation.
Segmentation of a database of 30,000+ leads by 12+ criteria.
3. Stages, timelines, and costs
Discovery & Foundation (n8n, Supabase, architecture): 5-7 days — $600 - $900.
Content & Video Engine (content and video in articles): 10-15 days — $1,500 - $2,500.
Distribution & CRM (Meta Ads, Framer, approval bot, CRM): 14-20 days — $1,500 - $2,500.
Analytics & AI Insight (dashboards and AI analysis): 7-10 days — $600 - $1,200.
Overall: 6-8 weeks | $4,200 - $7,100.
4. Clarifications
What CRM are you currently using?
Do we create creatives based on your templates (Figma/Canva) or set up AI image generation?
What volume of video is planned to be processed weekly?
Message me privately — we will discuss the details. Expand

Budget: 20000 UAH Deadline: 18 days

Good day! I have built similar connections — several LLM agents for different formats (landing pages, email chains, posts in Telegram), lead segmentation in CRM, and a mandatory step of human review before publication. Regarding the stack, I will decide based on your data: do you already have a CRM to store and segment leads, and where are the course materials located so that the agents can pull context from them? I would do it modularly — to add new agents without reworking the core. Approximately 2-3 weeks depending on the number of agents.

Budget: 27000 UAH Deadline: 21 days

Alexandre, I would not build six independent "magic agents." It is more reliable to create one managed conveyor with separate services, a shared data model, and mandatory human approval before any publication or ad launch.

Architecture of the first stage:
1. Postgres as the source of truth for materials, content versions, leads, hypotheses, and results.
2. Python/FastAPI orchestrator with a queue, retries, and a log of each step.
3. RAG layer over course programs and transcripts.
4. Separate format generators, but a single screen or Telegram gate "approve / return for revision."
5. Connectors to CRM, Telegram, Framer, and Meta as isolated modules.

Relevant experience: promAI generates content in batches through Gemini with a queue and progress, youtube_helper transforms Telegram discussions into an RAG database with daily updates, fasttravel has data pipelines, Telegram notifications, Postgres, and full Docker/CI/CD.

The rate of 27,000 UAH and 21 days covers the working MVP: content from materials, transcription, human review, one publication channel, lead segmentation into one CRM, and a basic hypothesis dashboard. The remaining integrations will be added in separate controlled stages after MVP verification.

I do not have direct production experience specifically with n8n, so I propose a code-first Python core. What CRM, Telegram bot, and access to Framer/Meta do you currently have? Expand

Budget: 1200 UAH Deadline: 7 days

Good day, Oleksandr!

Your task of automating the marketing and sales process through an AI-agent system is relevant. I understand the needs for content generation, lead segmentation, and analytics. I have experience in implementing similar projects, which allows me to create an effective system.

⚠️ To develop a specific SEO strategy, it is necessary to study your project and its goals. I propose the following work structure:
1. I will familiarize myself with your materials and requirements. A link to the website or other resources is needed.
2. Please specify which product categories are a priority and from which regions you expect clients.
3. After the analysis, I will provide an action plan. If it suits you, we will discuss the details of cooperation.

👉 There are two possible scenarios: your website already has positions on pages 1-5, or it requires comprehensive optimization. In the first case, we will work on increasing conversions; in the second - from scratch.

💰🕗 The preliminary cost of my services is from $400 per month. To achieve results for keywords with low competition, it takes 2-4 months; for high competition - 5-8 months.

Please take a look at my cases that demonstrate successful results:
- 243 conversions per month: insurance services - https://freelancehunt.com/showcase/work/243-konversii-mesyats-uslugi-strahovaniya/1865495.html
- 178 applications in 1.5 months: lawyer services in London - https://freelancehunt.com/showcase/work/178-lidov-za-mesyatsa-uslugi/1865493.html
- 263 thousand organic users in the first year of the website's operation - https://freelancehunt.com/showcase/work/263-tyis-organicheskih-polzovateley-za-1y/1865385.html

I suggest discussing your project in private messages to consider the goals and action plan in more detail. This will allow us to establish effective cooperation. Expand

Budget: 7000 UAH Deadline: 30 days

I build exactly such systems. I have worked with AI agents for lead qualification, automating communication with clients, and real-time reporting for management.

