I will build an ai chatbot for your website and knowledge base using rag


Informazioni su questo servizio
Help customers or employees get answers from your own knowledge: product information, FAQs, website content and business documents.
I build RAG chatbots that retrieve relevant information, answer with source references and signal when the supplied material is insufficient. Use them for website support, product questions or internal knowledge search.
Basic provides one knowledge base and a simple chat page. Standard adds website embedding, branding and content updates. Premium adds sign-in, document management and conversation history for an internal tool. Text limits and sample-question checks are listed in the FAQs.
You receive source code, deployment to one ready environment and a clear handover. My evidence comes from independent RAG and controlled AI workflow projects, including source checking and evaluation; these are not past commercial AI client deployments.
Send sample material and typical questions. Scanned files, detailed tables, multiple permission groups and live business-system connections can be assessed separately. Hosting and model usage are paid by you.
Scopri di più su Xiang W.
Python automation, websites, AI chatbots and web apps
- DaCina
- Membro daapr 2026
Lingue
Cinese, Inglese
FAQ
What does each package deliver?
Basic: a chat page and 20k-word knowledge base. Standard: 100k words, one website embed and protected content updates. Premium: 250k words, admin/member login, document management and user-owned chat history. Each uses one model provider and includes deployment.
Which sources and text limits are supported?
Clean text from PDF, Word, TXT, FAQs or approved website pages. Limits count total English words; other languages are assessed from a sample. Standard/Premium can ingest one public domain. OCR, complex tables and extensive content cleanup need a custom scope.
What is RAG, and will you train a new model?
RAG retrieves relevant passages from your material before generating an answer. I connect an existing language model to your knowledge base; these packages do not train a new model. The goal is useful, source-linked answers to your business questions.
How will answer quality be checked?
We agree on 10/20/30 sample questions for Basic/Standard/Premium, including missing-information cases. I check retrieval, answers and sources and report limitations. This is a sample evaluation, not a guarantee of perfect accuracy on every possible question.
Can this work with private knowledge or business systems?
Premium supports one organization with admin/member roles and shared document access; chats belong to their users. Fine-grained document permissions, SSO, multiple tenants, live API actions, messaging platforms or private hosting can be scoped separately.
How is private material handled?
Basic and Standard have a public-facing chat: use public/non-sensitive material. Premium adds sign-in. We agree on hosting, model provider and data handling before using private material; I do not promise that an external model keeps all data on your server.
What does a revision include?
One round of consolidated feedback within the agreed scope. I correct defects against the original requirements without counting them as new features. Additional features or a different direction can be quoted separately.
Are source code, hosting and running costs included?
Source code and handover notes are included. You own and pay for domains, hosting, model/API usage and paid services. The listed setup is included; ongoing operation and future changes can be arranged separately.
Do you use AI development tools?
I may use AI to assist development, and I review and test the result. Tell me about confidentiality, data handling or non-AI requirements before sharing materials so we can agree on a suitable workflow.

