
Creation of an AI agent for Claude (Opus 4.8)
- or -
Post a project like this- Posted:
- Proposals: 45
- Remote
- #4500359
- OPPORTUNITY
- Open for Proposals









Description
We're a company looking for an experienced developer to build a customer-facing AI chat assistant powered by Anthropic's Claude API, running on both our website and WhatsApp Business. The assistant answers visitor questions using content drawn exclusively from our own website, captures contact details, and emails each completed conversation to our team so they can follow up. This is a self-contained, end-to-end build with clear, documented deliverables.
Scope of work:
A Node.js + Express backend that calls the Claude API, keeping the full conversation server-side and never exposing credentials to the browser.
A lightweight, embeddable web widget: a floating launcher that invites visitors to ask, a branded chat panel, a custom agent name and logo, and a welcome message.
A WhatsApp Business integration through Meta's official Cloud API: webhook verification, message send/receive, and per-user session handling, reusing the same core logic as the web agent so both channels behave consistently.
A retrieval-augmented (RAG) layer that crawls our sitemap on a nightly schedule, indexes the entire site, and grounds every answer in that content, with strict anti-hallucination rules that direct users to a contact address when information isn't available.
Automatic email delivery of each transcript through our SMTP server (Postfix, SSL) on both channels, sent when a conversation finishes.
Spam and abuse protection: rate limiting, honeypot fields, input validation, and a configurable daily cost cap to protect the API budget.
Requirements: strong JavaScript/Node.js, solid experience with REST APIs and LLM integration, familiarity with embeddings and RAG pipelines, the WhatsApp Business Cloud API, SMTP configuration, and sound security practices. GDPR awareness is a plus.
Deliverables: production-ready, documented code with full setup instructions. Generic placeholders are provided for all company-specific values, so configuration is straightforward.
Please send a brief proposal, relevant work samples, and your estimated timeline and budget.
New Proposal
Login to your account and send a proposal now to get this project.
Log inClarification Board Ask a Question
-

Hi,
Can you please answer below questions
1. Do you already have WhatsApp Business Cloud API access and verified Meta assets?
2. Which hosting environment and database stack do you prefer for deployment? -

A few quick questions:
1. Approximately how many pages are currently available in your sitemap?
2. Is your website server-rendered or JavaScript-rendered?
3. Do you want conversation history stored in a database as well, or only emailed as transcripts?
-

What hosting environment will the application run on (VPS, dedicated server, cloud provider, shared hosting)?
Do you already have Anthropic API access and SMTP credentials available?
Approximately how many website pages are expected to be indexed initially?
Is multilingual support required from day one, or only the site's primary language?
Would you like chat conversations stored in a database in addition to email delivery?
Do you prefer a local JSON/vector index as described, or are you open to a vector database such as pgvector, Pinecone, or Qdrant?
What is the expected daily chat volume so we can size the infrastructure appropriately?
Should the chatbot be embedded on all pages or only selected sections of the website?
Is GDPR-compliant consent and retention handling required as part of the initial implementation?
Do you have a target delivery date in mind for the first production release?