This is particularly relevant for an educational project because a quick response to a new lead and proper segmentation are important. I implemented something similar for an online school where the response time was reduced from 3 hours to 2 minutes.

Please tell me more about which specific processes you want to automate first and what stack is currently being used?

Portfolio: https://interesting-galley-0eb.notion.site/Portfolio-ff53a82144bf8251bc1581a6cef25f7d

Budget: 10000 UAH Deadline: 10 days

Hello, Oleksandr! I am building such systems on n8n + OpenAI/Claude API: separate agents for each block - content generation (landing pages, email chains, Telegram, Meta ads with options for hypotheses), lead segmentation in CRM, and hypothesis analytics. There will definitely be a review stage: no text or campaign goes live without human approval (approve/rework buttons in Telegram or the dashboard). The solution is scalable and maintainable - I document all scenarios. I will first propose the architecture for approval, then assemble it block by block with a demo after each stage.

Budget: 20000 UAH Deadline: 15 days

Hello!
It seems that your main task is not just to create a set of AI agents, but to build a unified marketing ecosystem where content, advertising, CRM, and analytics work as one process with human review at every critical stage.

I specialize in developing AI marketing systems and automations based on Make.com, OpenAI, CRM (Zoho/Pipedrive), ActiveCampaign, and Looker Studio. I have implemented projects involving AI lead qualification, automated email funnels, CRM analytics, and end-to-end integrations that have reduced lead processing time and built a fully automated marketing process.

I would suggest implementing the project in stages: first, design the architecture of the agents and data flows, then implement the key modules (content, CRM, Meta, analytics), and after that, scale the system with new AI scenarios.

Could you please let me know if the CRM, Telegram bot, and stack for Framer/Meta API have already been chosen, or if they also need to be selected as part of the architecture?

Budget: 20000 UAH Deadline: 10 days

Good evening!

I professionally develop bots in Telegram and web applications in Python. I am ready to implement such an AI agent system for you with integration of Meta API, CRM, Framer, automatic PDF generation, and hypothesis analytics + deploy it on a server.

Message me privately, and we will clarify the details.

Budget: 8000 UAH Deadline: 7 days

Good day, Oleksandr.

My name is Ivan, I am the founder of Tyap Lyap AI. We develop AI automation and integrations for businesses.

We already have a relevant case — the Tyap Lyap content conveyor: research → generation → visual render → Gemini QA → Telegram approval → Publer.

I would not create a separate AI agent for each block.

I see the architecture like this:
- n8n — process orchestration, PostgreSQL/Supabase — data and shared knowledge base;
- AI modules — content generation, video transcription, and results analysis;
- Framer, Canva/Figma, and Meta API — creation of materials and drafts for campaigns;
- Telegram or a separate panel — mandatory approval before publication;
- CRM module — lead segmentation, tags, and linking to source, offer, and hypothesis;
- dashboard — expenses, leads, conversions, and AI recommendations without automated decisions without human intervention.

I suggest starting with a pilot: material → content → approval → publication → CRM → analytics. Estimated timeframe — 7 working days.

The price is 8,000 UAH.
We are currently consciously building a client base and feedback — so we will do it cheaper than others. Expand

Budget: 25000 UAH Deadline: 20 days

Good day! We have experience in developing AI agents for automating the sales funnel. We implement the system through the integration of LLM with your CRM, which will provide a complete cycle from content generation to lead qualification and analytics. This will significantly optimize the work of the marketing department. We are ready to discuss the technical details and architecture of the solution.

Budget: 14000 UAH Deadline: 10 days

😀 Hello, Oleksandr!

