I will build an ai customer support knowledge base chatbot with citations


About this gig
Does your support team keep searching the same policies, manuals, and FAQs?
I will configure a private AI knowledge base assistant that answers from your approved PDF, DOCX, TXT, or Markdown files and shows source references. If the available evidence is weak, it gives a safe refusal and adds the question to a human review queue.
Depending on your package, you receive document upload and indexing, cited answers, a review queue, answer analytics, branding, source code, documentation, and optional deployment.
This custom prototype is built from my self-built CiteDesk AI portfolio system. It is not presented as previous client work.
Stack: Python, FastAPI, Streamlit, SQLite vector search, an OpenAI-compatible interface, and Docker. Demo mode works without an API key; OpenAI, DeepSeek, a Gemini-compatible endpoint, or local Ollama can be configured when suitable.
API usage, hosting, OCR, scanned files, public user authentication, and channel or CRM integrations are excluded unless listed in your package or custom offer.
Please contact me before ordering with a sample file, document and page count, target users, model provider, and deployment needs.
Get to know hakimi
Python Automation Developer
- FromChina
- Member sinceJul 2026
- Avg. response time1 hour
Languages
English, Chinese, Spanish
My Portfolio
FAQ
Are model API or hosting fees included?
No. API tokens, cloud hosting, domains, databases, messaging, and other third-party charges are paid by the buyer. The key-free demo mode can be used for evaluation.
What do you need from me?
Your business goal, target users, authorized source files, approximate document/page count, brand details, required model/provider, deployment environment, and 5-15 acceptance questions with expected sources.
Is deployment included?
Basic is local handover. Standard includes the Docker package and guide, not a live server setup. Premium includes one agreed Docker-capable deployment target after access and environment review.
Is my data saved?
By default, uploaded files, extracted chunks, questions, answers, and review items are stored in the local project database/upload directory. Normal logs exclude document text and API keys. Retention, backups, and full deletion rules should be agreed before deployment.

