You will get an AI chatbot trained on your documents and data with citations
Top Rated
Project details
Generic chatbots guess. A RAG chatbot answers from YOUR documents, whether manuals, contracts, product docs or an internal knowledge base, with citations pointing to the source, so your team and your customers get accurate answers instead of hallucinations.
I have built RAG systems in production (Qdrant vector search, hybrid retrieval), including a document-trained assistant for an electronics business. I focus on the unglamorous part that makes RAG actually work: chunking strategy, retrieval quality and evaluation.
What you get:
• A chatbot trained on your own documents (PDF, website, Notion, Drive and more)
• Citations pointing to the source passage, so trust is built in
• A web widget or an API, and WhatsApp or Slack as an add-on
• Standard tier: an admin panel so you can add and update documents yourself
• Advanced tier: multi-source sync, an evaluation set proving accuracy, deployment and support
Process: you share your documents and 10 real questions your users ask, I build and tune retrieval until those questions are answered correctly, then we deploy and I hand everything over.
English and French corpora handled natively.
I have built RAG systems in production (Qdrant vector search, hybrid retrieval), including a document-trained assistant for an electronics business. I focus on the unglamorous part that makes RAG actually work: chunking strategy, retrieval quality and evaluation.
What you get:
• A chatbot trained on your own documents (PDF, website, Notion, Drive and more)
• Citations pointing to the source passage, so trust is built in
• A web widget or an API, and WhatsApp or Slack as an add-on
• Standard tier: an admin panel so you can add and update documents yourself
• Advanced tier: multi-source sync, an evaluation set proving accuracy, deployment and support
Process: you share your documents and 10 real questions your users ask, I build and tune retrieval until those questions are answered correctly, then we deploy and I hand everything over.
English and French corpora handled natively.
AI Algorithms
Large Language Model, Multimodal Large Language Model, Transformer ModelAI Applications
AI Chatbot, Conversational AI, Machine Translation, Natural Language Understanding, Text RecognitionAI Development Language
PythonAI Models
BERT, ChatGPT, GPT-4, LLaMAWhat's included
| Service Tiers |
Starter
$149
|
Standard
$249
|
Advanced
$449
|
|---|---|---|---|
| Delivery Time | 4 days | 6 days | 10 days |
Number of Revisions | 1 | 2 | 3 |
AI Model Integration | |||
Batch Normalization | - | - | - |
Database Integration | |||
Detailed Code Comments | - | - | |
Image Upscaling | - | - | - |
MLOps | - | - | - |
Model Deployment | - | - | - |
Model Documentation | - | - | - |
Model Monitoring | - | - | |
Model Testing & Optimization | - | ||
Model Tuning | - | - | - |
Natural Language Processing | |||
NLP Tokenization | - | - | - |
Pre-Training | - | - | - |
Prompt Engineering | |||
Setup File | |||
Source Code | - | - |
Optional add-ons
You can add these on the next page.
WhatsApp or Slack channel
(+ 2 Days)
+$79
Extra data source connector
(+ 1 Day)
+$69Frequently asked questions
4 reviews
(3)
(1)
(0)
(0)
(0)
This project doesn't have any reviews.
LH
Ludovic H.
May 1, 2026
AI Automation Specialist (n8n / WhatsApp / Calendar / Tasks)
Nafiou did a very good job. He is incredible. He handled the entire project seriously, respecting the initial brief. We worked on an AI agent in the field of tourism. He quickly understood the relevance of the project and became fully committed. After a week, the V1 was released. It was operational but not enough completed. So we reworked the agent taking into account the returned answers and Nafiou managed the entire project perfectly. He was very involved and very respectful. He advised me on the stack to have and on hosting my solution. I recommend it 100%!!
SS
SymbioteK S.
Jan 26, 2026
Développeurs Laravel pour projets avancés (Streaming, LMS, Tri des emails)
De bonne volonté et suffisamment compétent, tout en prévoyant un superviseur capable de réviser son travail, par rapport à sa jeune expérience actuellement, le temps qu'il devienne senior.