This is exactly the kind of systems I create — AI agents + n8n orchestration + CRM/Telegram/Meta integrations for a marketing-sales pipeline, not just a set of disparate scripts.

Here's how I will do it (modularly, as per your specifications):
1) Orchestrator on n8n, LLM layer — OpenAI/Claude (model selection based on the task — for higher quality/costly content use Claude, for mass operations use cheaper OpenAI)
2) Content agent: generation for channels (email/Telegram/Meta/blog) with a project knowledge base in a vector database (Supabase pgvector)
3) CRM agent: lead segmentation, auto-tagging, status synchronization
4) Video agent: transcription (Whisper) → structured articles/theses
5) Analytical agent: hypothesis dashboard (launch/stop campaigns) based on metrics
6) Human approval stage before publication — MANDATORY, as you suggested

I propose starting with Stage 1 (content + CRM + basic orchestration) — 10 days within your budget, then based on the results — Stage 2/3.

🎥 Profile: Freelancehunt

Question: what CRM is currently being used and is there already a Telegram bot/Framer project, or are we starting from scratch? Expand

Budget: 1000 UAH Deadline: 1 day

Good day!

The project is very interesting. This is no longer a separate automation, but the construction of an AI Marketing Operating System, where AI agents work as a unified system with mandatory human review before any publication or ad launch.

How I see the architecture

I would not build one large agent.

It is optimal to divide the system into approximately 8 independent AI services that interact with each other through an orchestrator. n8n
n8n is the "brain" of the entire automation,
while the AI models are just executors of specific tasks.
1. Knowledge Agent (RAG)
2. Content Agent
3. Creative Agent

There are a few points I would like to clarify:

What CRM is currently being used?
Is there already a Telegram bot, or does it also need to be developed?
Is there a Framer project, or does the website still need to be created?
Is fully automated generation of advertising campaigns in Meta via API needed, or is preparing materials for manual launch sufficient?
Where is the knowledge base planned to be stored (Google Drive, Notion, documents, videos, etc.)?
Are there requirements for local deployment (Docker/VPS), or is the use of cloud infrastructure allowed? Expand

Budget: 1000 UAH Deadline: 1 day

Good day, Oleksandr.

The six blocks of your conveyor—from copywriting to hypothesis analytics—require an orchestrated network of agents with review gates between stages, so that no material is published without your approval. The RAG database of course materials will serve as the single source of truth for all generating agents, while separate services for PDFs, Framer landing pages, and Meta creatives will isolate complex third-party integrations—thus adding new channels does not break the entire conveyor.

Options:

≈14 days: 12000–16000 UAH
Automation of text creation and a basic content approval mechanism.
- Multichannel text generator (landing pages, emails, social media)
- Review and human content approval system
- Basic module for transcribing videos into articles
- Integration with one publication channel (e.g., Telegram)

or

≈38 days: 32000–45000 UAH
Full cycle from idea generation to automatic uploading to advertising and CRM.
- All Basic features + automatic assembly of PDFs and Framer pages
- Automatic uploading of creatives and texts to Meta Ads
- Full lead segmentation in the Telegram bot and transfer to CRM
- Hypothesis analytics dashboard with AI recommendations for scaling
- Complete integration cycle of all 6 functional blocks

+ additionally

≈27 days: 23000–30000 UAH
Maximum automation with extended support, custom templates, and deep analytics.
- Development of custom design templates for PDF/Framer
- Prioritized optimization of prompts for your Tone of Voice (up to 10 iterations)
- Extended analytics of landing page conversion with automatic A/B testing
- Support and maintenance of the system during the first month after launch
- Scalable architecture with the ability to connect new advertising channels Expand

Budget: 1500 UAH Deadline: 4 days

To ensure scalability and ease of maintenance, I suggest not building a "monolith," but rather breaking the system into microservices with orchestration:

1. Orchestration and Agents: Python (FastAPI + LangGraph / CrewAI) or n8n / Make. LangGraph is perfect for custom loops with Human-in-the-loop (where the agent waits for webhook confirmation from a human before the next step).

2. Approval Module (Human Review UI): A convenient Telegram WebApp or a lightweight panel on Streamlit/Custom Python Admin, where agents send generated text/creative options. A person clicks "Approve" or makes edits directly in the interface.

3. Generation and Transcription:

▪️Text: OpenAI GPT-4o / Claude 3.5 Sonnet (for deep copywriting and maintaining TOV)
▪️Video $\rightarrow$ Text: Whisper API / AssemblyAI for transcribing workshops.

Artifact Production:

▪️Framer CMS API, automatic blog post creation, and landing page assembly.
▪️Canva / Figma API + Bannerbear, generating static creatives and resizes based on templates.

Integrations (Meta, CRM, TG):

▪️Meta Marketing API, uploading ads to drafts with tagging by hypotheses.
▪️Telegram Bot (aiogram) + CRM (SalesDrive/HubSpot/Pipedrive): Lead routing and auto-tagging via REST API.

To ensure you don't risk your budget and to demonstrate my competence, I propose starting with Phase One:

▪️Sprint 1 (Test): Building a multimodal content generator (Block 1) + deploying the human approval module (Telegram WebApp). You will be able to generate posts, texts for Meta, and email chains and approve them with one click.
▪️Sprint 1 Budget: 1,500 UAH (special price for the first case on the platform). Timeline: 3–4 days.
▪️Sprint 2: Framer CMS integration, video transcription (Blocks 2 and 3).
▪️Sprint 3: Integration with Meta Ads, CRM routing, and Analytics Dashboard (Blocks 4, 5, 6).

I am ready to demonstratively implement the first module for you and show the orchestration scheme. Please send the details of your CRM and examples of content formats in the chat! Expand

Budget: 9000 UAH Deadline: 5 days

Hello. I recommend starting the entire conveyor with a separate pilot: data map and integrations, human approval queue, and one vertical slice "material → OpenAI draft → manual approval → Telegram/CRM," without auto-publishing and launching Meta Ads. Secrets will remain only on the server; as a result, there will be an architecture scheme, a working prototype, and a phased assessment of the remaining blocks. There is a self-initiated n8n/OpenAI/Telegram workflow; Shatalov Agency — live Node.js/CMS/PostgreSQL/Supabase, but not a case of an identical AI marketing system. Pilot — 9,000 UAH, 5 days. What CRM, Telegram bot, Framer plan, and Meta access do you already have?

Budget: 27000 UAH Deadline: 35 days

Good day! The most important node in your technical specification is not the agents themselves, but the queue for human approval: it defines the interfaces of all 6 blocks, so I would design it first (statuses draft → under review → approved → published; nothing is published without approval).

1. Architecture. Orchestration — n8n; LLM layer — Claude/OpenAI through a single gateway with a log of expenses for each hypothesis. The foundation is the project knowledge base: course programs, transcripts of broadcasts (Whisper) → chunks in Supabase/pgvector. All content agents write from it, not "from their heads" — a single tone and minimal fabrication. Modules: content (landing pages, email chains, posts, Meta texts with options for hypotheses) → production (PDF lead magnets auto-generated HTML→PDF + layouts for refinement in Canva; Framer via CMS API) → video → blog (transcription, cutting into articles) → Meta (exporting texts/creatives via Marketing API, tagging campaigns by hypotheses) → CRM/lead routing (segmentation in Telegram bot: attended/not attended the event, funnel branch, offer → tags in CRM) → analytics (expenses/results for each funnel + periodic AI report "what to scale, what to stop").

2. Cases: AI conveyor belts for content generation, Telegram bots for lead transfer, automation of Google Sheets analytics, data parsing. I will show live examples in the dialogue.