DF
Doru P. F.
Sep 26, 2025
Instagram n8n workflow auto-posting based on digital avatars
reliable and supportive
DF
Doru P. F.
Aug 27, 2025
n8n Workflow for Automated Facebook Page Posting with Telegram Approval
About Nafiou
AI Automation Engineer | n8n, Claude Code, AI Agents, RAG, MCP, SaaS
100%
Job Success
Cotonou, Benin - 1:45 pm local time
I'm Tino, an AI Automation Engineer. I build AI agents, automation workflows and full SaaS systems that run in production — not demos that work once.
30+ projects delivered, 100% Job Success, Top Rated. Clients come to me when the AI part has to actually work: real users, real payments, real failure handling.
— 𝐖𝐇𝐀𝐓 𝐈'𝐕𝐄 𝐒𝐇𝐈𝐏𝐏𝐄𝐃 —
✅ Rostee — AI Chief of Staff on WhatsApp. Voice or text in, tasks and meetings out. Live with Stripe payments and n8n workflows (rostee.so)
✅ Journy — multi-tenant AI concierge for hotels, one WhatsApp number per property
✅ Sambaty — logistics SaaS between France and West Africa, WhatsApp AI + Stripe
✅ A custom MCP server exposing 110+ tools to Claude Code and n8n — messaging, email, CRM, content, all callable by an agent
✅ Tourism agency (France): AI agent handling WhatsApp requests, calendar and tasks end to end. Under 4 weeks — 5.0★ "He is incredible. He handled the entire project seriously."
✅ Content creator: manual Instagram posting replaced by a hands-free daily pipeline — 5.0★
✅ Facebook page owner: n8n + Telegram approval flow, one tap publishes. 10 days — 5.0★
— 𝐖𝐇𝐀𝐓 𝐈 𝐃𝐎 —
• AI agents — Claude Code, tool calling, multi-step reasoning, state that survives restarts
• n8n automation — webhooks, schedules, multi-service orchestration, retries and error handling
• MCP servers — new to MCP? It's how you give an AI agent real tools (your CRM, your database, your inbox) instead of stuffing everything into one prompt. I build those servers.
• RAG chatbots — your documents, your data, answers with citations instead of guesses
• WhatsApp systems — Cloud API, Evolution, Waha, voice transcription, multilingual
• Full-stack SaaS — TypeScript/Hono, React/Next.js, Supabase, Stripe LIVE, Docker, Coolify
• Rescue work — Lovable, Bolt or AI-generated apps stuck at 80%. I finish them and put them in production.
— 𝐇𝐎𝐖 𝐈 𝐖𝐎𝐑𝐊 —
You get a written scope before I touch anything, async updates as I go, and a system you own: your repo, your keys, your infrastructure. No black box, no lock-in.
Demos are easy. Production is where things break — error handling, rate limits, retries, monitoring. That's the part I'm paid for.
🇫🇷 Fully bilingual French/English — specs, calls and documentation in either language. Based in Cotonou, Benin (GMT+1, same timezone as Paris), with evening availability to overlap US hours.
𝐑𝐞𝐚𝐝𝐲 𝐭𝐨 𝐬𝐭𝐚𝐫𝐭? Send me a short message describing what you're trying to automate. You'll get a clear plan, an honest feasibility answer, and I can usually start within 48h.
Steps for completing your project
After purchasing the project, send requirements so Nafiou can start the project.
Delivery time starts when Nafiou receives requirements from you.
Nafiou works on your project following the steps below.
Revisions may occur after the delivery date.
Documents & questions
You share your documents and the 10 real questions users ask. I analyse the corpus and design the chunking and retrieval strategy.
Build & accuracy tuning
I build the RAG pipeline with citations, then tune retrieval until your 10 reference questions are answered correctly and consistently.