3. Evaluation (in stages, total 27,000 UAH): Stage 1 — knowledge base + content agent + approval queue + lead segmentation in the bot with tags in CRM, ~2 weeks, 12,000. Stage 2 — production (PDF, Framer CMS) + Meta integration, ~2 weeks, 10,000. Stage 3 — hypothesis analytics + AI recommendations, ~1 week, 5,000. Payment in stages, each stage — a separate work result; after answering questions, I will clarify the breakdown.

4. Questions: what CRM do you use and is there already a Telegram bot? Who on your side will approve the content (one person or a team)? Are leads currently coming in through registration on the website or through the bot?

Honestly: pixel-perfect design of creatives — is up to the designer, I provide working layouts and texts. I work in an AI-assisted format — I disclose this openly, I personally check the final logic and integrations. Expand

Budget: 15000 UAH Deadline: 10 days

Good day!

The project looks interesting, especially since you want to build a single system rather than a set of separate automations that can then be scaled.

1. How I see the architecture

I would divide the system into several independent modules so that they can be developed separately:

* Content agent — generates texts for various channels (email, Telegram, Meta, blog, landing pages) based on a knowledge base about the project.
* Production agent — collects final materials: PDFs via Canva/Figma, creatives, pages in Framer.
* Video agent — transcribes videos, forms articles, and submits them for publication.
* Marketing agent — prepares materials for Meta Ads, tags campaigns and hypotheses.
* CRM agent — segments leads, transfers them to CRM, updates statuses and tags.
* Analytical agent — collects statistics, forms dashboards, and generates recommendations for scaling or stopping hypotheses.

As an orchestrator, I would use n8n, for AI — OpenAI or Claude, for data storage — Supabase/PostgreSQL, and for context and reuse of materials — a vector knowledge base.

I would definitely add a human approval stage before publishing content or launching ads.

2. Experience

I specialize in AI automations and workflows based on n8n, integrations via API, OpenAI, Telegram, CRM, Google Workspace. I have created systems for content generation, automatic lead processing, AI assistants, and business processes. Some case studies can be viewed in my portfolio in my profile.

3. Timelines and budget

I would develop the system in stages.

* Stage 1 — content + CRM + basic automation — about 2–3 weeks.
* Stage 2 — generation of creatives, Framer, Meta Ads — another 2–3 weeks.
* Stage 3 — analytics, AI recommendations, and optimization — approximately 1–2 weeks.

After clarifying the details, I will be able to provide a more accurate budget estimate.

4. Questions

I would like to clarify a few points:

* What CRM is being used?
* Is there already a Framer project and Telegram bot?
* Is there a ready base of materials (courses, broadcasts, interviews), or does it also need to be structured?
* Do you plan to use the Meta API directly or through third-party services?

I would be happy to discuss the details and propose the optimal implementation option. Expand

Budget: 20000 UAH Deadline: 29 days

Orchestration on n8n, content generation — through Claude (API), data storage — Google Sheets/Airtable at the start (easy to migrate to CRM later). Each block is a separate workflow agent that communicates through a shared task queue:
Block 1-2 (content + final materials): generating agent on n8n+Claude → drafts of texts, PDF lead magnets (Canva API/templates), Meta creatives (resizing via scripts), landing pages on Framer CMS
Block 3 (video→blog): transcription (Whisper API) → cutting into articles → publication
Block 4 (Meta): uploading via Meta Marketing API, tagging campaigns
Block 5 (CRM/lead routing): segmentation in Telegram bot → transfer to CRM with auto-tags
Block 6 (analytics): dashboard of hypotheses + periodic AI analysis
Key — Telegram bot for review: each material before publication goes for confirmation to a person (Yes/No button), as I have already implemented in other projects.
Timeline: step by step, 4 weeks
Week 1: Block 1-2 (content + materials) + review mechanism
Week 2: Block 3 (video→blog) + Block 4 (Meta)
Week 3: Block 5 (CRM/routing)
Week 4: Block 6 (analytics) + testing of the entire system Expand